[PATCH 1/7] e1000e: Return 1 instead of a non-zero value for link up indication

[PATCH 1/7] e1000e: Return 1 instead of a non-zero value for link up indication

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

A number of users have mentioned they have tools that rely on a link-up
indication having a return value of 1 rather than a non-zero value.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/ethtool.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/e1000e/ethtool.c b/drivers/net/e1000e/ethtool.c
index cf9679f..e21c9e0 100644
--- a/drivers/net/e1000e/ethtool.c
+++ b/drivers/net/e1000e/ethtool.c
@@ -177,7 +177,7 @@ static u32 e1000_get_link(struct net_device *netdev)
 	u32 status;
 	
 	status = er32(STATUS);
-	return (status & E1000_STATUS_LU);
+	return (status & E1000_STATUS_LU) ? 1 : 0;
 }
 
 static int e1000_set_spd_dplx(struct e1000_adapter *adapter, u16 spddplx)

[PATCH 2/7] e1000e: Set InterruptThrottleRate to default when invalid value used

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

During module load, seting the InterruptThrottleRate parameter to an
invalid value would result in the itr/itr_setting pair being set to
unexpected values which would result in poor performance.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/param.c |   23 +++++++++++++++++++----
 1 files changed, 19 insertions(+), 4 deletions(-)

diff --git a/drivers/net/e1000e/param.c b/drivers/net/e1000e/param.c
index 8effc31..b0499f2 100644
--- a/drivers/net/e1000e/param.c
+++ b/drivers/net/e1000e/param.c
@@ -327,11 +327,26 @@ void __devinit e1000e_check_options(struct e1000_adapter *adapter)
 				e1000_validate_option(&adapter->itr, &opt,
 					adapter);
 				/*
-				 * save the setting, because the dynamic bits
-				 * change itr. clear the lower two bits
-				 * because they are used as control
+				 * Save the setting, because the dynamic bits
+				 * change itr.
 				 */
-				adapter->itr_setting = adapter->itr & ~3;
+				if (e1000_validate_option(&adapter->itr, &opt,
+							  adapter) &&
+				    (adapter->itr == 3)) {
+					/*
+					 * In case of invalid user value,
+					 * default to conservative mode.
+					 */
+					adapter->itr_setting = adapter->itr;
+					adapter->itr = 20000;
+				} else {
+					/*
+					 * Clear the lower two bits because
+					 * they are used as control.
+					 */
+					adapter->itr_setting =
+						adapter->itr & ~3;
+				}
 				break;
 			}
 		} else {

[PATCH 3/7] e1000e: Use skb_copy_to_linear_data_offset introduced in 2.6.22

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

The e1000e driver was based on a version of e1000 prior to acme's
introduction of skb_copy_to_linear_data_offset, and was submitted
after acme went through and coverted all the drivers to use it.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/netdev.c |    9 ++++++---
 1 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/net/e1000e/netdev.c b/drivers/net/e1000e/netdev.c
index 05b0b2f..efa9445 100644
--- a/drivers/net/e1000e/netdev.c
+++ b/drivers/net/e1000e/netdev.c
@@ -510,9 +510,12 @@ static bool e1000_clean_rx_irq(struct e1000_adapter *adapter,
 			    netdev_alloc_skb(netdev, length + NET_IP_ALIGN);
 			if (new_skb) {
 				skb_reserve(new_skb, NET_IP_ALIGN);
-				memcpy(new_skb->data - NET_IP_ALIGN,
-				       skb->data - NET_IP_ALIGN,
-				       length + NET_IP_ALIGN);
+				skb_copy_to_linear_data_offset(new_skb,
+							       -NET_IP_ALIGN,
+							       (skb->data -
+								NET_IP_ALIGN),
+							       (length +
+								NET_IP_ALIGN));
 				/* save the skb in buffer_info as good */
 				buffer_info->skb = skb;
 				skb = new_skb;

[PATCH 4/7] e1000e: Increase Tx timeout factor for 10Mbps

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

Prevent Tx hangs from happening on 10Mb flood ping by increasing the
timeout factor.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/netdev.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/e1000e/netdev.c b/drivers/net/e1000e/netdev.c
index efa9445..4573c7b 100644
--- a/drivers/net/e1000e/netdev.c
+++ b/drivers/net/e1000e/netdev.c
@@ -3058,7 +3058,7 @@ static void e1000_watchdog_task(struct work_struct *work)
 			case SPEED_10:
 				txb2b = 0;
 				netdev->tx_queue_len = 10;
-				adapter->tx_timeout_factor = 14;
+				adapter->tx_timeout_factor = 16;
 				break;
 			case SPEED_100:
 				txb2b = 0;

[PATCH 5/7] e1000e: increase minimum frame size allowed

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

Setting an MTU value below 68 was disabling the network connection and
would not reconnect until the driver was reloaded.  Prevent changing the
MTU to anything below 68.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/netdev.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/net/e1000e/netdev.c b/drivers/net/e1000e/netdev.c
index 4573c7b..b8b26cf 100644
--- a/drivers/net/e1000e/netdev.c
+++ b/drivers/net/e1000e/netdev.c
@@ -3724,7 +3724,7 @@ static int e1000_change_mtu(struct net_device *netdev, int new_mtu)
 	struct e1000_adapter *adapter = netdev_priv(netdev);
 	int max_frame = new_mtu + ETH_HLEN + ETH_FCS_LEN;
 
-	if ((max_frame < ETH_ZLEN + ETH_FCS_LEN) ||
+	if ((new_mtu < ETH_ZLEN + ETH_FCS_LEN + VLAN_HLEN) ||
 	    (max_frame > MAX_JUMBO_FRAME_SIZE)) {
 		e_err("Invalid MTU setting\n");
 		return -EINVAL;

[PATCH 6/7] e1000e: test for unusable MSI support

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

Some systems do not like 82571/2 use of 16-bit MSI messages and some
other systems claim to support MSI, but neither really works.  Setup a
test MSI handler to detect whether or not MSI is working properly, and
if not, fallback to legacy INTx interrupts.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/defines.h |    2 
 drivers/net/e1000e/e1000.h   |    1 
 drivers/net/e1000e/netdev.c  |  172 ++++++++++++++++++++++++++++++++++++++++--
 3 files changed, 164 insertions(+), 11 deletions(-)

diff --git a/drivers/net/e1000e/defines.h b/drivers/net/e1000e/defines.h
index f823b8b..14b0e6c 100644
--- a/drivers/net/e1000e/defines.h
+++ b/drivers/net/e1000e/defines.h
@@ -389,7 +389,7 @@
 
 /* Interrupt Cause Set */
 #define E1000_ICS_LSC       E1000_ICR_LSC       /* Link Status Change */
-#define E1000_ICS_RXDMT0    E1000_ICR_RXDMT0    /* rx desc min. threshold */
+#define E1000_ICS_RXSEQ     E1000_ICR_RXSEQ     /* Rx sequence error */
 #define E1000_ICS_RXDMT0    E1000_ICR_RXDMT0    /* Rx desc min. threshold */
 
 /* Transmit Descriptor Control */
diff --git a/drivers/net/e1000e/e1000.h b/drivers/net/e1000e/e1000.h
index cf57050..ac4e506 100644
--- a/drivers/net/e1000e/e1000.h
+++ b/drivers/net/e1000e/e1000.h
@@ -326,6 +326,7 @@ struct e1000_info {
 #define FLAG_RX_CSUM_ENABLED              (1 << 28)
 #define FLAG_TSO_FORCE                    (1 << 29)
 #define FLAG_RX_RESTART_NOW               (1 << 30)
+#define FLAG_MSI_TEST_FAILED              (1 << 31)
 
 #define E1000_RX_DESC_PS(R, i)	    \
 	(&(((union e1000_rx_desc_packet_split *)((R).desc))[i]))
diff --git a/drivers/net/e1000e/netdev.c b/drivers/net/e1000e/netdev.c
index b8b26cf..d266510 100644
--- a/drivers/net/e1000e/netdev.c
+++ b/drivers/net/e1000e/netdev.c
@@ -1236,26 +1236,36 @@ static irqreturn_t e1000_intr(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
+/**
+ * e1000_request_irq - initialize interrupts
+ *
+ * Attempts to configure interrupts using the best available
+ * capabilities of the hardware and kernel.
+ **/
 static int e1000_request_irq(struct e1000_adapter *adapter)
 {
 	struct net_device *netdev = adapter->netdev;
-	irq_handler_t handler = e1000_intr;
 	int irq_flags = IRQF_SHARED;
 	int err;
 
-	if (!pci_enable_msi(adapter->pdev)) {
-		adapter->flags |= FLAG_MSI_ENABLED;
-		handler = e1000_intr_msi;
-		irq_flags = 0;
+	if (!(adapter->flags & FLAG_MSI_TEST_FAILED)) {
+		err = pci_enable_msi(adapter->pdev);
+		if (!err) {
+			adapter->flags |= FLAG_MSI_ENABLED;
+			irq_flags = 0;
+		}
 	}
 
-	err = request_irq(adapter->pdev->irq, handler, irq_flags, netdev->name,
-			  netdev);
+	err = request_irq(adapter->pdev->irq,
+			  ((adapter->flags & FLAG_MSI_ENABLED) ?
+				&e1000_intr_msi : &e1000_intr),
+			  irq_flags, netdev->name, netdev);
 	if (err) {
-		e_err("Unable to allocate %s interrupt (return: %d)\n",
-		      adapter->flags & FLAG_MSI_ENABLED ? "MSI":"INTx",	err);
-		if (adapter->flags & FLAG_MSI_ENABLED)
+		if (adapter->flags & FLAG_MSI_ENABLED) {
 			pci_disable_msi(adapter->pdev);
+			adapter->flags &= ~FLAG_MSI_ENABLED;
+		}
+		e_err("Unable to allocate interrupt, Error: %d\n", err);
 	}
 
 	return err;
@@ -2595,6 +2605,135 @@ err:
 }
 
 /**
+ * e1000_intr_msi_test - Interrupt Handler
+ * @irq: interrupt number
+ * @data: pointer to a network interface device structure
+ **/
+static irqreturn_t e1000_intr_msi_test(int irq, void *data)
+{
+	struct net_device *netdev = data;
+	struct e1000_adapter *adapter = netdev_priv(netdev);
+	struct e1000_hw *hw = &adapter->hw;
+	u32 icr = er32(ICR);
+
+	e_dbg("%s: icr is %08X\n", netdev->name, icr);
+	if (icr & E1000_ICR_RXSEQ) {
+		adapter->flags &= ~FLAG_MSI_TEST_FAILED;
+		wmb();
+	}
+
+	return IRQ_HANDLED;
+}
+
+/**
+ * e1000_test_msi_interrupt - Returns 0 for successful test
+ * @adapter: board private struct
+ *
+ * code flow taken from tg3.c
+ **/
+static int e1000_test_msi_interrupt(struct e1000_adapter *adapter)
+{
+	struct net_device *netdev = adapter->netdev;
+	struct e1000_hw *hw = &adapter->hw;
+	int err;
+
+	/* poll_enable hasn't been called yet, so don't need disable */
+	/* clear any pending events */
+	er32(ICR);
+
+	/* free the real vector and request a test handler */
+	e1000_free_irq(adapter);
+
+	/* Assume that the test fails, if it succeeds then the test
+	 * MSI irq handler will unset this flag */
+	adapter->flags |= FLAG_MSI_TEST_FAILED;
+
+	err = pci_enable_msi(adapter->pdev);
+	if (err)
+		goto msi_test_failed;
+
+	err = request_irq(adapter->pdev->irq, &e1000_intr_msi_test, 0,
+			  netdev->name, netdev);
+	if (err) {
+		pci_disable_msi(adapter->pdev);
+		goto msi_test_failed;
+	}
+
+	wmb();
+
+	e1000_irq_enable(adapter);
+
+	/* fire an unusual interrupt on the test handler */
+	ew32(ICS, E1000_ICS_RXSEQ);
+	e1e_flush();
+	msleep(50);
+
+	e1000_irq_disable(adapter);
+
+	rmb();
+
+	if (adapter->flags & FLAG_MSI_TEST_FAILED) {
+		err = -EIO;
+		e_info("MSI interrupt test failed!\n");
+	}
+
+	free_irq(adapter->pdev->irq, netdev);
+	pci_disable_msi(adapter->pdev);
+
+	if (err == -EIO)
+		goto msi_test_failed;
+
+	/* okay so the test worked, restore settings */
+	e_dbg("%s: MSI interrupt test succeeded!\n", netdev->name);
+msi_test_failed:
+	/* restore the original vector, even if it failed */
+	e1000_request_irq(adapter);
+	return err;
+}
+
+/**
+ * e1000_test_msi - Returns 0 if MSI test succeeds or INTx mode is restored
+ * @adapter: board private struct
+ *
+ * code flow taken from tg3.c, called with e1000 interrupts disabled.
+ **/
+static int e1000_test_msi(struct e1000_adapter *adapter)
+{
+	int err;
+	u16 pci_cmd;
+
+	if (!(adapter->flags & FLAG_MSI_ENABLED))
+		return 0;
+
+	/* disable SERR in case the MSI write causes a master abort */
+	pci_read_config_word(adapter->pdev, PCI_COMMAND, &pci_cmd);
+	pci_write_config_word(adapter->pdev, PCI_COMMAND,
+			      pci_cmd & ~PCI_COMMAND_SERR);
+
+	err = e1000_test_msi_interrupt(adapter);
+
+	/* restore previous setting of command word */
+	pci_write_config_word(adapter->pdev, PCI_COMMAND, pci_cmd);
+
+	/* success ! */
+	if (!err)
+		return 0;
+
+	/* EIO means MSI test failed */
+	if (err != -EIO)
+		return err;
+
+	/* back to INTx mode */
+	e_warn("MSI interrupt test failed, using legacy interrupt.\n");
+
+	e1000_free_irq(adapter);
+
+	err = e1000_request_irq(adapter);
+
+	return err;
+}
+
+/**
  * e1000_open - Called when a network interface is made active
  * @netdev: network interface device structure
  *
@@ -2652,6 +2791,19 @@ static int e1000_open(struct net_device *netdev)
 	if (err)
 		goto err_req_irq;
 
+	/*
+	 * Work around PCIe errata with MSI interrupts causing some chipsets to
+	 * ignore e1000e MSI messages, which means we need to test our MSI
+	 * interrupt now
+	 */
+	{
+		err = e1000_test_msi(adapter);
+		if (err) {
+			e_err("Interrupt allocation failed\n");
+			goto err_req_irq;
+		}
+	}
+
 	/* From here on the code is the same as e1000e_up() */
 	clear_bit(__E1000_DOWN, &adapter->state);
 

[PATCH 7/7] e1000e: remove unnecessary snippet missed in prior check_options update

[Jeff Kirsher]

From: Bruce Allan <bruce.w.allan@intel.com>

The removal of this bit of code was missed in an earlier patch submittal.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---

 drivers/net/e1000e/param.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)

diff --git a/drivers/net/e1000e/param.c b/drivers/net/e1000e/param.c
index b0499f2..ed912e0 100644
--- a/drivers/net/e1000e/param.c
+++ b/drivers/net/e1000e/param.c
@@ -324,8 +324,6 @@ void __devinit e1000e_check_options(struct e1000_adapter *adapter)
 				adapter->itr = 20000;
 				break;
 			default:
-				e1000_validate_option(&adapter->itr, &opt,
-					adapter);
 				/*
 				 * Save the setting, because the dynamic bits
 				 * change itr.

Re: [PATCH 1/7] e1000e: Return 1 instead of a non-zero value for link up indication

[Jeff Garzik]

Jeff Kirsher wrote:
> From: Bruce Allan <bruce.w.allan@intel.com>
> 
> A number of users have mentioned they have tools that rely on a link-up
> indication having a return value of 1 rather than a non-zero value.
> 
> Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> ---
> 
>  drivers/net/e1000e/ethtool.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/net/e1000e/ethtool.c b/drivers/net/e1000e/ethtool.c
> index cf9679f..e21c9e0 100644
> --- a/drivers/net/e1000e/ethtool.c
> +++ b/drivers/net/e1000e/ethtool.c
> @@ -177,7 +177,7 @@ static u32 e1000_get_link(struct net_device *netdev)
>  	u32 status;
>  	
>  	status = er32(STATUS);
> -	return (status & E1000_STATUS_LU);
> +	return (status & E1000_STATUS_LU) ? 1 : 0;

applied 1-7