[net 0/6][pull request] Intel Wired LAN Driver Updates

[net 0/6][pull request] Intel Wired LAN Driver Updates

[Jeff Kirsher]

This series contains updates to ixgbe and e1000e.

Jacob provides a ixgbe patch to fix the configure_rx patch to properly
disable RSC hardware logic when a user disables it.  Previously we only
disabled RSC in the queue settings, but this does not fully disable
hardware RSC logic which can lead to unexpected performance issues.

Emil provides three fixes for ixgbe.  First fixes the ethtool loopback
test when DCB is enabled, where the frames may be modified on Tx
(by adding VLAN tag) which will fail the check on receive.  Then a fix
for QSFP+ modules, limit the speed setting to advertise only one speed
at a time since the QSFP+ modules do not support auto negotiation.
Lastly, resolve an issue where the driver will display incorrect info
for QSFP+ modules that were inserted after the driver has been loaded.

David Ertman provides to fixes for e1000e, one removes a comparison to
the boolean value true where evaluating the lvalue will produce the
same result.  The other fixes an error in the calculation of the
rar_entry_count, which causes a write of unkown/undefined register
space in the MAC to unknown/undefined register space in the PHY.

The following are changes since commit 38463e2c290426262cd9a75fe66cbbe2925a68c2:
  net/mlx4_en: Check device state when setting coalescing
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net master

David Ertman (2):
  e1000e: cleanup boolean comparison to true
  e1000e: fix overrun of PHY RAR array

Emil Tantilov (3):
  ixgbe: fix ethtool loopback diagnostic with DCB enabled
  ixgbe: limit setting speed to only one at a time for QSFP modules
  ixgbe: fix ethtool reporting of supported links for SFP modules

Jacob Keller (1):
  ixgbe: fully disable hardware RSC logic when disabling RSC

 drivers/net/ethernet/intel/e1000e/ethtool.c      |  8 ++++++++
 drivers/net/ethernet/intel/e1000e/ich8lan.c      | 13 +++++++-----
 drivers/net/ethernet/intel/e1000e/ich8lan.h      |  2 +-
 drivers/net/ethernet/intel/e1000e/netdev.c       |  2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 25 ++++++++++++++++++++++++
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c    | 19 ++++++++++++++++--
 drivers/net/ethernet/intel/ixgbe/ixgbe_type.h    |  1 +
 7 files changed, 61 insertions(+), 9 deletions(-)

-- 
1.8.3.1

[net 1/6] ixgbe: fully disable hardware RSC logic when disabling RSC

[Jeff Kirsher]

From: Jacob Keller <jacob.e.keller@intel.com>

This patch modifies the configure_rx path in order to properly disable RSC
hardware logic when the user disables it. Previously we only disabled RSC in the
queue settings, but this does not fully disable hardware RSC logic which can
lead to some unexpected performance issues.

Signed-off-by: Jacob Keller <jacob.e.keller@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_main.c | 9 ++++++++-
 drivers/net/ethernet/intel/ixgbe/ixgbe_type.h | 1 +
 2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 7aba452..ae26786 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -3571,7 +3571,7 @@ static void ixgbe_configure_rx(struct ixgbe_adapter *adapter)
 {
 	struct ixgbe_hw *hw = &adapter->hw;
 	int i;
-	u32 rxctrl;
+	u32 rxctrl, rfctl;
 
 	/* disable receives while setting up the descriptors */
 	rxctrl = IXGBE_READ_REG(hw, IXGBE_RXCTRL);
@@ -3580,6 +3580,13 @@ static void ixgbe_configure_rx(struct ixgbe_adapter *adapter)
 	ixgbe_setup_psrtype(adapter);
 	ixgbe_setup_rdrxctl(adapter);
 
+	/* RSC Setup */
+	rfctl = IXGBE_READ_REG(hw, IXGBE_RFCTL);
+	rfctl &= ~IXGBE_RFCTL_RSC_DIS;
+	if (!(adapter->flags2 & IXGBE_FLAG2_RSC_ENABLED))
+		rfctl |= IXGBE_RFCTL_RSC_DIS;
+	IXGBE_WRITE_REG(hw, IXGBE_RFCTL, rfctl);
+
 	/* Program registers for the distribution of queues */
 	ixgbe_setup_mrqc(adapter);
 
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
index 6442cf8..10775cb 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
@@ -1861,6 +1861,7 @@ enum {
 #define IXGBE_RFCTL_ISCSI_DIS       0x00000001
 #define IXGBE_RFCTL_ISCSI_DWC_MASK  0x0000003E
 #define IXGBE_RFCTL_ISCSI_DWC_SHIFT 1
+#define IXGBE_RFCTL_RSC_DIS		0x00000020
 #define IXGBE_RFCTL_NFSW_DIS        0x00000040
 #define IXGBE_RFCTL_NFSR_DIS        0x00000080
 #define IXGBE_RFCTL_NFS_VER_MASK    0x00000300
-- 
1.8.3.1

[net 2/6] ixgbe: fix ethtool loopback diagnostic with DCB enabled

[Jeff Kirsher]

From: Emil Tantilov <emil.s.tantilov@intel.com>

This patch disables DCB prior to running the loopback test.
When DCB is enabled the frames may be modified on Tx (by adding vlan tag)
which will fail the check on Rx.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Jack Morgan <jack.morgan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index 0e1b973..08efc253 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -1805,6 +1805,10 @@ static int ixgbe_run_loopback_test(struct ixgbe_adapter *adapter)
 	unsigned int size = 1024;
 	netdev_tx_t tx_ret_val;
 	struct sk_buff *skb;
+	u32 flags_orig = adapter->flags;
+
+	/* DCB can modify the frames on Tx */
+	adapter->flags &= ~IXGBE_FLAG_DCB_ENABLED;
 
 	/* allocate test skb */
 	skb = alloc_skb(size, GFP_KERNEL);
@@ -1857,6 +1861,7 @@ static int ixgbe_run_loopback_test(struct ixgbe_adapter *adapter)
 
 	/* free the original skb */
 	kfree_skb(skb);
+	adapter->flags = flags_orig;
 
 	return ret_val;
 }
-- 
1.8.3.1

[net 3/6] ixgbe: limit setting speed to only one at a time for QSFP modules

[Jeff Kirsher]

From: Emil Tantilov <emil.s.tantilov@intel.com>

QSFP+ modules do not support auto negotiation and should advertise only
one speed at a time.

This patch adds logic in ethtool to allow setting and reporting the
advertised speed at either 1Gbps or 10Gbps, but not both. Also limits
the speed set in ixgbe_sfp_link_config_subtask() to highest supported.
Previously the link was set to whatever the supported speeds were.

Signed-off-by: Emil Tantilov <emil.s.tantilov@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_ethtool.c | 13 +++++++++++++
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c    | 10 +++++++++-
 2 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index 08efc253..b41db3b 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -186,6 +186,11 @@ static int ixgbe_get_settings(struct net_device *netdev,
 			ecmd->advertising |= ADVERTISED_1000baseT_Full;
 		if (supported_link & IXGBE_LINK_SPEED_100_FULL)
 			ecmd->advertising |= ADVERTISED_100baseT_Full;
+
+		if (hw->phy.multispeed_fiber && !autoneg) {
+			if (supported_link & IXGBE_LINK_SPEED_10GB_FULL)
+				ecmd->advertising = ADVERTISED_10000baseT_Full;
+		}
 	}
 
 	if (autoneg) {
@@ -314,6 +319,14 @@ static int ixgbe_set_settings(struct net_device *netdev,
 		if (ecmd->advertising & ~ecmd->supported)
 			return -EINVAL;
 
+		/* only allow one speed at a time if no autoneg */
+		if (!ecmd->autoneg && hw->phy.multispeed_fiber) {
+			if (ecmd->advertising ==
+			    (ADVERTISED_10000baseT_Full |
+			     ADVERTISED_1000baseT_Full))
+				return -EINVAL;
+		}
+
 		old = hw->phy.autoneg_advertised;
 		advertised = 0;
 		if (ecmd->advertising & ADVERTISED_10000baseT_Full)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index ae26786..0ade0cd 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -6000,8 +6000,16 @@ static void ixgbe_sfp_link_config_subtask(struct ixgbe_adapter *adapter)
 	adapter->flags &= ~IXGBE_FLAG_NEED_LINK_CONFIG;
 
 	speed = hw->phy.autoneg_advertised;
-	if ((!speed) && (hw->mac.ops.get_link_capabilities))
+	if ((!speed) && (hw->mac.ops.get_link_capabilities)) {
 		hw->mac.ops.get_link_capabilities(hw, &speed, &autoneg);
+
+		/* setup the highest link when no autoneg */
+		if (!autoneg) {
+			if (speed & IXGBE_LINK_SPEED_10GB_FULL)
+				speed = IXGBE_LINK_SPEED_10GB_FULL;
+		}
+	}
+
 	if (hw->mac.ops.setup_link)
 		hw->mac.ops.setup_link(hw, speed, true);
 
-- 
1.8.3.1

[net 4/6] ixgbe: fix ethtool reporting of supported links for SFP modules

[Jeff Kirsher]

From: Emil Tantilov <emil.s.tantilov@intel.com>

This patch resolves an issue where the driver will display incorrect info
for Q/SFP+ modules that were inserted after the driver has been loaded.

This patch adds a call to identify_phy() in ixgbe_get_settings() prior to
calling get_link_capabilities() which needs the PHY data in order to
determine the correct settings.

Signed-off-by: Emil Tantilov <emil.s.tantilov@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_ethtool.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index b41db3b..e8649ab 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -160,6 +160,13 @@ static int ixgbe_get_settings(struct net_device *netdev,
 	bool autoneg = false;
 	bool link_up;
 
+	/* SFP type is needed for get_link_capabilities */
+	if (hw->phy.media_type & (ixgbe_media_type_fiber |
+				  ixgbe_media_type_fiber_qsfp)) {
+		if (hw->phy.sfp_type == ixgbe_sfp_type_not_present)
+				hw->phy.ops.identify_sfp(hw);
+	}
+
 	hw->mac.ops.get_link_capabilities(hw, &supported_link, &autoneg);
 
 	/* set the supported link speeds */
-- 
1.8.3.1

[net 5/6] e1000e: cleanup boolean comparison to true

[Jeff Kirsher]

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

Removing a comparison to the boolean value true where simply interrogating
the lvalue will produce the same result.

Signed-off-by: David Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/e1000e/netdev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c
index e87e9b0..4ef7867 100644
--- a/drivers/net/ethernet/intel/e1000e/netdev.c
+++ b/drivers/net/ethernet/intel/e1000e/netdev.c
@@ -4868,7 +4868,7 @@ static void e1000_watchdog_task(struct work_struct *work)
 			 */
 			if ((hw->phy.type == e1000_phy_igp_3 ||
 			     hw->phy.type == e1000_phy_bm) &&
-			    (hw->mac.autoneg == true) &&
+			    hw->mac.autoneg &&
 			    (adapter->link_speed == SPEED_10 ||
 			     adapter->link_speed == SPEED_100) &&
 			    (adapter->link_duplex == HALF_DUPLEX)) {
-- 
1.8.3.1

[net 6/6] e1000e: fix overrun of PHY RAR array

[Jeff Kirsher]

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

When copying the MAC RAR registers to PHY there is an error in the
calculation of the rar_entry_count, which causes a write of unknown/
undefined register space in the MAC to unknown/undefined register space in
the PHY.

This patch fixes the overrun with writing to the PHY RAR and also fixes the
ethtool offline register tests so that the correctly addressed registers
have the appropriate bitmasks for R/W and RO bits for affected parts.

Shawn Rader gets credit for finding and fixing the register overrun.

Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
CC: Shawn Rader <shawn.t.rader@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/e1000e/ethtool.c |  8 ++++++++
 drivers/net/ethernet/intel/e1000e/ich8lan.c | 13 ++++++++-----
 drivers/net/ethernet/intel/e1000e/ich8lan.h |  2 +-
 3 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/intel/e1000e/ethtool.c b/drivers/net/ethernet/intel/e1000e/ethtool.c
index a8633b8..d14c8f5 100644
--- a/drivers/net/ethernet/intel/e1000e/ethtool.c
+++ b/drivers/net/ethernet/intel/e1000e/ethtool.c
@@ -922,6 +922,14 @@ static int e1000_reg_test(struct e1000_adapter *adapter, u64 *data)
 			else
 				mask &= ~(1 << 30);
 		}
+		if (mac->type == e1000_pch2lan) {
+			/* SHRAH[0,1,2] different than previous */
+			if (i == 7)
+				mask &= 0xFFF4FFFF;
+			/* SHRAH[3] different than SHRAH[0,1,2] */
+			if (i == 10)
+				mask |= (1 << 30);
+		}
 
 		REG_PATTERN_TEST_ARRAY(E1000_RA, ((i << 1) + 1), mask,
 				       0xFFFFFFFF);
diff --git a/drivers/net/ethernet/intel/e1000e/ich8lan.c b/drivers/net/ethernet/intel/e1000e/ich8lan.c
index af08188..42f0f67 100644
--- a/drivers/net/ethernet/intel/e1000e/ich8lan.c
+++ b/drivers/net/ethernet/intel/e1000e/ich8lan.c
@@ -1371,7 +1371,10 @@ static void e1000_rar_set_pch2lan(struct e1000_hw *hw, u8 *addr, u32 index)
 		return;
 	}
 
-	if (index < hw->mac.rar_entry_count) {
+	/* RAR[1-6] are owned by manageability.  Skip those and program the
+	 * next address into the SHRA register array.
+	 */
+	if (index < (u32)(hw->mac.rar_entry_count - 6)) {
 		s32 ret_val;
 
 		ret_val = e1000_acquire_swflag_ich8lan(hw);
@@ -1962,8 +1965,8 @@ void e1000_copy_rx_addrs_to_phy_ich8lan(struct e1000_hw *hw)
 	if (ret_val)
 		goto release;
 
-	/* Copy both RAL/H (rar_entry_count) and SHRAL/H (+4) to PHY */
-	for (i = 0; i < (hw->mac.rar_entry_count + 4); i++) {
+	/* Copy both RAL/H (rar_entry_count) and SHRAL/H to PHY */
+	for (i = 0; i < (hw->mac.rar_entry_count); i++) {
 		mac_reg = er32(RAL(i));
 		hw->phy.ops.write_reg_page(hw, BM_RAR_L(i),
 					   (u16)(mac_reg & 0xFFFF));
@@ -2007,10 +2010,10 @@ s32 e1000_lv_jumbo_workaround_ich8lan(struct e1000_hw *hw, bool enable)
 		return ret_val;
 
 	if (enable) {
-		/* Write Rx addresses (rar_entry_count for RAL/H, +4 for
+		/* Write Rx addresses (rar_entry_count for RAL/H, and
 		 * SHRAL/H) and initial CRC values to the MAC
 		 */
-		for (i = 0; i < (hw->mac.rar_entry_count + 4); i++) {
+		for (i = 0; i < hw->mac.rar_entry_count; i++) {
 			u8 mac_addr[ETH_ALEN] = { 0 };
 			u32 addr_high, addr_low;
 
diff --git a/drivers/net/ethernet/intel/e1000e/ich8lan.h b/drivers/net/ethernet/intel/e1000e/ich8lan.h
index 5986569..217090d 100644
--- a/drivers/net/ethernet/intel/e1000e/ich8lan.h
+++ b/drivers/net/ethernet/intel/e1000e/ich8lan.h
@@ -98,7 +98,7 @@
 #define PCIE_ICH8_SNOOP_ALL	PCIE_NO_SNOOP_ALL
 
 #define E1000_ICH_RAR_ENTRIES	7
-#define E1000_PCH2_RAR_ENTRIES	5	/* RAR[0], SHRA[0-3] */
+#define E1000_PCH2_RAR_ENTRIES	11      /* RAR[0-6], SHRA[0-3] */
 #define E1000_PCH_LPT_RAR_ENTRIES	12	/* RAR[0], SHRA[0-10] */
 
 #define PHY_PAGE_SHIFT		5
-- 
1.8.3.1

Re: [net 0/6][pull request] Intel Wired LAN Driver Updates

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Fri, 13 Sep 2013 14:12:04 -0700

> This series contains updates to ixgbe and e1000e.
> 
> Jacob provides a ixgbe patch to fix the configure_rx patch to properly
> disable RSC hardware logic when a user disables it.  Previously we only
> disabled RSC in the queue settings, but this does not fully disable
> hardware RSC logic which can lead to unexpected performance issues.
> 
> Emil provides three fixes for ixgbe.  First fixes the ethtool loopback
> test when DCB is enabled, where the frames may be modified on Tx
> (by adding VLAN tag) which will fail the check on receive.  Then a fix
> for QSFP+ modules, limit the speed setting to advertise only one speed
> at a time since the QSFP+ modules do not support auto negotiation.
> Lastly, resolve an issue where the driver will display incorrect info
> for QSFP+ modules that were inserted after the driver has been loaded.
> 
> David Ertman provides to fixes for e1000e, one removes a comparison to
> the boolean value true where evaluating the lvalue will produce the
> same result.  The other fixes an error in the calculation of the
> rar_entry_count, which causes a write of unkown/undefined register
> space in the MAC to unknown/undefined register space in the PHY.

Pulled, thanks Jeff.

Re: [net 6/6] e1000e: fix overrun of PHY RAR array

[Jason Wang]

On 09/14/2013 05:12 AM, Jeff Kirsher wrote:
> From: David Ertman <davidx.m.ertman@intel.com>
>
> When copying the MAC RAR registers to PHY there is an error in the
> calculation of the rar_entry_count, which causes a write of unknown/
> undefined register space in the MAC to unknown/undefined register space in
> the PHY.
>
> This patch fixes the overrun with writing to the PHY RAR and also fixes the
> ethtool offline register tests so that the correctly addressed registers
> have the appropriate bitmasks for R/W and RO bits for affected parts.
>
> Shawn Rader gets credit for finding and fixing the register overrun.
>
> Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
> CC: Shawn Rader <shawn.t.rader@intel.com>
> Tested-by: Aaron Brown <aaron.f.brown@intel.com>
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> ---
>  drivers/net/ethernet/intel/e1000e/ethtool.c |  8 ++++++++
>  drivers/net/ethernet/intel/e1000e/ich8lan.c | 13 ++++++++-----
>  drivers/net/ethernet/intel/e1000e/ich8lan.h |  2 +-
>  3 files changed, 17 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/e1000e/ethtool.c b/drivers/net/ethernet/intel/e1000e/ethtool.c
> index a8633b8..d14c8f5 100644
> --- a/drivers/net/ethernet/intel/e1000e/ethtool.c
> +++ b/drivers/net/ethernet/intel/e1000e/ethtool.c
> @@ -922,6 +922,14 @@ static int e1000_reg_test(struct e1000_adapter *adapter, u64 *data)
>  			else
>  				mask &= ~(1 << 30);
>  		}
> +		if (mac->type == e1000_pch2lan) {
> +			/* SHRAH[0,1,2] different than previous */
> +			if (i == 7)
> +				mask &= 0xFFF4FFFF;
> +			/* SHRAH[3] different than SHRAH[0,1,2] */
> +			if (i == 10)
> +				mask |= (1 << 30);
> +		}
>  
>  		REG_PATTERN_TEST_ARRAY(E1000_RA, ((i << 1) + 1), mask,
>  				       0xFFFFFFFF);
> diff --git a/drivers/net/ethernet/intel/e1000e/ich8lan.c b/drivers/net/ethernet/intel/e1000e/ich8lan.c
> index af08188..42f0f67 100644
> --- a/drivers/net/ethernet/intel/e1000e/ich8lan.c
> +++ b/drivers/net/ethernet/intel/e1000e/ich8lan.c
> @@ -1371,7 +1371,10 @@ static void e1000_rar_set_pch2lan(struct e1000_hw *hw, u8 *addr, u32 index)
>  		return;
>  	}
>  
> -	if (index < hw->mac.rar_entry_count) {
> +	/* RAR[1-6] are owned by manageability.  Skip those and program the
> +	 * next address into the SHRA register array.
> +	 */
> +	if (index < (u32)(hw->mac.rar_entry_count - 6)) {
>  		s32 ret_val;

This looks wrong. The caller writes the addresses in the reverse order,
so the first index we get here is E1000_PCH2_RAR_ENTRIES - 1 which
obviously fails the check. So nothing were write to RAR/SHA.

This breaks macvtap guest, and after I revert this patch, it works.

Also the comment above looks confused, it said the RAR[1-6] should be
skipped, but E1000_PCH2_RAR_ENTRIES takes it into account which may need
some explanation.
>  
>  		ret_val = e1000_acquire_swflag_ich8lan(hw);
> @@ -1962,8 +1965,8 @@ void e1000_copy_rx_addrs_to_phy_ich8lan(struct e1000_hw *hw)
>  	if (ret_val)
>  		goto release;
>  
> -	/* Copy both RAL/H (rar_entry_count) and SHRAL/H (+4) to PHY */
> -	for (i = 0; i < (hw->mac.rar_entry_count + 4); i++) {
> +	/* Copy both RAL/H (rar_entry_count) and SHRAL/H to PHY */
> +	for (i = 0; i < (hw->mac.rar_entry_count); i++) {
>  		mac_reg = er32(RAL(i));
>  		hw->phy.ops.write_reg_page(hw, BM_RAR_L(i),
>  					   (u16)(mac_reg & 0xFFFF));
> @@ -2007,10 +2010,10 @@ s32 e1000_lv_jumbo_workaround_ich8lan(struct e1000_hw *hw, bool enable)
>  		return ret_val;
>  
>  	if (enable) {
> -		/* Write Rx addresses (rar_entry_count for RAL/H, +4 for
> +		/* Write Rx addresses (rar_entry_count for RAL/H, and
>  		 * SHRAL/H) and initial CRC values to the MAC
>  		 */
> -		for (i = 0; i < (hw->mac.rar_entry_count + 4); i++) {
> +		for (i = 0; i < hw->mac.rar_entry_count; i++) {
>  			u8 mac_addr[ETH_ALEN] = { 0 };
>  			u32 addr_high, addr_low;
>  
> diff --git a/drivers/net/ethernet/intel/e1000e/ich8lan.h b/drivers/net/ethernet/intel/e1000e/ich8lan.h
> index 5986569..217090d 100644
> --- a/drivers/net/ethernet/intel/e1000e/ich8lan.h
> +++ b/drivers/net/ethernet/intel/e1000e/ich8lan.h
> @@ -98,7 +98,7 @@
>  #define PCIE_ICH8_SNOOP_ALL	PCIE_NO_SNOOP_ALL
>  
>  #define E1000_ICH_RAR_ENTRIES	7
> -#define E1000_PCH2_RAR_ENTRIES	5	/* RAR[0], SHRA[0-3] */
> +#define E1000_PCH2_RAR_ENTRIES	11      /* RAR[0-6], SHRA[0-3] */
>  #define E1000_PCH_LPT_RAR_ENTRIES	12	/* RAR[0], SHRA[0-10] */
>  
>  #define PHY_PAGE_SHIFT		5