[net-next 00/13][pull request] Intel Wired LAN Driver Updates

[net-next 00/13][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains fixes to e1000e, igb, ixgbe, ixgebvf, i40e and
i40evf.

David provides a fix for e1000e to resolve an issue where the device is
capable of transmitting packets but is unable to receive packets until
a previously introduced workaround is called.

Jakub Kicinski provides PTP fixes for ixgbe, which include removing a
redundant if clause and make sure we are not generating both a software and
hardware timestamp.  As well as fix a race condition and leaking skbs
when multiple transmit rings try to claim time stamping.

Jean Sacren fixes a function declaration in ixgbe which was introduced
in commit c97506ab0e22 ("ixgbe: Add check for FW veto bit").  In addition
fixes a function header comment in i40e and fixes the error checking
by binding the check to the pertinent DMA bit mask.

Mark provides several fixes for ixgbe and ixgbevf.  Most notably are fixes
to resolve namespace issues and fix ECU warnings induced by LER for ixgbe
and ixgbevf.

Joe Perches fixes up unnecessary casts in i40e and i40evf.

Peter Senna Tschudin fixes igb to use pci_iounmap when the virtual mapping
was done with pci_iomap.
 
The following are changes since commit 0b70195e0c3206103be991e196c26fcf168d0334:
  Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-next master

David Ertman (1):
  e1000e: Fix no connectivity when driver loaded with cable out

Jakub Kicinski (3):
  ixgbe: remove redundant if clause from PTP work
  ixgbe: never generate both software and hardware timestamps
  ixgbe: fix race conditions on queuing skb for HW time stamp

Jean Sacren (3):
  ixgbe: fix ixgbe_check_reset_blocked() declaration
  i40e: fix function kernel doc description
  i40e/i40evf: fix error checking path

Joe Perches (2):
  i40e/i40evf: Remove addressof casts to same type
  i40e: Remove casts of pointer to same type

Mark Rustad (3):
  ixgbevf: Change ixgbe_read_reg to ixgbevf_read_reg
  ixgbe: Fix rcu warnings induced by LER
  ixgbevf: Fix rcu warnings induced by LER

Peter Senna Tschudin (1):
  INTEL-IGB: Convert iounmap to pci_iounmap

 drivers/net/ethernet/intel/e1000e/netdev.c        | 20 +++++++++++----
 drivers/net/ethernet/intel/i40e/i40e_common.c     |  4 +--
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c    |  4 +--
 drivers/net/ethernet/intel/i40e/i40e_main.c       | 11 ++++----
 drivers/net/ethernet/intel/i40e/i40e_txrx.c       |  2 +-
 drivers/net/ethernet/intel/i40evf/i40e_common.c   |  3 +--
 drivers/net/ethernet/intel/i40evf/i40evf_main.c   | 11 ++++----
 drivers/net/ethernet/intel/igb/igb_main.c         |  4 +--
 drivers/net/ethernet/intel/ixgbe/ixgbe.h          |  2 ++
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c     | 31 +++++++++++++++++------
 drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c      |  2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h      |  2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c      |  7 +++--
 drivers/net/ethernet/intel/ixgbevf/ethtool.c      |  8 +++---
 drivers/net/ethernet/intel/ixgbevf/ixgbevf.h      |  1 +
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 26 ++++++++++++++-----
 drivers/net/ethernet/intel/ixgbevf/vf.h           |  6 ++---
 17 files changed, 92 insertions(+), 52 deletions(-)

-- 
1.9.0

[net-next 01/13] e1000e: Fix no connectivity when driver loaded with cable out

[Jeff Kirsher]

From: David Ertman <davidx.m.ertman@intel.com>

In commit da1e2046e5, the flow for enabling/disabling an Si errata
workaround (e1000_lv_jumbo_workaround_ich8lan) was changed to fix a problem
with iAMT connections dropping on interface down with jumbo frames set.
Part of this change was to move the function call disabling the workaround
to e1000e_down() from the e1000_setup_rctl() function.  The mechanic for
disabling of this workaround involves writing several MAC and PHY registers
back to hardware defaults.

After this commit, when the driver is loaded with the cable out, the PHY
registers are not programmed with the correct default values.  This causes
the device to be capable of transmitting packets, but is unable to recieve
them until this workaround is called.

The flow of e1000e's open code relies upon calling the above workaround to
expicitly program these registers either with jumbo frame appropriate settings
or h/w defaults on 82579 and newer hardware.

Fix this issue by adding logic to e1000_setup_rctl() that not only calls
e1000_lv_jumbo_workaround_ich8lan() when jumbo frames are set, to enable the
workaround, but also calls this function to explicitly disable the workaround
in the case that jumbo frames are not set.

Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/e1000e/netdev.c | 20 +++++++++++++++-----
 1 file changed, 15 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c
index f1cce59..dce377b 100644
--- a/drivers/net/ethernet/intel/e1000e/netdev.c
+++ b/drivers/net/ethernet/intel/e1000e/netdev.c
@@ -2991,11 +2991,21 @@ static void e1000_setup_rctl(struct e1000_adapter *adapter)
 	u32 rctl, rfctl;
 	u32 pages = 0;
 
-	/* Workaround Si errata on PCHx - configure jumbo frame flow */
-	if ((hw->mac.type >= e1000_pch2lan) &&
-	    (adapter->netdev->mtu > ETH_DATA_LEN) &&
-	    e1000_lv_jumbo_workaround_ich8lan(hw, true))
-		e_dbg("failed to enable jumbo frame workaround mode\n");
+	/* Workaround Si errata on PCHx - configure jumbo frame flow.
+	 * If jumbo frames not set, program related MAC/PHY registers
+	 * to h/w defaults
+	 */
+	if (hw->mac.type >= e1000_pch2lan) {
+		s32 ret_val;
+
+		if (adapter->netdev->mtu > ETH_DATA_LEN)
+			ret_val = e1000_lv_jumbo_workaround_ich8lan(hw, true);
+		else
+			ret_val = e1000_lv_jumbo_workaround_ich8lan(hw, false);
+
+		if (ret_val)
+			e_dbg("failed to enable|disable jumbo frame workaround mode\n");
+	}
 
 	/* Program MC offset vector base */
 	rctl = er32(RCTL);
-- 
1.9.0

[net-next 02/13] ixgbe: remove redundant if clause from PTP work

[Jeff Kirsher]

From: Jakub Kicinski <kubakici@wp.pl>

ptp_tx_skb is always set before work is scheduled,
work is cancelled before ptp_tx_skb is set to NULL.
PTP work cannot ever see ptp_tx_skb set to NULL.

Signed-off-by: Jakub Kicinski <kubakici@wp.pl>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c
index 44ac9ae..8b527d7 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c
@@ -512,10 +512,6 @@ static void ixgbe_ptp_tx_hwtstamp_work(struct work_struct *work)
 					      IXGBE_PTP_TX_TIMEOUT);
 	u32 tsynctxctl;
 
-	/* we have to have a valid skb */
-	if (!adapter->ptp_tx_skb)
-		return;
-
 	if (timeout) {
 		dev_kfree_skb_any(adapter->ptp_tx_skb);
 		adapter->ptp_tx_skb = NULL;
-- 
1.9.0

[net-next 03/13] ixgbe: never generate both software and hardware timestamps

[Jeff Kirsher]

From: Jakub Kicinski <kubakici@wp.pl>

skb_tx_timestamp() does not report software time stamp
if SKBTX_IN_PROGRESS is set. According to timestamping.txt
software time stamps are a fallback and should not be
generated if hardware time stamp is provided.

Move call to skb_tx_timestamp() after setting
SKBTX_IN_PROGRESS.

Signed-off-by: Jakub Kicinski <kubakici@wp.pl>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 9e5a366..24538cb 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -7042,8 +7042,6 @@ netdev_tx_t ixgbe_xmit_frame_ring(struct sk_buff *skb,
 		tx_flags |= IXGBE_TX_FLAGS_SW_VLAN;
 	}
 
-	skb_tx_timestamp(skb);
-
 	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP)) {
 		skb_shinfo(skb)->tx_flags |= SKBTX_IN_PROGRESS;
 		tx_flags |= IXGBE_TX_FLAGS_TSTAMP;
@@ -7054,6 +7052,8 @@ netdev_tx_t ixgbe_xmit_frame_ring(struct sk_buff *skb,
 		schedule_work(&adapter->ptp_tx_work);
 	}
 
+	skb_tx_timestamp(skb);
+
 #ifdef CONFIG_PCI_IOV
 	/*
 	 * Use the l2switch_enable flag - would be false if the DMA
-- 
1.9.0

[net-next 04/13] ixgbe: fix race conditions on queuing skb for HW time stamp

[Jeff Kirsher]

From: Jakub Kicinski <kubakici@wp.pl>

ixgbe has a single set of TX time stamping resources per NIC.
Use a simple bit lock to avoid race conditions and leaking skbs
when multiple TX rings try to claim time stamping.

Signed-off-by: Jakub Kicinski <kubakici@wp.pl>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe.h      | 1 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 4 +++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c  | 3 +++
 3 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe.h b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
index 2fff0fc..26d27aa 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
@@ -812,6 +812,7 @@ enum ixgbe_state_t {
 	__IXGBE_SERVICE_SCHED,
 	__IXGBE_IN_SFP_INIT,
 	__IXGBE_PTP_RUNNING,
+	__IXGBE_PTP_TX_IN_PROGRESS,
 };
 
 struct ixgbe_cb {
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 24538cb..c4b930c 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -7042,7 +7042,9 @@ netdev_tx_t ixgbe_xmit_frame_ring(struct sk_buff *skb,
 		tx_flags |= IXGBE_TX_FLAGS_SW_VLAN;
 	}
 
-	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP)) {
+	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP &&
+		     !test_and_set_bit_lock(__IXGBE_PTP_TX_IN_PROGRESS,
+					    &adapter->state))) {
 		skb_shinfo(skb)->tx_flags |= SKBTX_IN_PROGRESS;
 		tx_flags |= IXGBE_TX_FLAGS_TSTAMP;
 
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c
index 8b527d7..63515a6 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ptp.c
@@ -493,6 +493,7 @@ static void ixgbe_ptp_tx_hwtstamp(struct ixgbe_adapter *adapter)
 
 	dev_kfree_skb_any(adapter->ptp_tx_skb);
 	adapter->ptp_tx_skb = NULL;
+	clear_bit_unlock(__IXGBE_PTP_TX_IN_PROGRESS, &adapter->state);
 }
 
 /**
@@ -515,6 +516,7 @@ static void ixgbe_ptp_tx_hwtstamp_work(struct work_struct *work)
 	if (timeout) {
 		dev_kfree_skb_any(adapter->ptp_tx_skb);
 		adapter->ptp_tx_skb = NULL;
+		clear_bit_unlock(__IXGBE_PTP_TX_IN_PROGRESS, &adapter->state);
 		e_warn(drv, "clearing Tx Timestamp hang");
 		return;
 	}
@@ -925,6 +927,7 @@ void ixgbe_ptp_stop(struct ixgbe_adapter *adapter)
 	if (adapter->ptp_tx_skb) {
 		dev_kfree_skb_any(adapter->ptp_tx_skb);
 		adapter->ptp_tx_skb = NULL;
+		clear_bit_unlock(__IXGBE_PTP_TX_IN_PROGRESS, &adapter->state);
 	}
 
 	if (adapter->ptp_clock) {
-- 
1.9.0

[net-next 05/13] ixgbe: fix ixgbe_check_reset_blocked() declaration

[Jeff Kirsher]

From: Jean Sacren <sakiwit@gmail.com>

The commit c97506ab0e22 ("ixgbe: Add check for FW veto bit")
introduced the new function ixgbe_check_reset_blocked() with a minor
issue in declaration. Fix the declaration by changing the type
specifier to bool as the definition returns a boolean value.
Additionally all ixgbe_check_reset_blocked() callers are expected to
return a boolean value.

Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Acked-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c | 2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
index ad51c12..23f7652 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
@@ -107,7 +107,7 @@ s32 ixgbe_identify_phy_generic(struct ixgbe_hw *hw)
  * have this bit just return false since the link can not be blocked
  * via this method.
  **/
-s32 ixgbe_check_reset_blocked(struct ixgbe_hw *hw)
+bool ixgbe_check_reset_blocked(struct ixgbe_hw *hw)
 {
 	u32 mmngc;
 
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
index 4a456c9..0bb047f 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
@@ -128,7 +128,7 @@ s32 ixgbe_setup_phy_link_speed_generic(struct ixgbe_hw *hw,
 s32 ixgbe_get_copper_link_capabilities_generic(struct ixgbe_hw *hw,
                                                ixgbe_link_speed *speed,
                                                bool *autoneg);
-s32 ixgbe_check_reset_blocked(struct ixgbe_hw *hw);
+bool ixgbe_check_reset_blocked(struct ixgbe_hw *hw);
 
 /* PHY specific */
 s32 ixgbe_check_phy_link_tnx(struct ixgbe_hw *hw,
-- 
1.9.0

[net-next 06/13] ixgbevf: Change ixgbe_read_reg to ixgbevf_read_reg

[Jeff Kirsher]

From: Mark Rustad <mark.d.rustad@intel.com>

Change the ixgbe_read_reg function name to ixgbevf_read_reg to
avoid a namespace clash with the ixgbe driver. This will allow
ixgbe to take its register read function out-of-line to reduce
memory footprint.

Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbevf/ethtool.c      | 8 ++++----
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 4 ++--
 drivers/net/ethernet/intel/ixgbevf/vf.h           | 6 +++---
 3 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbevf/ethtool.c b/drivers/net/ethernet/intel/ixgbevf/ethtool.c
index b2d0023..1baecb6 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ethtool.c
@@ -540,10 +540,10 @@ static bool reg_pattern_test(struct ixgbevf_adapter *adapter, u64 *data,
 		return true;
 	}
 	for (pat = 0; pat < ARRAY_SIZE(register_test_patterns); pat++) {
-		before = ixgbe_read_reg(&adapter->hw, reg);
+		before = ixgbevf_read_reg(&adapter->hw, reg);
 		ixgbe_write_reg(&adapter->hw, reg,
 				register_test_patterns[pat] & write);
-		val = ixgbe_read_reg(&adapter->hw, reg);
+		val = ixgbevf_read_reg(&adapter->hw, reg);
 		if (val != (register_test_patterns[pat] & write & mask)) {
 			hw_dbg(&adapter->hw,
 			       "pattern test reg %04X failed: got 0x%08X expected 0x%08X\n",
@@ -567,9 +567,9 @@ static bool reg_set_and_check(struct ixgbevf_adapter *adapter, u64 *data,
 		*data = 1;
 		return true;
 	}
-	before = ixgbe_read_reg(&adapter->hw, reg);
+	before = ixgbevf_read_reg(&adapter->hw, reg);
 	ixgbe_write_reg(&adapter->hw, reg, write & mask);
-	val = ixgbe_read_reg(&adapter->hw, reg);
+	val = ixgbevf_read_reg(&adapter->hw, reg);
 	if ((write & mask) != (val & mask)) {
 		pr_err("set/check reg %04X test failed: got 0x%08X expected 0x%08X\n",
 		       reg, (val & mask), write & mask);
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
index a50e892..5d93759 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
@@ -124,12 +124,12 @@ static void ixgbevf_check_remove(struct ixgbe_hw *hw, u32 reg)
 		ixgbevf_remove_adapter(hw);
 		return;
 	}
-	value = ixgbe_read_reg(hw, IXGBE_VFSTATUS);
+	value = ixgbevf_read_reg(hw, IXGBE_VFSTATUS);
 	if (value == IXGBE_FAILED_READ_REG)
 		ixgbevf_remove_adapter(hw);
 }
 
-u32 ixgbe_read_reg(struct ixgbe_hw *hw, u32 reg)
+u32 ixgbevf_read_reg(struct ixgbe_hw *hw, u32 reg)
 {
 	u8 __iomem *reg_addr = ACCESS_ONCE(hw->hw_addr);
 	u32 value;
diff --git a/drivers/net/ethernet/intel/ixgbevf/vf.h b/drivers/net/ethernet/intel/ixgbevf/vf.h
index 096d33a..3061d18 100644
--- a/drivers/net/ethernet/intel/ixgbevf/vf.h
+++ b/drivers/net/ethernet/intel/ixgbevf/vf.h
@@ -186,8 +186,8 @@ static inline void ixgbe_write_reg(struct ixgbe_hw *hw, u32 reg, u32 value)
 }
 #define IXGBE_WRITE_REG(h, r, v) ixgbe_write_reg(h, r, v)
 
-u32 ixgbe_read_reg(struct ixgbe_hw *hw, u32 reg);
-#define IXGBE_READ_REG(h, r) ixgbe_read_reg(h, r)
+u32 ixgbevf_read_reg(struct ixgbe_hw *hw, u32 reg);
+#define IXGBE_READ_REG(h, r) ixgbevf_read_reg(h, r)
 
 static inline void ixgbe_write_reg_array(struct ixgbe_hw *hw, u32 reg,
 					  u32 offset, u32 value)
@@ -199,7 +199,7 @@ static inline void ixgbe_write_reg_array(struct ixgbe_hw *hw, u32 reg,
 static inline u32 ixgbe_read_reg_array(struct ixgbe_hw *hw, u32 reg,
 					u32 offset)
 {
-	return ixgbe_read_reg(hw, reg + (offset << 2));
+	return ixgbevf_read_reg(hw, reg + (offset << 2));
 }
 #define IXGBE_READ_REG_ARRAY(h, r, o) ixgbe_read_reg_array(h, r, o)
 
-- 
1.9.0

[net-next 07/13] i40e: fix function kernel doc description

[Jeff Kirsher]

From: Jean Sacren <sakiwit@gmail.com>

The commit c7d05ca89f8e ("i40e: driver ethtool core") introduced the
new function i40e_add_del_fdir_sctpv4() with the kernel doc
description a little bit off. The trivial error was copied over to a
different file by the commit 17a73f6b1401 ("i40e: Flow Director
sideband accounting") most recently. Fix the kernel doc with the
correct description for clarity.

Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Joseph Gasparakis <joseph.gasparakis@intel.com>
Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_txrx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.c b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
index a329aac..0f5d96a 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_txrx.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
@@ -299,7 +299,7 @@ static int i40e_add_del_fdir_tcpv4(struct i40e_vsi *vsi,
  * @raw_packet: the pre-allocated packet buffer for FDir
  * @add: true adds a filter, false removes it
  *
- * Returns 0 if the filters were successfully added or removed
+ * Always returns -EOPNOTSUPP
  **/
 static int i40e_add_del_fdir_sctpv4(struct i40e_vsi *vsi,
 				    struct i40e_fdir_filter *fd_data,
-- 
1.9.0

[net-next 08/13] i40e/i40evf: fix error checking path

[Jeff Kirsher]

From: Jean Sacren <sakiwit@gmail.com>

The commit 6494294f277fd ("i40e/i40evf: Use
dma_set_mask_and_coherent") uses dma_set_mask_and_coherent() to
replace dma_set_coherent_mask() for the benefit of return error.
The conversion brings some confusion in error checking as whether
against DMA_BIT_MASK(64) or DMA_BIT_MASK(32). For one, if error is
zero, the check will be against DMA_BIT_MASK(64) twice. Fix this
error checking by binding the check to the pertinent one.

Cc: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_main.c     | 11 ++++++-----
 drivers/net/ethernet/intel/i40evf/i40evf_main.c | 11 ++++++-----
 2 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index a1ec793..861b722 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -8091,12 +8091,13 @@ static int i40e_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 	/* set up for high or low dma */
 	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
-	if (err)
-		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
 	if (err) {
-		dev_err(&pdev->dev,
-			"DMA configuration failed: 0x%x\n", err);
-		goto err_dma;
+		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
+		if (err) {
+			dev_err(&pdev->dev,
+				"DMA configuration failed: 0x%x\n", err);
+			goto err_dma;
+		}
 	}
 
 	/* set up pci connections */
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 51c84c1..e35e66f 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -2191,12 +2191,13 @@ static int i40evf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		return err;
 
 	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
-	if (err)
-		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
 	if (err) {
-		dev_err(&pdev->dev,
-			"DMA configuration failed: 0x%x\n", err);
-		goto err_dma;
+		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
+		if (err) {
+			dev_err(&pdev->dev,
+				"DMA configuration failed: 0x%x\n", err);
+			goto err_dma;
+		}
 	}
 
 	err = pci_request_regions(pdev, i40evf_driver_name);
-- 
1.9.0

[net-next 09/13] i40e/i40evf: Remove addressof casts to same type

[Jeff Kirsher]

From: Joe Perches <joe@perches.com>

Using addressof then casting to the original type is pointless,
so remove these unnecessary casts.

Done via coccinelle script:

$ cat typecast.cocci
@@
type T;
T foo;
@@

-	(T *)&foo
+	&foo

Signed-off-by: Joe Perches <joe@perches.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_common.c   | 4 ++--
 drivers/net/ethernet/intel/i40evf/i40e_common.c | 3 +--
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_common.c b/drivers/net/ethernet/intel/i40e/i40e_common.c
index bb948dd..922cdcc 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_common.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_common.c
@@ -1775,9 +1775,9 @@ static void i40e_parse_discover_capabilities(struct i40e_hw *hw, void *buff,
 	cap = (struct i40e_aqc_list_capabilities_element_resp *) buff;
 
 	if (list_type_opc == i40e_aqc_opc_list_dev_capabilities)
-		p = (struct i40e_hw_capabilities *)&hw->dev_caps;
+		p = &hw->dev_caps;
 	else if (list_type_opc == i40e_aqc_opc_list_func_capabilities)
-		p = (struct i40e_hw_capabilities *)&hw->func_caps;
+		p = &hw->func_caps;
 	else
 		return;
 
diff --git a/drivers/net/ethernet/intel/i40evf/i40e_common.c b/drivers/net/ethernet/intel/i40evf/i40e_common.c
index c688a0f..ae08437 100644
--- a/drivers/net/ethernet/intel/i40evf/i40e_common.c
+++ b/drivers/net/ethernet/intel/i40evf/i40e_common.c
@@ -565,8 +565,7 @@ i40e_status i40e_aq_send_msg_to_pf(struct i40e_hw *hw,
 		details.async = true;
 		cmd_details = &details;
 	}
-	status = i40evf_asq_send_command(hw, (struct i40e_aq_desc *)&desc, msg,
-				       msglen, cmd_details);
+	status = i40evf_asq_send_command(hw, &desc, msg, msglen, cmd_details);
 	return status;
 }
 
-- 
1.9.0

[net-next 10/13] i40e: Remove casts of pointer to same type

[Jeff Kirsher]

From: Joe Perches <joe@perches.com>

Casting a pointer to a pointer of the same type is pointless,
so remove these unnecessary casts.

Done via coccinelle script:

$ cat typecast_2.cocci
@@
type T;
T *foo;
@@

-	(T *)foo
+	foo

Signed-off-by: Joe Perches <joe@perches.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_ethtool.c b/drivers/net/ethernet/intel/i40e/i40e_ethtool.c
index aa123f4..03d99cb 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_ethtool.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_ethtool.c
@@ -396,7 +396,7 @@ static int i40e_get_eeprom(struct net_device *netdev,
 		ret_val = i40e_aq_read_nvm(hw, 0x0,
 				eeprom->offset + (I40E_NVM_SECTOR_SIZE * i),
 				len,
-				(u8 *)eeprom_buff + (I40E_NVM_SECTOR_SIZE * i),
+				eeprom_buff + (I40E_NVM_SECTOR_SIZE * i),
 				last, NULL);
 		if (ret_val) {
 			dev_info(&pf->pdev->dev,
@@ -408,7 +408,7 @@ static int i40e_get_eeprom(struct net_device *netdev,
 
 release_nvm:
 	i40e_release_nvm(hw);
-	memcpy(bytes, (u8 *)eeprom_buff, eeprom->len);
+	memcpy(bytes, eeprom_buff, eeprom->len);
 free_buff:
 	kfree(eeprom_buff);
 	return ret_val;
-- 
1.9.0

[net-next 11/13] INTEL-IGB: Convert iounmap to pci_iounmap

[Jeff Kirsher]

From: Peter Senna Tschudin <peter.senna@gmail.com>

Use pci_iounmap instead of iounmap when the virtual mapping was done
with pci_iomap. A simplified version of the semantic patch that finds this
issue is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
expression addr;
@@
addr = pci_iomap(...)

@rr@
expression r.addr;
@@
* iounmap(addr)
// </smpl>

Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index 55fc559..3019818 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -2652,7 +2652,7 @@ err_eeprom:
 		iounmap(hw->flash_address);
 err_sw_init:
 	igb_clear_interrupt_scheme(adapter);
-	iounmap(hw->hw_addr);
+	pci_iounmap(pdev, hw->hw_addr);
 err_ioremap:
 	free_netdev(netdev);
 err_alloc_etherdev:
@@ -2819,7 +2819,7 @@ static void igb_remove(struct pci_dev *pdev)
 	igb_disable_sriov(pdev);
 #endif
 
-	iounmap(hw->hw_addr);
+	pci_iounmap(pdev, hw->hw_addr);
 	if (hw->flash_address)
 		iounmap(hw->flash_address);
 	pci_release_selected_regions(pdev,
-- 
1.9.0

[net-next 12/13] ixgbe: Fix rcu warnings induced by LER

[Jeff Kirsher]

From: Mark Rustad <mark.d.rustad@intel.com>

Resolve some rcu warnings produced when LER actions take place.
This appears to be due to not holding the rtnl lock when calling
ixgbe_down, so hold the lock. Also avoid disabling the device
when it is already disabled. This check is necessary because the
callback can be called more than once in some cases.

Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe.h      |  1 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 23 ++++++++++++++++++-----
 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe.h b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
index 26d27aa..55c53a1 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
@@ -808,6 +808,7 @@ enum ixgbe_state_t {
 	__IXGBE_TESTING,
 	__IXGBE_RESETTING,
 	__IXGBE_DOWN,
+	__IXGBE_DISABLED,
 	__IXGBE_REMOVING,
 	__IXGBE_SERVICE_SCHED,
 	__IXGBE_IN_SFP_INIT,
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index c4b930c..8436c65 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -5566,6 +5566,8 @@ static int ixgbe_resume(struct pci_dev *pdev)
 		e_dev_err("Cannot enable PCI device from suspend\n");
 		return err;
 	}
+	smp_mb__before_clear_bit();
+	clear_bit(__IXGBE_DISABLED, &adapter->state);
 	pci_set_master(pdev);
 
 	pci_wake_from_d3(pdev, false);
@@ -5663,7 +5665,8 @@ static int __ixgbe_shutdown(struct pci_dev *pdev, bool *enable_wake)
 
 	ixgbe_release_hw_control(adapter);
 
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBE_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
 
 	return 0;
 }
@@ -8313,7 +8316,8 @@ err_alloc_etherdev:
 				     pci_select_bars(pdev, IORESOURCE_MEM));
 err_pci_reg:
 err_dma:
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBE_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
 	return err;
 }
 
@@ -8382,7 +8386,8 @@ static void ixgbe_remove(struct pci_dev *pdev)
 
 	pci_disable_pcie_error_reporting(pdev);
 
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBE_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
 }
 
 /**
@@ -8489,14 +8494,20 @@ static pci_ers_result_t ixgbe_io_error_detected(struct pci_dev *pdev,
 
 skip_bad_vf_detection:
 #endif /* CONFIG_PCI_IOV */
+	rtnl_lock();
 	netif_device_detach(netdev);
 
-	if (state == pci_channel_io_perm_failure)
+	if (state == pci_channel_io_perm_failure) {
+		rtnl_unlock();
 		return PCI_ERS_RESULT_DISCONNECT;
+	}
 
 	if (netif_running(netdev))
 		ixgbe_down(adapter);
-	pci_disable_device(pdev);
+
+	if (!test_and_set_bit(__IXGBE_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
+	rtnl_unlock();
 
 	/* Request a slot reset. */
 	return PCI_ERS_RESULT_NEED_RESET;
@@ -8518,6 +8529,8 @@ static pci_ers_result_t ixgbe_io_slot_reset(struct pci_dev *pdev)
 		e_err(probe, "Cannot re-enable PCI device after reset.\n");
 		result = PCI_ERS_RESULT_DISCONNECT;
 	} else {
+		smp_mb__before_clear_bit();
+		clear_bit(__IXGBE_DISABLED, &adapter->state);
 		adapter->hw.hw_addr = adapter->io_addr;
 		pci_set_master(pdev);
 		pci_restore_state(pdev);
-- 
1.9.0

[net-next 13/13] ixgbevf: Fix rcu warnings induced by LER

[Jeff Kirsher]

From: Mark Rustad <mark.d.rustad@intel.com>

Resolve some rcu warnings produced when LER actions take place.
This appears to be due to not holding the rtnl lock when calling
ixgbe_down, so hold the lock. Also avoid disabling the device
when it is already disabled. This check is necessary because the
callback can be called more than once in some cases.

Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbevf/ixgbevf.h      |  1 +
 drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 22 +++++++++++++++++-----
 2 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h b/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h
index a08bd7c..e7e7d69 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h
@@ -419,6 +419,7 @@ enum ixbgevf_state_t {
 	__IXGBEVF_TESTING,
 	__IXGBEVF_RESETTING,
 	__IXGBEVF_DOWN,
+	__IXGBEVF_DISABLED,
 	__IXGBEVF_REMOVING,
 };
 
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
index 5d93759..4ba139b 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
@@ -3329,7 +3329,8 @@ static int ixgbevf_suspend(struct pci_dev *pdev, pm_message_t state)
 		return retval;
 
 #endif
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBEVF_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
 
 	return 0;
 }
@@ -3353,6 +3354,8 @@ static int ixgbevf_resume(struct pci_dev *pdev)
 		dev_err(&pdev->dev, "Cannot enable PCI device from suspend\n");
 		return err;
 	}
+	smp_mb__before_clear_bit();
+	clear_bit(__IXGBEVF_DISABLED, &adapter->state);
 	pci_set_master(pdev);
 
 	ixgbevf_reset(adapter);
@@ -3607,7 +3610,8 @@ err_alloc_etherdev:
 	pci_release_regions(pdev);
 err_pci_reg:
 err_dma:
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBEVF_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
 	return err;
 }
 
@@ -3645,7 +3649,8 @@ static void ixgbevf_remove(struct pci_dev *pdev)
 
 	free_netdev(netdev);
 
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBEVF_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
 }
 
 /**
@@ -3662,15 +3667,20 @@ static pci_ers_result_t ixgbevf_io_error_detected(struct pci_dev *pdev,
 	struct net_device *netdev = pci_get_drvdata(pdev);
 	struct ixgbevf_adapter *adapter = netdev_priv(netdev);
 
+	rtnl_lock();
 	netif_device_detach(netdev);
 
-	if (state == pci_channel_io_perm_failure)
+	if (state == pci_channel_io_perm_failure) {
+		rtnl_unlock();
 		return PCI_ERS_RESULT_DISCONNECT;
+	}
 
 	if (netif_running(netdev))
 		ixgbevf_down(adapter);
 
-	pci_disable_device(pdev);
+	if (!test_and_set_bit(__IXGBEVF_DISABLED, &adapter->state))
+		pci_disable_device(pdev);
+	rtnl_unlock();
 
 	/* Request a slot slot reset. */
 	return PCI_ERS_RESULT_NEED_RESET;
@@ -3694,6 +3704,8 @@ static pci_ers_result_t ixgbevf_io_slot_reset(struct pci_dev *pdev)
 		return PCI_ERS_RESULT_DISCONNECT;
 	}
 
+	smp_mb__before_clear_bit();
+	clear_bit(__IXGBEVF_DISABLED, &adapter->state);
 	pci_set_master(pdev);
 
 	ixgbevf_reset(adapter);
-- 
1.9.0

Re: [net-next 00/13][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Mon, 31 Mar 2014 16:34:46 -0700

> This series contains fixes to e1000e, igb, ixgbe, ixgebvf, i40e and
> i40evf.

Pulled, thanks Jeff.

Re: [net-next 01/13] e1000e: Fix no connectivity when driver loaded with cable out

[Sergei Shtylyov]

Hello.

On 01-04-2014 3:34, Jeff Kirsher wrote:

> From: David Ertman <davidx.m.ertman@intel.com>

> In commit da1e2046e5, the flow for enabling/disabling an Si errata

    Please also specify that commit's summary line in parens.

> workaround (e1000_lv_jumbo_workaround_ich8lan) was changed to fix a problem
> with iAMT connections dropping on interface down with jumbo frames set.
> Part of this change was to move the function call disabling the workaround
> to e1000e_down() from the e1000_setup_rctl() function.  The mechanic for
> disabling of this workaround involves writing several MAC and PHY registers
> back to hardware defaults.

> After this commit, when the driver is loaded with the cable out, the PHY
> registers are not programmed with the correct default values.  This causes
> the device to be capable of transmitting packets, but is unable to recieve
> them until this workaround is called.

> The flow of e1000e's open code relies upon calling the above workaround to
> expicitly program these registers either with jumbo frame appropriate settings
> or h/w defaults on 82579 and newer hardware.

> Fix this issue by adding logic to e1000_setup_rctl() that not only calls
> e1000_lv_jumbo_workaround_ich8lan() when jumbo frames are set, to enable the
> workaround, but also calls this function to explicitly disable the workaround
> in the case that jumbo frames are not set.

> Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
> Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> ---
>   drivers/net/ethernet/intel/e1000e/netdev.c | 20 +++++++++++++++-----
>   1 file changed, 15 insertions(+), 5 deletions(-)

[...]

WBR, Sergei

Re: [net-next 08/13] i40e/i40evf: fix error checking path

[Sergei Shtylyov]

Hello.

On 01-04-2014 3:34, Jeff Kirsher wrote:

> From: Jean Sacren <sakiwit@gmail.com>

> The commit 6494294f277fd ("i40e/i40evf: Use
> dma_set_mask_and_coherent") uses dma_set_mask_and_coherent() to
> replace dma_set_coherent_mask() for the benefit of return error.
> The conversion brings some confusion in error checking as whether
> against DMA_BIT_MASK(64) or DMA_BIT_MASK(32). For one, if error is
> zero, the check will be against DMA_BIT_MASK(64) twice.

    So what?

> Fix this
> error checking by binding the check to the pertinent one.

    There's nothing wrong with the current code. I don't know what you're 
fixing here.

> Cc: Mitch Williams <mitch.a.williams@intel.com>
> Signed-off-by: Jean Sacren <sakiwit@gmail.com>
> Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> ---
>   drivers/net/ethernet/intel/i40e/i40e_main.c     | 11 ++++++-----
>   drivers/net/ethernet/intel/i40evf/i40evf_main.c | 11 ++++++-----
>   2 files changed, 12 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
> index a1ec793..861b722 100644
> --- a/drivers/net/ethernet/intel/i40e/i40e_main.c
> +++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
> @@ -8091,12 +8091,13 @@ static int i40e_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>
>   	/* set up for high or low dma */
>   	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
> -	if (err)
> -		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
>   	if (err) {
> -		dev_err(&pdev->dev,
> -			"DMA configuration failed: 0x%x\n", err);
> -		goto err_dma;
> +		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
> +		if (err) {
> +			dev_err(&pdev->dev,
> +				"DMA configuration failed: 0x%x\n", err);
> +			goto err_dma;
> +		}
>   	}
>
>   	/* set up pci connections */
> diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
> index 51c84c1..e35e66f 100644
> --- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
> +++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
> @@ -2191,12 +2191,13 @@ static int i40evf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>   		return err;
>
>   	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
> -	if (err)
> -		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
>   	if (err) {
> -		dev_err(&pdev->dev,
> -			"DMA configuration failed: 0x%x\n", err);
> -		goto err_dma;
> +		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
> +		if (err) {
> +			dev_err(&pdev->dev,
> +				"DMA configuration failed: 0x%x\n", err);
> +			goto err_dma;
> +		}
>   	}
>
>   	err = pci_request_regions(pdev, i40evf_driver_name);
>

WBR, Sergei

Re: [net-next 01/13] e1000e: Fix no connectivity when driver loaded with cable out

[Fabio Estevam]

On Mon, Mar 31, 2014 at 8:34 PM, Jeff Kirsher
<jeffrey.t.kirsher@intel.com> wrote:
> From: David Ertman <davidx.m.ertman@intel.com>
>
> In commit da1e2046e5, the flow for enabling/disabling an Si errata
> workaround (e1000_lv_jumbo_workaround_ich8lan) was changed to fix a problem
> with iAMT connections dropping on interface down with jumbo frames set.
> Part of this change was to move the function call disabling the workaround
> to e1000e_down() from the e1000_setup_rctl() function.  The mechanic for
> disabling of this workaround involves writing several MAC and PHY registers
> back to hardware defaults.
>
> After this commit, when the driver is loaded with the cable out, the PHY
> registers are not programmed with the correct default values.  This causes
> the device to be capable of transmitting packets, but is unable to recieve
> them until this workaround is called.
>
> The flow of e1000e's open code relies upon calling the above workaround to
> expicitly program these registers either with jumbo frame appropriate settings
> or h/w defaults on 82579 and newer hardware.
>
> Fix this issue by adding logic to e1000_setup_rctl() that not only calls
> e1000_lv_jumbo_workaround_ich8lan() when jumbo frames are set, to enable the
> workaround, but also calls this function to explicitly disable the workaround
> in the case that jumbo frames are not set.
>
> Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
> Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Maybe a Reported-by tag from Thomas is missing?

Regards,

Fabio Estevam

Re: [net-next 01/13] e1000e: Fix no connectivity when driver loaded with cable out

[Thomas Gleixner]

On Tue, 1 Apr 2014, Fabio Estevam wrote:

> On Mon, Mar 31, 2014 at 8:34 PM, Jeff Kirsher
> <jeffrey.t.kirsher@intel.com> wrote:
> > From: David Ertman <davidx.m.ertman@intel.com>
> >
> > In commit da1e2046e5, the flow for enabling/disabling an Si errata
> > workaround (e1000_lv_jumbo_workaround_ich8lan) was changed to fix a problem
> > with iAMT connections dropping on interface down with jumbo frames set.
> > Part of this change was to move the function call disabling the workaround
> > to e1000e_down() from the e1000_setup_rctl() function.  The mechanic for
> > disabling of this workaround involves writing several MAC and PHY registers
> > back to hardware defaults.
> >
> > After this commit, when the driver is loaded with the cable out, the PHY
> > registers are not programmed with the correct default values.  This causes
> > the device to be capable of transmitting packets, but is unable to recieve
> > them until this workaround is called.
> >
> > The flow of e1000e's open code relies upon calling the above workaround to
> > expicitly program these registers either with jumbo frame appropriate settings
> > or h/w defaults on 82579 and newer hardware.
> >
> > Fix this issue by adding logic to e1000_setup_rctl() that not only calls
> > e1000_lv_jumbo_workaround_ich8lan() when jumbo frames are set, to enable the
> > workaround, but also calls this function to explicitly disable the workaround
> > in the case that jumbo frames are not set.
> >
> > Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
> > Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
> > Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> 
> Maybe a Reported-by tag from Thomas is missing?

Plus a Tested-by: me :)

Re: [net-next 01/13] e1000e: Fix no connectivity when driver loaded with cable out

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1315 bytes --]

On Mon, 2014-03-31 at 16:34 -0700, Jeff Kirsher wrote:
> From: David Ertman <davidx.m.ertman@intel.com>
> 
> In commit da1e2046e5, the flow for enabling/disabling an Si errata
> workaround (e1000_lv_jumbo_workaround_ich8lan) was changed to fix a problem
> with iAMT connections dropping on interface down with jumbo frames set.
> Part of this change was to move the function call disabling the workaround
> to e1000e_down() from the e1000_setup_rctl() function.  The mechanic for
> disabling of this workaround involves writing several MAC and PHY registers
> back to hardware defaults.
> 
> After this commit, when the driver is loaded with the cable out, the PHY
> registers are not programmed with the correct default values.  This causes
> the device to be capable of transmitting packets, but is unable to recieve
> them until this workaround is called.
[...]

Which devices are affected by this?  I just did a backport of e1000e
from 3.13 for Debian 7.5 and didn't spot this fix until later.

I couldn't reproduce this problem on my laptop, which has:

00:19.0 Ethernet controller [0200]: Intel Corporation 82579LM Gigabit Network Connection [8086:1502] (rev 04)

Ben.

-- 
Ben Hutchings
Lowery's Law:
             If it jams, force it. If it breaks, it needed replacing anyway.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 811 bytes --]

RE: [net-next 01/13] e1000e: Fix no connectivity when driver loaded with cable out

[Ertman, DavidX M]

> -----Original Message-----
> From: Ben Hutchings [mailto:ben@decadent.org.uk]
> Sent: Friday, April 18, 2014 6:16 AM
> To: Kirsher, Jeffrey T; Ertman, DavidX M
> Cc: davem@davemloft.net; netdev@vger.kernel.org; gospo@redhat.com;
> sassmann@redhat.com
> Subject: Re: [net-next 01/13] e1000e: Fix no connectivity when driver loaded
> with cable out
> 
> On Mon, 2014-03-31 at 16:34 -0700, Jeff Kirsher wrote:
> > From: David Ertman <davidx.m.ertman@intel.com>
> >
> > In commit da1e2046e5, the flow for enabling/disabling an Si errata
> > workaround (e1000_lv_jumbo_workaround_ich8lan) was changed to fix a
> problem
> > with iAMT connections dropping on interface down with jumbo frames set.
> > Part of this change was to move the function call disabling the workaround
> > to e1000e_down() from the e1000_setup_rctl() function.  The mechanic for
> > disabling of this workaround involves writing several MAC and PHY registers
> > back to hardware defaults.
> >
> > After this commit, when the driver is loaded with the cable out, the PHY
> > registers are not programmed with the correct default values.  This causes
> > the device to be capable of transmitting packets, but is unable to recieve
> > them until this workaround is called.
> [...]
> 
> Which devices are affected by this?  I just did a backport of e1000e
> from 3.13 for Debian 7.5 and didn't spot this fix until later.
> 
> I couldn't reproduce this problem on my laptop, which has:
> 
> 00:19.0 Ethernet controller [0200]: Intel Corporation 82579LM Gigabit
> Network Connection [8086:1502] (rev 04)
> 
> Ben.
> 
> --
> Ben Hutchings
> Lowery's Law:
>              If it jams, force it. If it breaks, it needed replacing anyway.

i217 parts are affected by this issue.

It is possible that 82579 parts are not affected (I don't think a repro was done with one of these devices), but the code involved was originally put in for an 82579 workaround.

Dave Ertman