Netdev List
 help / color / mirror / Atom feed
* Re: [PATCH 01/12] sfc: Remove redundant writes to INT_ADR_KER
From: David Miller @ 2009-11-30  0:47 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502123.3709.40.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:42:03 +0000

> This register only needs to be written after reset, not each time we
> enable interrupts.
> 
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 02/12] sfc: Remove duplicate hardware structure definitions
From: David Miller @ 2009-11-30  0:47 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502130.3709.41.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:42:10 +0000

> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 03/12] sfc: Turn pause frame generation on and off at the MAC, not the RX FIFO
From: David Miller @ 2009-11-30  0:47 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502138.3709.42.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:42:18 +0000

> Pause frame generation is gated by both RX_XOFF_MAC_EN and an enable
> bit in each MAC.  RX_XOFF_MAC_EN bit always reads back as 0 so we need
> to set it correctly every time we modify RX_CFG_REG.  Simplify this by
> always setting it to 1 and only changing the enable bits in the MACs.
> 
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 04/12] sfc: Move Falcon NIC operations to efx_nic_type
From: David Miller @ 2009-11-30  0:47 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502151.3709.43.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:42:31 +0000

> This is preparation for adding differing implementations for new NICs.
> 
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 05/12] sfc: Refactor link configuration
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502161.3709.44.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:42:41 +0000

> Refactor PHY, MAC and NIC configuration operations so that the
> existing link configuration can be re-pushed with:
 ...
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 06/12] sfc: Generalise link state monitoring
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502180.3709.45.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:00 +0000

> From: Steve Hodgson <shodgson@solarflare.com>
> 
> Use the efx_nic_type::monitor operation or event handling as
> appropriate.
> 
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 07/12] sfc: Add power-management and wake-on-LAN support
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502187.3709.46.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:07 +0000

> Wake-on-LAN is a stub for Falcon, but will be implemented fully for
> new NICs.
> 
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 08/12] sfc: Implement ethtool reset operation
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502195.3709.47.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:15 +0000

> Refactor efx_reset_down() and efx_reset_up() accordingly.
> 
> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 09/12] sfc: Add efx_nic_type operation for register self-test
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502203.3709.48.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:23 +0000

> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 10/12] sfc: Add efx_nic_type operation for NVRAM self-test
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502213.3709.49.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:33 +0000

> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 11/12] sfc: Add efx_nic_type operation for identity LED control
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502223.3709.50.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:43 +0000

> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Applied.

^ permalink raw reply

* Re: [PATCH 12/12] sfc: Separate shared NIC code from Falcon-specific and rename accordingly
From: David Miller @ 2009-11-30  0:48 UTC (permalink / raw)
  To: bhutchings; +Cc: netdev, linux-net-drivers
In-Reply-To: <1259502236.3709.51.camel@localhost>

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 29 Nov 2009 13:43:56 +0000

> Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Also applied, thanks a lot Ben.

^ permalink raw reply

* Re: [PATCH net-next-2.6 PATCH] net: Move && and || to end of previous line
From: David Miller @ 2009-11-30  0:55 UTC (permalink / raw)
  To: joe; +Cc: netdev
In-Reply-To: <1259528449.29779.194.camel@Joe-Laptop.home>

From: Joe Perches <joe@perches.com>
Date: Sun, 29 Nov 2009 13:00:49 -0800

> Not including net/atm/
> 
> Compiled tested x86 allyesconfig only
> Added a > 80 column line or two, which I ignored.
> Existing checkpatch plaints willfully, cheerfully ignored.
> 
> Signed-off-by: Joe Perches <joe@perches.com>

Applied, thanks a lot Joe.

^ permalink raw reply

* Re: mv643xx: convert to netdev_tx_t
From: David Miller @ 2009-11-30  1:04 UTC (permalink / raw)
  To: "kirjanov, kirjanov; +Cc: netdev
In-Reply-To: <20091129162123.GA17782@coldcone>


Applied, thank you.

^ permalink raw reply

* [PATCH 01/12] sfc: Fold falcon_probe_nic_variant() into falcon_probe_nic()
From: Ben Hutchings @ 2009-11-30  1:08 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

falcon_probe_nic_variant() does a lot less than it used to, and a
lot less than it claims to.  Fold the remainder into its caller.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/falcon.c |   64 +++++++++++++++++----------------------------
 1 files changed, 24 insertions(+), 40 deletions(-)

diff --git a/drivers/net/sfc/falcon.c b/drivers/net/sfc/falcon.c
index 2e4c711..63e6734 100644
--- a/drivers/net/sfc/falcon.c
+++ b/drivers/net/sfc/falcon.c
@@ -2806,40 +2806,6 @@ u32 efx_nic_fpga_ver(struct efx_nic *efx)
 	return EFX_OWORD_FIELD(altera_build, FRF_AZ_ALTERA_BUILD_VER);
 }
 
-/* Probe the NIC variant (revision, ASIC vs FPGA, function count, port
- * count, port speed).  Set workaround and feature flags accordingly.
- */
-static int falcon_probe_nic_variant(struct efx_nic *efx)
-{
-	efx_oword_t nic_stat;
-
-	if (efx_nic_fpga_ver(efx) != 0) {
-		EFX_ERR(efx, "Falcon FPGA not supported\n");
-		return -ENODEV;
-	}
-
-	efx_reado(efx, &nic_stat, FR_AB_NIC_STAT);
-
-	if (efx_nic_rev(efx) <= EFX_REV_FALCON_A1) {
-		u8 pci_rev = efx->pci_dev->revision;
-
-		if ((pci_rev == 0xff) || (pci_rev == 0)) {
-			EFX_ERR(efx, "Falcon rev A0 not supported\n");
-			return -ENODEV;
-		}
-		if (EFX_OWORD_FIELD(nic_stat, FRF_AB_STRAP_10G) == 0) {
-			EFX_ERR(efx, "Falcon rev A1 1G not supported\n");
-			return -ENODEV;
-		}
-		if (EFX_OWORD_FIELD(nic_stat, FRF_AA_STRAP_PCIE) == 0) {
-			EFX_ERR(efx, "Falcon rev A1 PCI-X not supported\n");
-			return -ENODEV;
-		}
-	}
-
-	return 0;
-}
-
 /* Probe all SPI devices on the NIC */
 static void falcon_probe_spi_devices(struct efx_nic *efx)
 {
@@ -2891,15 +2857,33 @@ static int falcon_probe_nic(struct efx_nic *efx)
 		return -ENOMEM;
 	efx->nic_data = nic_data;
 
-	/* Determine number of ports etc. */
-	rc = falcon_probe_nic_variant(efx);
-	if (rc)
+	rc = -ENODEV;
+
+	if (efx_nic_fpga_ver(efx) != 0) {
+		EFX_ERR(efx, "Falcon FPGA not supported\n");
 		goto fail1;
+	}
+
+	if (efx_nic_rev(efx) <= EFX_REV_FALCON_A1) {
+		efx_oword_t nic_stat;
+		struct pci_dev *dev;
+		u8 pci_rev = efx->pci_dev->revision;
 
-	/* Probe secondary function if expected */
-	if (efx_nic_is_dual_func(efx)) {
-		struct pci_dev *dev = pci_dev_get(efx->pci_dev);
+		if ((pci_rev == 0xff) || (pci_rev == 0)) {
+			EFX_ERR(efx, "Falcon rev A0 not supported\n");
+			goto fail1;
+		}
+		efx_reado(efx, &nic_stat, FR_AB_NIC_STAT);
+		if (EFX_OWORD_FIELD(nic_stat, FRF_AB_STRAP_10G) == 0) {
+			EFX_ERR(efx, "Falcon rev A1 1G not supported\n");
+			goto fail1;
+		}
+		if (EFX_OWORD_FIELD(nic_stat, FRF_AA_STRAP_PCIE) == 0) {
+			EFX_ERR(efx, "Falcon rev A1 PCI-X not supported\n");
+			goto fail1;
+		}
 
+		dev = pci_dev_get(efx->pci_dev);
 		while ((dev = pci_get_device(EFX_VENDID_SFC, FALCON_A_S_DEVID,
 					     dev))) {
 			if (dev->bus == efx->pci_dev->bus &&
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 02/12] sfc: Extend loopback mode enumeration
From: Ben Hutchings @ 2009-11-30  1:08 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

New NICs and PHYs support a wider variety of loopback modes.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/efx.c         |   19 +++++++-
 drivers/net/sfc/enum.h        |  106 +++++++++++++++++++++++++++++++++--------
 drivers/net/sfc/falcon_xmac.c |    2 +-
 drivers/net/sfc/mdio_10g.c    |    4 +-
 drivers/net/sfc/net_driver.h  |    2 +-
 drivers/net/sfc/qt202x_phy.c  |    2 +-
 drivers/net/sfc/tenxpress.c   |    4 +-
 7 files changed, 111 insertions(+), 28 deletions(-)

diff --git a/drivers/net/sfc/efx.c b/drivers/net/sfc/efx.c
index e5c33c6..75dcaae 100644
--- a/drivers/net/sfc/efx.c
+++ b/drivers/net/sfc/efx.c
@@ -36,15 +36,32 @@
 const unsigned int efx_loopback_mode_max = LOOPBACK_MAX;
 const char *efx_loopback_mode_names[] = {
 	[LOOPBACK_NONE]		= "NONE",
+	[LOOPBACK_DATA]		= "DATAPATH",
 	[LOOPBACK_GMAC]		= "GMAC",
 	[LOOPBACK_XGMII]	= "XGMII",
 	[LOOPBACK_XGXS]		= "XGXS",
 	[LOOPBACK_XAUI]  	= "XAUI",
+	[LOOPBACK_GMII] 	= "GMII",
+	[LOOPBACK_SGMII] 	= "SGMII",
+	[LOOPBACK_XGBR]		= "XGBR",
+	[LOOPBACK_XFI]		= "XFI",
+	[LOOPBACK_XAUI_FAR]	= "XAUI_FAR",
+	[LOOPBACK_GMII_FAR]	= "GMII_FAR",
+	[LOOPBACK_SGMII_FAR]	= "SGMII_FAR",
+	[LOOPBACK_XFI_FAR]	= "XFI_FAR",
 	[LOOPBACK_GPHY]		= "GPHY",
 	[LOOPBACK_PHYXS]	= "PHYXS",
 	[LOOPBACK_PCS]	 	= "PCS",
 	[LOOPBACK_PMAPMD] 	= "PMA/PMD",
-	[LOOPBACK_NETWORK]	= "NETWORK",
+	[LOOPBACK_XPORT]	= "XPORT",
+	[LOOPBACK_XGMII_WS]	= "XGMII_WS",
+	[LOOPBACK_XAUI_WS]  	= "XAUI_WS",
+	[LOOPBACK_XAUI_WS_FAR]  = "XAUI_WS_FAR",
+	[LOOPBACK_XAUI_WS_NEAR] = "XAUI_WS_NEAR",
+	[LOOPBACK_GMII_WS] 	= "GMII_WS",
+	[LOOPBACK_XFI_WS]	= "XFI_WS",
+	[LOOPBACK_XFI_WS_FAR]	= "XFI_WS_FAR",
+	[LOOPBACK_PHYXS_WS]  	= "PHYXS_WS",
 };
 
 /* Interrupt mode names (see INT_MODE())) */
diff --git a/drivers/net/sfc/enum.h b/drivers/net/sfc/enum.h
index fcd14b7..7a9e79a 100644
--- a/drivers/net/sfc/enum.h
+++ b/drivers/net/sfc/enum.h
@@ -13,38 +13,101 @@
 /**
  * enum efx_loopback_mode - loopback modes
  * @LOOPBACK_NONE: no loopback
- * @LOOPBACK_GMAC: loopback within GMAC at unspecified level
- * @LOOPBACK_XGMII: loopback within XMAC at XGMII level
- * @LOOPBACK_XGXS: loopback within XMAC at XGXS level
- * @LOOPBACK_XAUI: loopback within XMAC at XAUI level
+ * @LOOPBACK_DATA: data path loopback
+ * @LOOPBACK_GMAC: loopback within GMAC
+ * @LOOPBACK_XGMII: loopback after XMAC
+ * @LOOPBACK_XGXS: loopback within BPX after XGXS
+ * @LOOPBACK_XAUI: loopback within BPX before XAUI serdes
+ * @LOOPBACK_GMII: loopback within BPX after GMAC
+ * @LOOPBACK_SGMII: loopback within BPX within SGMII
+ * @LOOPBACK_XGBR: loopback within BPX within XGBR
+ * @LOOPBACK_XFI: loopback within BPX before XFI serdes
+ * @LOOPBACK_XAUI_FAR: loopback within BPX after XAUI serdes
+ * @LOOPBACK_GMII_FAR: loopback within BPX before SGMII
+ * @LOOPBACK_SGMII_FAR: loopback within BPX after SGMII
+ * @LOOPBACK_XFI_FAR: loopback after XFI serdes
  * @LOOPBACK_GPHY: loopback within 1G PHY at unspecified level
  * @LOOPBACK_PHYXS: loopback within 10G PHY at PHYXS level
  * @LOOPBACK_PCS: loopback within 10G PHY at PCS level
  * @LOOPBACK_PMAPMD: loopback within 10G PHY at PMAPMD level
- * @LOOPBACK_NETWORK: reflecting loopback (even further than furthest!)
+ * @LOOPBACK_XPORT: cross port loopback
+ * @LOOPBACK_XGMII_WS: wireside loopback excluding XMAC
+ * @LOOPBACK_XAUI_WS: wireside loopback within BPX within XAUI serdes
+ * @LOOPBACK_XAUI_WS_FAR: wireside loopback within BPX including XAUI serdes
+ * @LOOPBACK_XAUI_WS_NEAR: wireside loopback within BPX excluding XAUI serdes
+ * @LOOPBACK_GMII_WS: wireside loopback excluding GMAC
+ * @LOOPBACK_XFI_WS: wireside loopback excluding XFI serdes
+ * @LOOPBACK_XFI_WS_FAR: wireside loopback including XFI serdes
+ * @LOOPBACK_PHYXS_WS: wireside loopback within 10G PHY at PHYXS level
  */
-/* Please keep in order and up-to-date w.r.t the following two #defines */
+/* Please keep up-to-date w.r.t the following two #defines */
 enum efx_loopback_mode {
 	LOOPBACK_NONE = 0,
-	LOOPBACK_GMAC = 1,
-	LOOPBACK_XGMII = 2,
-	LOOPBACK_XGXS = 3,
-	LOOPBACK_XAUI = 4,
-	LOOPBACK_GPHY = 5,
-	LOOPBACK_PHYXS = 6,
-	LOOPBACK_PCS = 7,
-	LOOPBACK_PMAPMD = 8,
-	LOOPBACK_NETWORK = 9,
+	LOOPBACK_DATA = 1,
+	LOOPBACK_GMAC = 2,
+	LOOPBACK_XGMII = 3,
+	LOOPBACK_XGXS = 4,
+	LOOPBACK_XAUI = 5,
+	LOOPBACK_GMII = 6,
+	LOOPBACK_SGMII = 7,
+	LOOPBACK_XGBR = 8,
+	LOOPBACK_XFI = 9,
+	LOOPBACK_XAUI_FAR = 10,
+	LOOPBACK_GMII_FAR = 11,
+	LOOPBACK_SGMII_FAR = 12,
+	LOOPBACK_XFI_FAR = 13,
+	LOOPBACK_GPHY = 14,
+	LOOPBACK_PHYXS = 15,
+	LOOPBACK_PCS = 16,
+	LOOPBACK_PMAPMD = 17,
+	LOOPBACK_XPORT = 18,
+	LOOPBACK_XGMII_WS = 19,
+	LOOPBACK_XAUI_WS = 20,
+	LOOPBACK_XAUI_WS_FAR = 21,
+	LOOPBACK_XAUI_WS_NEAR = 22,
+	LOOPBACK_GMII_WS = 23,
+	LOOPBACK_XFI_WS = 24,
+	LOOPBACK_XFI_WS_FAR = 25,
+	LOOPBACK_PHYXS_WS = 26,
 	LOOPBACK_MAX
 };
-
 #define LOOPBACK_TEST_MAX LOOPBACK_PMAPMD
 
 /* These loopbacks occur within the controller */
-#define LOOPBACKS_INTERNAL ((1 << LOOPBACK_GMAC) |     \
-			    (1 << LOOPBACK_XGMII)|     \
-			    (1 << LOOPBACK_XGXS) |     \
-			    (1 << LOOPBACK_XAUI))
+#define LOOPBACKS_INTERNAL ((1 << LOOPBACK_DATA) |		\
+			    (1 << LOOPBACK_GMAC) |		\
+			    (1 << LOOPBACK_XGMII)|		\
+			    (1 << LOOPBACK_XGXS) |		\
+			    (1 << LOOPBACK_XAUI) |		\
+			    (1 << LOOPBACK_GMII) |		\
+			    (1 << LOOPBACK_SGMII) |		\
+			    (1 << LOOPBACK_SGMII) |		\
+			    (1 << LOOPBACK_XGBR) |		\
+			    (1 << LOOPBACK_XFI) |		\
+			    (1 << LOOPBACK_XAUI_FAR) |		\
+			    (1 << LOOPBACK_GMII_FAR) |		\
+			    (1 << LOOPBACK_SGMII_FAR) |		\
+			    (1 << LOOPBACK_XFI_FAR) |		\
+			    (1 << LOOPBACK_XGMII_WS) |		\
+			    (1 << LOOPBACK_XAUI_WS) |		\
+			    (1 << LOOPBACK_XAUI_WS_FAR) |	\
+			    (1 << LOOPBACK_XAUI_WS_NEAR) |	\
+			    (1 << LOOPBACK_GMII_WS) |		\
+			    (1 << LOOPBACK_XFI_WS) |		\
+			    (1 << LOOPBACK_XFI_WS_FAR))
+
+#define LOOPBACKS_WS ((1 << LOOPBACK_XGMII_WS) |		\
+		      (1 << LOOPBACK_XAUI_WS) |			\
+		      (1 << LOOPBACK_XAUI_WS_FAR) |		\
+		      (1 << LOOPBACK_XAUI_WS_NEAR) |		\
+		      (1 << LOOPBACK_GMII_WS) |			\
+		      (1 << LOOPBACK_XFI_WS) |			\
+		      (1 << LOOPBACK_XFI_WS_FAR) |		\
+		      (1 << LOOPBACK_PHYXS_WS))
+
+#define LOOPBACKS_EXTERNAL(_efx)					\
+	((_efx)->loopback_modes & ~LOOPBACKS_INTERNAL &			\
+	 ~(1 << LOOPBACK_NONE))
 
 #define LOOPBACK_MASK(_efx)			\
 	(1 << (_efx)->loopback_mode)
@@ -52,6 +115,9 @@ enum efx_loopback_mode {
 #define LOOPBACK_INTERNAL(_efx)				\
 	(!!(LOOPBACKS_INTERNAL & LOOPBACK_MASK(_efx)))
 
+#define LOOPBACK_EXTERNAL(_efx)				\
+	(!!(LOOPBACK_MASK(_efx) & LOOPBACKS_EXTERNAL(_efx)))
+
 #define LOOPBACK_CHANGED(_from, _to, _mask)				\
 	(!!((LOOPBACK_MASK(_from) ^ LOOPBACK_MASK(_to)) & (_mask)))
 
diff --git a/drivers/net/sfc/falcon_xmac.c b/drivers/net/sfc/falcon_xmac.c
index 898afc1..83da792 100644
--- a/drivers/net/sfc/falcon_xmac.c
+++ b/drivers/net/sfc/falcon_xmac.c
@@ -249,7 +249,7 @@ static bool falcon_check_xaui_link_up(struct efx_nic *efx, int tries)
 {
 	bool mac_up = falcon_xaui_link_ok(efx);
 
-	if ((efx->loopback_mode == LOOPBACK_NETWORK) ||
+	if (LOOPBACK_MASK(efx) & LOOPBACKS_EXTERNAL(efx) & LOOPBACKS_WS ||
 	    efx_phy_mode_disabled(efx->phy_mode))
 		/* XAUI link is expected to be down */
 		return mac_up;
diff --git a/drivers/net/sfc/mdio_10g.c b/drivers/net/sfc/mdio_10g.c
index 1f62a5c..e6ca988 100644
--- a/drivers/net/sfc/mdio_10g.c
+++ b/drivers/net/sfc/mdio_10g.c
@@ -173,7 +173,7 @@ bool efx_mdio_links_ok(struct efx_nic *efx, unsigned int mmd_mask)
 	 * of mmd's */
 	if (LOOPBACK_INTERNAL(efx))
 		return true;
-	else if (efx->loopback_mode == LOOPBACK_NETWORK)
+	else if (LOOPBACK_MASK(efx) & LOOPBACKS_WS)
 		return false;
 	else if (efx_phy_mode_disabled(efx->phy_mode))
 		return false;
@@ -210,7 +210,7 @@ void efx_mdio_phy_reconfigure(struct efx_nic *efx)
 			  efx->loopback_mode == LOOPBACK_PCS);
 	efx_mdio_set_flag(efx, MDIO_MMD_PHYXS,
 			  MDIO_CTRL1, MDIO_PHYXS_CTRL1_LOOPBACK,
-			  efx->loopback_mode == LOOPBACK_NETWORK);
+			  efx->loopback_mode == LOOPBACK_PHYXS_WS);
 }
 
 static void efx_mdio_set_mmd_lpower(struct efx_nic *efx,
diff --git a/drivers/net/sfc/net_driver.h b/drivers/net/sfc/net_driver.h
index 62b2089..cd2debb 100644
--- a/drivers/net/sfc/net_driver.h
+++ b/drivers/net/sfc/net_driver.h
@@ -824,7 +824,7 @@ struct efx_nic {
 
 	atomic_t rx_reset;
 	enum efx_loopback_mode loopback_mode;
-	unsigned int loopback_modes;
+	u64 loopback_modes;
 
 	void *loopback_selftest;
 };
diff --git a/drivers/net/sfc/qt202x_phy.c b/drivers/net/sfc/qt202x_phy.c
index 4c38516..49a5ab5 100644
--- a/drivers/net/sfc/qt202x_phy.c
+++ b/drivers/net/sfc/qt202x_phy.c
@@ -23,7 +23,7 @@
 
 #define QT202X_LOOPBACKS ((1 << LOOPBACK_PCS) |		\
 			  (1 << LOOPBACK_PMAPMD) |	\
-			  (1 << LOOPBACK_NETWORK))
+			  (1 << LOOPBACK_PHYXS_WS))
 
 /****************************************************************************/
 /* Quake-specific MDIO registers */
diff --git a/drivers/net/sfc/tenxpress.c b/drivers/net/sfc/tenxpress.c
index c301853..0dfb227 100644
--- a/drivers/net/sfc/tenxpress.c
+++ b/drivers/net/sfc/tenxpress.c
@@ -30,13 +30,13 @@
 #define SFX7101_LOOPBACKS ((1 << LOOPBACK_PHYXS) |	\
 			   (1 << LOOPBACK_PCS) |	\
 			   (1 << LOOPBACK_PMAPMD) |	\
-			   (1 << LOOPBACK_NETWORK))
+			   (1 << LOOPBACK_PHYXS_WS))
 
 #define SFT9001_LOOPBACKS ((1 << LOOPBACK_GPHY) |	\
 			   (1 << LOOPBACK_PHYXS) |	\
 			   (1 << LOOPBACK_PCS) |	\
 			   (1 << LOOPBACK_PMAPMD) |	\
-			   (1 << LOOPBACK_NETWORK))
+			   (1 << LOOPBACK_PHYXS_WS))
 
 /* We complain if we fail to see the link partner as 10G capable this many
  * times in a row (must be > 1 as sampling the autoneg. registers is racy)
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 03/12] sfc: Remove static PHY data and enumerations
From: Ben Hutchings @ 2009-11-30  1:08 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

New NICs have firmware managing the PHY, and we will discover the PHY
capabilities at run-time.  Replace the static data with probe() and
test_name() operations.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/ethtool.c     |   18 ++++++++++++--
 drivers/net/sfc/falcon.c      |   15 +++---------
 drivers/net/sfc/falcon.h      |   20 ++++++++++++++++
 drivers/net/sfc/falcon_xmac.c |    2 +-
 drivers/net/sfc/mdio_10g.c    |    4 +-
 drivers/net/sfc/net_driver.h  |   35 +++++------------------------
 drivers/net/sfc/qt202x_phy.c  |   12 +++++++--
 drivers/net/sfc/selftest.c    |    5 +---
 drivers/net/sfc/tenxpress.c   |   49 +++++++++++++++++++++++++++++++----------
 9 files changed, 95 insertions(+), 65 deletions(-)

diff --git a/drivers/net/sfc/ethtool.c b/drivers/net/sfc/ethtool.c
index e86cbca..4fe8740 100644
--- a/drivers/net/sfc/ethtool.c
+++ b/drivers/net/sfc/ethtool.c
@@ -365,9 +365,21 @@ static int efx_ethtool_fill_self_tests(struct efx_nic *efx,
 	efx_fill_test(n++, strings, data, &tests->registers,
 		      "core", 0, "registers", NULL);
 
-	for (i = 0; i < efx->phy_op->num_tests; i++)
-		efx_fill_test(n++, strings, data, &tests->phy[i],
-			      "phy", 0, efx->phy_op->test_names[i], NULL);
+	if (efx->phy_op->run_tests != NULL) {
+		EFX_BUG_ON_PARANOID(efx->phy_op->test_name == NULL);
+
+		for (i = 0; true; ++i) {
+			const char *name;
+
+			EFX_BUG_ON_PARANOID(i >= EFX_MAX_PHY_TESTS);
+			name = efx->phy_op->test_name(efx, i);
+			if (name == NULL)
+				break;
+
+			efx_fill_test(n++, strings, data, &tests->phy[i],
+				      "phy", 0, name, NULL);
+		}
+	}
 
 	/* Loopback tests */
 	for (mode = LOOPBACK_NONE; mode <= LOOPBACK_TEST_MAX; mode++) {
diff --git a/drivers/net/sfc/falcon.c b/drivers/net/sfc/falcon.c
index 63e6734..29d4537 100644
--- a/drivers/net/sfc/falcon.c
+++ b/drivers/net/sfc/falcon.c
@@ -2291,19 +2291,12 @@ static int falcon_probe_port(struct efx_nic *efx)
 		return -ENODEV;
 	}
 
-	if (efx->phy_op->macs & EFX_XMAC)
-		efx->loopback_modes |= ((1 << LOOPBACK_XGMII) |
-					(1 << LOOPBACK_XGXS) |
-					(1 << LOOPBACK_XAUI));
-	if (efx->phy_op->macs & EFX_GMAC)
-		efx->loopback_modes |= (1 << LOOPBACK_GMAC);
-	efx->loopback_modes |= efx->phy_op->loopbacks;
-
-	/* Set up MDIO structure for PHY */
-	efx->mdio.mmds = efx->phy_op->mmds;
-	efx->mdio.mode_support = MDIO_SUPPORTS_C45 | MDIO_EMULATE_C22;
+	/* Fill out MDIO structure and loopback modes */
 	efx->mdio.mdio_read = falcon_mdio_read;
 	efx->mdio.mdio_write = falcon_mdio_write;
+	rc = efx->phy_op->probe(efx);
+	if (rc != 0)
+		return rc;
 
 	/* Initial assumption */
 	efx->link_state.speed = 10000;
diff --git a/drivers/net/sfc/falcon.h b/drivers/net/sfc/falcon.h
index 875b58e..3085ecf 100644
--- a/drivers/net/sfc/falcon.h
+++ b/drivers/net/sfc/falcon.h
@@ -38,6 +38,26 @@ static inline bool efx_nic_is_dual_func(struct efx_nic *efx)
 	return efx_nic_rev(efx) < EFX_REV_FALCON_B0;
 }
 
+enum {
+	PHY_TYPE_NONE = 0,
+	PHY_TYPE_TXC43128 = 1,
+	PHY_TYPE_88E1111 = 2,
+	PHY_TYPE_SFX7101 = 3,
+	PHY_TYPE_QT2022C2 = 4,
+	PHY_TYPE_PM8358 = 6,
+	PHY_TYPE_SFT9001A = 8,
+	PHY_TYPE_QT2025C = 9,
+	PHY_TYPE_SFT9001B = 10,
+};
+
+#define FALCON_XMAC_LOOPBACKS			\
+	((1 << LOOPBACK_XGMII) |		\
+	 (1 << LOOPBACK_XGXS) |			\
+	 (1 << LOOPBACK_XAUI))
+
+#define FALCON_GMAC_LOOPBACKS			\
+	(1 << LOOPBACK_GMAC)
+
 /**
  * struct falcon_board_type - board operations and type information
  * @id: Board type id, as found in NVRAM
diff --git a/drivers/net/sfc/falcon_xmac.c b/drivers/net/sfc/falcon_xmac.c
index 83da792..643622d 100644
--- a/drivers/net/sfc/falcon_xmac.c
+++ b/drivers/net/sfc/falcon_xmac.c
@@ -137,7 +137,7 @@ static bool falcon_xaui_link_ok(struct efx_nic *efx)
 
 	/* If the link is up, then check the phy side of the xaui link */
 	if (efx->link_state.up && link_ok)
-		if (efx->phy_op->mmds & (1 << MDIO_MMD_PHYXS))
+		if (efx->mdio.mmds & (1 << MDIO_MMD_PHYXS))
 			link_ok = efx_mdio_phyxgxs_lane_sync(efx);
 
 	return link_ok;
diff --git a/drivers/net/sfc/mdio_10g.c b/drivers/net/sfc/mdio_10g.c
index e6ca988..20e6274 100644
--- a/drivers/net/sfc/mdio_10g.c
+++ b/drivers/net/sfc/mdio_10g.c
@@ -15,6 +15,7 @@
 #include "net_driver.h"
 #include "mdio_10g.h"
 #include "workarounds.h"
+#include "falcon.h"
 
 unsigned efx_mdio_id_oui(u32 id)
 {
@@ -312,8 +313,7 @@ void efx_mdio_an_reconfigure(struct efx_nic *efx)
 	/* Enable and restart AN */
 	reg = efx_mdio_read(efx, MDIO_MMD_AN, MDIO_CTRL1);
 	reg |= MDIO_AN_CTRL1_ENABLE;
-	if (!(EFX_WORKAROUND_15195(efx) &&
-	      LOOPBACK_MASK(efx) & efx->phy_op->loopbacks))
+	if (!(EFX_WORKAROUND_15195(efx) && LOOPBACK_EXTERNAL(efx)))
 		reg |= MDIO_AN_CTRL1_RESTART;
 	if (xnp)
 		reg |= MDIO_AN_CTRL1_XNP;
diff --git a/drivers/net/sfc/net_driver.h b/drivers/net/sfc/net_driver.h
index cd2debb..452f835 100644
--- a/drivers/net/sfc/net_driver.h
+++ b/drivers/net/sfc/net_driver.h
@@ -428,19 +428,6 @@ enum efx_int_mode {
 };
 #define EFX_INT_MODE_USE_MSI(x) (((x)->interrupt_mode) <= EFX_INT_MODE_MSI)
 
-enum phy_type {
-	PHY_TYPE_NONE = 0,
-	PHY_TYPE_TXC43128 = 1,
-	PHY_TYPE_88E1111 = 2,
-	PHY_TYPE_SFX7101 = 3,
-	PHY_TYPE_QT2022C2 = 4,
-	PHY_TYPE_PM8358 = 6,
-	PHY_TYPE_SFT9001A = 8,
-	PHY_TYPE_QT2025C = 9,
-	PHY_TYPE_SFT9001B = 10,
-	PHY_TYPE_MAX	/* Insert any new items before this */
-};
-
 #define EFX_IS10G(efx) ((efx)->link_state.speed == 10000)
 
 enum nic_state {
@@ -483,12 +470,6 @@ enum efx_fc_type {
 	EFX_FC_AUTO = 4,
 };
 
-/* Supported MAC bit-mask */
-enum efx_mac_type {
-	EFX_GMAC = 1,
-	EFX_XMAC = 2,
-};
-
 /**
  * struct efx_link_state - Current state of the link
  * @up: Link is up
@@ -524,6 +505,8 @@ struct efx_mac_operations {
 
 /**
  * struct efx_phy_operations - Efx PHY operations table
+ * @probe: Probe PHY and initialise efx->mdio.mode_support, efx->mdio.mmds,
+ *	efx->loopback_modes.
  * @init: Initialise PHY
  * @fini: Shut down PHY
  * @reconfigure: Reconfigure PHY (e.g. for new link parameters)
@@ -533,15 +516,12 @@ struct efx_mac_operations {
  * @set_settings: Set ethtool settings. Serialised by the mac_lock.
  * @set_npage_adv: Set abilities advertised in (Extended) Next Page
  *	(only needed where AN bit is set in mmds)
- * @num_tests: Number of PHY-specific tests/results
- * @test_names: Names of the tests/results
+ * @test_name: Get the name of a PHY-specific test/result
  * @run_tests: Run tests and record results as appropriate.
  *	Flags are the ethtool tests flags.
- * @mmds: MMD presence mask
- * @loopbacks: Supported loopback modes mask
  */
 struct efx_phy_operations {
-	enum efx_mac_type macs;
+	int (*probe) (struct efx_nic *efx);
 	int (*init) (struct efx_nic *efx);
 	void (*fini) (struct efx_nic *efx);
 	int (*reconfigure) (struct efx_nic *efx);
@@ -551,11 +531,8 @@ struct efx_phy_operations {
 	int (*set_settings) (struct efx_nic *efx,
 			     struct ethtool_cmd *ecmd);
 	void (*set_npage_adv) (struct efx_nic *efx, u32);
-	u32 num_tests;
-	const char *const *test_names;
+	const char *(*test_name) (struct efx_nic *efx, unsigned int index);
 	int (*run_tests) (struct efx_nic *efx, int *results, unsigned flags);
-	int mmds;
-	unsigned loopbacks;
 };
 
 /**
@@ -806,7 +783,7 @@ struct efx_nic {
 	struct efx_mac_operations *mac_op;
 	unsigned char mac_address[ETH_ALEN];
 
-	enum phy_type phy_type;
+	unsigned int phy_type;
 	struct mutex mdio_lock;
 	struct efx_phy_operations *phy_op;
 	void *phy_data;
diff --git a/drivers/net/sfc/qt202x_phy.c b/drivers/net/sfc/qt202x_phy.c
index 49a5ab5..22b0e89 100644
--- a/drivers/net/sfc/qt202x_phy.c
+++ b/drivers/net/sfc/qt202x_phy.c
@@ -135,6 +135,14 @@ static int qt202x_reset_phy(struct efx_nic *efx)
 	return rc;
 }
 
+static int qt202x_phy_probe(struct efx_nic *efx)
+{
+	efx->mdio.mmds = QT202X_REQUIRED_DEVS;
+	efx->mdio.mode_support = MDIO_SUPPORTS_C45 | MDIO_EMULATE_C22;
+	efx->loopback_modes = QT202X_LOOPBACKS | FALCON_XMAC_LOOPBACKS;
+	return 0;
+}
+
 static int qt202x_phy_init(struct efx_nic *efx)
 {
 	struct qt202x_phy_data *phy_data;
@@ -224,13 +232,11 @@ static void qt202x_phy_fini(struct efx_nic *efx)
 }
 
 struct efx_phy_operations falcon_qt202x_phy_ops = {
-	.macs		 = EFX_XMAC,
+	.probe		 = qt202x_phy_probe,
 	.init		 = qt202x_phy_init,
 	.reconfigure	 = qt202x_phy_reconfigure,
 	.poll	     	 = qt202x_phy_poll,
 	.fini	  	 = qt202x_phy_fini,
 	.get_settings	 = qt202x_phy_get_settings,
 	.set_settings	 = efx_mdio_set_settings,
-	.mmds            = QT202X_REQUIRED_DEVS,
-	.loopbacks       = QT202X_LOOPBACKS,
 };
diff --git a/drivers/net/sfc/selftest.c b/drivers/net/sfc/selftest.c
index 9a24053..16258d8 100644
--- a/drivers/net/sfc/selftest.c
+++ b/drivers/net/sfc/selftest.c
@@ -100,7 +100,7 @@ static int efx_test_mdio(struct efx_nic *efx, struct efx_self_tests *tests)
 	}
 
 	if (EFX_IS10G(efx)) {
-		rc = efx_mdio_check_mmds(efx, efx->phy_op->mmds, 0);
+		rc = efx_mdio_check_mmds(efx, efx->mdio.mmds, 0);
 		if (rc)
 			goto out;
 	}
@@ -253,9 +253,6 @@ static int efx_test_phy(struct efx_nic *efx, struct efx_self_tests *tests,
 	if (!efx->phy_op->run_tests)
 		return 0;
 
-	EFX_BUG_ON_PARANOID(efx->phy_op->num_tests == 0 ||
-			    efx->phy_op->num_tests > EFX_MAX_PHY_TESTS);
-
 	mutex_lock(&efx->mac_lock);
 	rc = efx->phy_op->run_tests(efx, tests->phy, flags);
 	mutex_unlock(&efx->mac_lock);
diff --git a/drivers/net/sfc/tenxpress.c b/drivers/net/sfc/tenxpress.c
index 0dfb227..8de97a9 100644
--- a/drivers/net/sfc/tenxpress.c
+++ b/drivers/net/sfc/tenxpress.c
@@ -298,6 +298,23 @@ static int tenxpress_init(struct efx_nic *efx)
 	return 0;
 }
 
+static int sfx7101_phy_probe(struct efx_nic *efx)
+{
+	efx->mdio.mmds = TENXPRESS_REQUIRED_DEVS;
+	efx->mdio.mode_support = MDIO_SUPPORTS_C45 | MDIO_EMULATE_C22;
+	efx->loopback_modes = SFX7101_LOOPBACKS | FALCON_XMAC_LOOPBACKS;
+	return 0;
+}
+
+static int sft9001_phy_probe(struct efx_nic *efx)
+{
+	efx->mdio.mmds = TENXPRESS_REQUIRED_DEVS;
+	efx->mdio.mode_support = MDIO_SUPPORTS_C45 | MDIO_EMULATE_C22;
+	efx->loopback_modes = (SFT9001_LOOPBACKS | FALCON_XMAC_LOOPBACKS |
+			       FALCON_GMAC_LOOPBACKS);
+	return 0;
+}
+
 static int tenxpress_phy_init(struct efx_nic *efx)
 {
 	struct tenxpress_phy_data *phy_data;
@@ -512,7 +529,7 @@ static int tenxpress_phy_reconfigure(struct efx_nic *efx)
 
 	phy_mode_change = (efx->phy_mode == PHY_MODE_NORMAL &&
 			   phy_data->phy_mode != PHY_MODE_NORMAL);
-	loop_reset = (LOOPBACK_OUT_OF(phy_data, efx, efx->phy_op->loopbacks) ||
+	loop_reset = (LOOPBACK_OUT_OF(phy_data, efx, LOOPBACKS_EXTERNAL(efx)) ||
 		      LOOPBACK_CHANGED(phy_data, efx, 1 << LOOPBACK_GPHY));
 
 	if (loop_reset || phy_mode_change) {
@@ -627,6 +644,13 @@ static const char *const sfx7101_test_names[] = {
 	"bist"
 };
 
+static const char *sfx7101_test_name(struct efx_nic *efx, unsigned int index)
+{
+	if (index < ARRAY_SIZE(sfx7101_test_names))
+		return sfx7101_test_names[index];
+	return NULL;
+}
+
 static int
 sfx7101_run_tests(struct efx_nic *efx, int *results, unsigned flags)
 {
@@ -656,6 +680,13 @@ static const char *const sft9001_test_names[] = {
 	"cable.pairD.length",
 };
 
+static const char *sft9001_test_name(struct efx_nic *efx, unsigned int index)
+{
+	if (index < ARRAY_SIZE(sft9001_test_names))
+		return sft9001_test_names[index];
+	return NULL;
+}
+
 static int sft9001_run_tests(struct efx_nic *efx, int *results, unsigned flags)
 {
 	int rc = 0, rc2, i, ctrl_reg, res_reg;
@@ -758,7 +789,7 @@ tenxpress_get_settings(struct efx_nic *efx, struct ethtool_cmd *ecmd)
 	 * but doesn't advertise the correct speed. So override it */
 	if (efx->loopback_mode == LOOPBACK_GPHY)
 		ecmd->speed = SPEED_1000;
-	else if (LOOPBACK_MASK(efx) & efx->phy_op->loopbacks)
+	else if (LOOPBACK_EXTERNAL(efx))
 		ecmd->speed = SPEED_10000;
 }
 
@@ -788,7 +819,7 @@ static void sft9001_set_npage_adv(struct efx_nic *efx, u32 advertising)
 }
 
 struct efx_phy_operations falcon_sfx7101_phy_ops = {
-	.macs		  = EFX_XMAC,
+	.probe		  = sfx7101_phy_probe,
 	.init             = tenxpress_phy_init,
 	.reconfigure      = tenxpress_phy_reconfigure,
 	.poll             = tenxpress_phy_poll,
@@ -796,15 +827,12 @@ struct efx_phy_operations falcon_sfx7101_phy_ops = {
 	.get_settings	  = tenxpress_get_settings,
 	.set_settings	  = tenxpress_set_settings,
 	.set_npage_adv    = sfx7101_set_npage_adv,
-	.num_tests	  = ARRAY_SIZE(sfx7101_test_names),
-	.test_names	  = sfx7101_test_names,
+	.test_name	  = sfx7101_test_name,
 	.run_tests	  = sfx7101_run_tests,
-	.mmds             = TENXPRESS_REQUIRED_DEVS,
-	.loopbacks        = SFX7101_LOOPBACKS,
 };
 
 struct efx_phy_operations falcon_sft9001_phy_ops = {
-	.macs		  = EFX_GMAC | EFX_XMAC,
+	.probe		  = sft9001_phy_probe,
 	.init             = tenxpress_phy_init,
 	.reconfigure      = tenxpress_phy_reconfigure,
 	.poll             = tenxpress_phy_poll,
@@ -812,9 +840,6 @@ struct efx_phy_operations falcon_sft9001_phy_ops = {
 	.get_settings	  = tenxpress_get_settings,
 	.set_settings	  = tenxpress_set_settings,
 	.set_npage_adv    = sft9001_set_npage_adv,
-	.num_tests	  = ARRAY_SIZE(sft9001_test_names),
-	.test_names	  = sft9001_test_names,
+	.test_name	  = sft9001_test_name,
 	.run_tests	  = sft9001_run_tests,
-	.mmds             = TENXPRESS_REQUIRED_DEVS,
-	.loopbacks        = SFT9001_LOOPBACKS,
 };
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 04/12] sfc: Extend MTD driver for use with new NICs
From: Ben Hutchings @ 2009-11-30  1:10 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-mtd, linux-net-drivers

In new NICs flash is managed by firmware and we will use high-level
operations on partitions rather than direct SPI commands.  Add support
for multiple MTD partitions per flash device and remove the direct
link between MTD and SPI devices.  Maintain a list of MTD partitions
in struct efx_nic.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/efx.c        |    3 +
 drivers/net/sfc/ethtool.c    |    6 +-
 drivers/net/sfc/falcon.c     |   32 ++---
 drivers/net/sfc/mtd.c        |  312 +++++++++++++++++++++++++++++-------------
 drivers/net/sfc/net_driver.h |    8 +-
 drivers/net/sfc/spi.h        |   18 +--
 6 files changed, 255 insertions(+), 124 deletions(-)

diff --git a/drivers/net/sfc/efx.c b/drivers/net/sfc/efx.c
index 75dcaae..4fe6d63 100644
--- a/drivers/net/sfc/efx.c
+++ b/drivers/net/sfc/efx.c
@@ -1977,6 +1977,9 @@ static int efx_init_struct(struct efx_nic *efx, struct efx_nic_type *type,
 	spin_lock_init(&efx->biu_lock);
 	mutex_init(&efx->mdio_lock);
 	mutex_init(&efx->spi_lock);
+#ifdef CONFIG_SFC_MTD
+	INIT_LIST_HEAD(&efx->mtd_list);
+#endif
 	INIT_WORK(&efx->reset_work, efx_reset_work);
 	INIT_DELAYED_WORK(&efx->monitor_work, efx_monitor);
 	efx->pci_dev = pci_dev;
diff --git a/drivers/net/sfc/ethtool.c b/drivers/net/sfc/ethtool.c
index 4fe8740..08a9db9 100644
--- a/drivers/net/sfc/ethtool.c
+++ b/drivers/net/sfc/ethtool.c
@@ -567,7 +567,8 @@ static int efx_ethtool_get_eeprom(struct net_device *net_dev,
 	rc = mutex_lock_interruptible(&efx->spi_lock);
 	if (rc)
 		return rc;
-	rc = falcon_spi_read(spi, eeprom->offset + EFX_EEPROM_BOOTCONFIG_START,
+	rc = falcon_spi_read(efx, spi,
+			     eeprom->offset + EFX_EEPROM_BOOTCONFIG_START,
 			     eeprom->len, &len, buf);
 	mutex_unlock(&efx->spi_lock);
 
@@ -590,7 +591,8 @@ static int efx_ethtool_set_eeprom(struct net_device *net_dev,
 	rc = mutex_lock_interruptible(&efx->spi_lock);
 	if (rc)
 		return rc;
-	rc = falcon_spi_write(spi, eeprom->offset + EFX_EEPROM_BOOTCONFIG_START,
+	rc = falcon_spi_write(efx, spi,
+			      eeprom->offset + EFX_EEPROM_BOOTCONFIG_START,
 			      eeprom->len, &len, buf);
 	mutex_unlock(&efx->spi_lock);
 
diff --git a/drivers/net/sfc/falcon.c b/drivers/net/sfc/falcon.c
index 29d4537..950de84 100644
--- a/drivers/net/sfc/falcon.c
+++ b/drivers/net/sfc/falcon.c
@@ -1640,11 +1640,10 @@ static int falcon_spi_wait(struct efx_nic *efx)
 	}
 }
 
-int falcon_spi_cmd(const struct efx_spi_device *spi,
+int falcon_spi_cmd(struct efx_nic *efx, const struct efx_spi_device *spi,
 		   unsigned int command, int address,
 		   const void *in, void *out, size_t len)
 {
-	struct efx_nic *efx = spi->efx;
 	bool addressed = (address >= 0);
 	bool reading = (out != NULL);
 	efx_oword_t reg;
@@ -1713,15 +1712,15 @@ efx_spi_munge_command(const struct efx_spi_device *spi,
 }
 
 /* Wait up to 10 ms for buffered write completion */
-int falcon_spi_wait_write(const struct efx_spi_device *spi)
+int
+falcon_spi_wait_write(struct efx_nic *efx, const struct efx_spi_device *spi)
 {
-	struct efx_nic *efx = spi->efx;
 	unsigned long timeout = jiffies + 1 + DIV_ROUND_UP(HZ, 100);
 	u8 status;
 	int rc;
 
 	for (;;) {
-		rc = falcon_spi_cmd(spi, SPI_RDSR, -1, NULL,
+		rc = falcon_spi_cmd(efx, spi, SPI_RDSR, -1, NULL,
 				    &status, sizeof(status));
 		if (rc)
 			return rc;
@@ -1737,8 +1736,8 @@ int falcon_spi_wait_write(const struct efx_spi_device *spi)
 	}
 }
 
-int falcon_spi_read(const struct efx_spi_device *spi, loff_t start,
-		    size_t len, size_t *retlen, u8 *buffer)
+int falcon_spi_read(struct efx_nic *efx, const struct efx_spi_device *spi,
+		    loff_t start, size_t len, size_t *retlen, u8 *buffer)
 {
 	size_t block_len, pos = 0;
 	unsigned int command;
@@ -1748,7 +1747,7 @@ int falcon_spi_read(const struct efx_spi_device *spi, loff_t start,
 		block_len = min(len - pos, FALCON_SPI_MAX_LEN);
 
 		command = efx_spi_munge_command(spi, SPI_READ, start + pos);
-		rc = falcon_spi_cmd(spi, command, start + pos, NULL,
+		rc = falcon_spi_cmd(efx, spi, command, start + pos, NULL,
 				    buffer + pos, block_len);
 		if (rc)
 			break;
@@ -1767,8 +1766,9 @@ int falcon_spi_read(const struct efx_spi_device *spi, loff_t start,
 	return rc;
 }
 
-int falcon_spi_write(const struct efx_spi_device *spi, loff_t start,
-		     size_t len, size_t *retlen, const u8 *buffer)
+int
+falcon_spi_write(struct efx_nic *efx, const struct efx_spi_device *spi,
+		 loff_t start, size_t len, size_t *retlen, const u8 *buffer)
 {
 	u8 verify_buffer[FALCON_SPI_MAX_LEN];
 	size_t block_len, pos = 0;
@@ -1776,24 +1776,24 @@ int falcon_spi_write(const struct efx_spi_device *spi, loff_t start,
 	int rc = 0;
 
 	while (pos < len) {
-		rc = falcon_spi_cmd(spi, SPI_WREN, -1, NULL, NULL, 0);
+		rc = falcon_spi_cmd(efx, spi, SPI_WREN, -1, NULL, NULL, 0);
 		if (rc)
 			break;
 
 		block_len = min(len - pos,
 				falcon_spi_write_limit(spi, start + pos));
 		command = efx_spi_munge_command(spi, SPI_WRITE, start + pos);
-		rc = falcon_spi_cmd(spi, command, start + pos,
+		rc = falcon_spi_cmd(efx, spi, command, start + pos,
 				    buffer + pos, NULL, block_len);
 		if (rc)
 			break;
 
-		rc = falcon_spi_wait_write(spi);
+		rc = falcon_spi_wait_write(efx, spi);
 		if (rc)
 			break;
 
 		command = efx_spi_munge_command(spi, SPI_READ, start + pos);
-		rc = falcon_spi_cmd(spi, command, start + pos,
+		rc = falcon_spi_cmd(efx, spi, command, start + pos,
 				    NULL, verify_buffer, block_len);
 		if (memcmp(verify_buffer, buffer + pos, block_len)) {
 			rc = -EIO;
@@ -2352,7 +2352,7 @@ falcon_read_nvram(struct efx_nic *efx, struct falcon_nvconfig *nvconfig_out)
 	nvconfig = region + FALCON_NVCONFIG_OFFSET;
 
 	mutex_lock(&efx->spi_lock);
-	rc = falcon_spi_read(spi, 0, FALCON_NVCONFIG_END, NULL, region);
+	rc = falcon_spi_read(efx, spi, 0, FALCON_NVCONFIG_END, NULL, region);
 	mutex_unlock(&efx->spi_lock);
 	if (rc) {
 		EFX_ERR(efx, "Failed to read %s\n",
@@ -2710,8 +2710,6 @@ static int falcon_spi_device_init(struct efx_nic *efx,
 		spi_device->block_size =
 			1 << SPI_DEV_TYPE_FIELD(device_type,
 						SPI_DEV_TYPE_BLOCK_SIZE);
-
-		spi_device->efx = efx;
 	} else {
 		spi_device = NULL;
 	}
diff --git a/drivers/net/sfc/mtd.c b/drivers/net/sfc/mtd.c
index 820c233..3121e24 100644
--- a/drivers/net/sfc/mtd.c
+++ b/drivers/net/sfc/mtd.c
@@ -11,26 +11,58 @@
 #include <linux/module.h>
 #include <linux/mtd/mtd.h>
 #include <linux/delay.h>
+#include <linux/rtnetlink.h>
 
 #define EFX_DRIVER_NAME "sfc_mtd"
 #include "net_driver.h"
 #include "spi.h"
 #include "efx.h"
+#include "falcon.h"
 
 #define EFX_SPI_VERIFY_BUF_LEN 16
 
-struct efx_mtd {
-	const struct efx_spi_device *spi;
+struct efx_mtd_partition {
 	struct mtd_info mtd;
+	size_t offset;
+	const char *type_name;
 	char name[IFNAMSIZ + 20];
 };
 
+struct efx_mtd_ops {
+	int (*read)(struct mtd_info *mtd, loff_t start, size_t len,
+		    size_t *retlen, u8 *buffer);
+	int (*erase)(struct mtd_info *mtd, loff_t start, size_t len);
+	int (*write)(struct mtd_info *mtd, loff_t start, size_t len,
+		     size_t *retlen, const u8 *buffer);
+	int (*sync)(struct mtd_info *mtd);
+};
+
+struct efx_mtd {
+	struct list_head node;
+	struct efx_nic *efx;
+	const struct efx_spi_device *spi;
+	const char *name;
+	const struct efx_mtd_ops *ops;
+	size_t n_parts;
+	struct efx_mtd_partition part[0];
+};
+
+#define efx_for_each_partition(part, efx_mtd)			\
+	for ((part) = &(efx_mtd)->part[0];			\
+	     (part) != &(efx_mtd)->part[(efx_mtd)->n_parts];	\
+	     (part)++)
+
+#define to_efx_mtd_partition(mtd)				\
+	container_of(mtd, struct efx_mtd_partition, mtd)
+
+static int falcon_mtd_probe(struct efx_nic *efx);
+
 /* SPI utilities */
 
 static int efx_spi_slow_wait(struct efx_mtd *efx_mtd, bool uninterruptible)
 {
 	const struct efx_spi_device *spi = efx_mtd->spi;
-	struct efx_nic *efx = spi->efx;
+	struct efx_nic *efx = efx_mtd->efx;
 	u8 status;
 	int rc, i;
 
@@ -39,7 +71,7 @@ static int efx_spi_slow_wait(struct efx_mtd *efx_mtd, bool uninterruptible)
 		__set_current_state(uninterruptible ?
 				    TASK_UNINTERRUPTIBLE : TASK_INTERRUPTIBLE);
 		schedule_timeout(HZ / 10);
-		rc = falcon_spi_cmd(spi, SPI_RDSR, -1, NULL,
+		rc = falcon_spi_cmd(efx, spi, SPI_RDSR, -1, NULL,
 				    &status, sizeof(status));
 		if (rc)
 			return rc;
@@ -52,32 +84,35 @@ static int efx_spi_slow_wait(struct efx_mtd *efx_mtd, bool uninterruptible)
 	return -ETIMEDOUT;
 }
 
-static int efx_spi_unlock(const struct efx_spi_device *spi)
+static int
+efx_spi_unlock(struct efx_nic *efx, const struct efx_spi_device *spi)
 {
 	const u8 unlock_mask = (SPI_STATUS_BP2 | SPI_STATUS_BP1 |
 				SPI_STATUS_BP0);
 	u8 status;
 	int rc;
 
-	rc = falcon_spi_cmd(spi, SPI_RDSR, -1, NULL, &status, sizeof(status));
+	rc = falcon_spi_cmd(efx, spi, SPI_RDSR, -1, NULL,
+			    &status, sizeof(status));
 	if (rc)
 		return rc;
 
 	if (!(status & unlock_mask))
 		return 0; /* already unlocked */
 
-	rc = falcon_spi_cmd(spi, SPI_WREN, -1, NULL, NULL, 0);
+	rc = falcon_spi_cmd(efx, spi, SPI_WREN, -1, NULL, NULL, 0);
 	if (rc)
 		return rc;
-	rc = falcon_spi_cmd(spi, SPI_SST_EWSR, -1, NULL, NULL, 0);
+	rc = falcon_spi_cmd(efx, spi, SPI_SST_EWSR, -1, NULL, NULL, 0);
 	if (rc)
 		return rc;
 
 	status &= ~unlock_mask;
-	rc = falcon_spi_cmd(spi, SPI_WRSR, -1, &status, NULL, sizeof(status));
+	rc = falcon_spi_cmd(efx, spi, SPI_WRSR, -1, &status,
+			    NULL, sizeof(status));
 	if (rc)
 		return rc;
-	rc = falcon_spi_wait_write(spi);
+	rc = falcon_spi_wait_write(efx, spi);
 	if (rc)
 		return rc;
 
@@ -87,6 +122,7 @@ static int efx_spi_unlock(const struct efx_spi_device *spi)
 static int efx_spi_erase(struct efx_mtd *efx_mtd, loff_t start, size_t len)
 {
 	const struct efx_spi_device *spi = efx_mtd->spi;
+	struct efx_nic *efx = efx_mtd->efx;
 	unsigned pos, block_len;
 	u8 empty[EFX_SPI_VERIFY_BUF_LEN];
 	u8 buffer[EFX_SPI_VERIFY_BUF_LEN];
@@ -98,13 +134,14 @@ static int efx_spi_erase(struct efx_mtd *efx_mtd, loff_t start, size_t len)
 	if (spi->erase_command == 0)
 		return -EOPNOTSUPP;
 
-	rc = efx_spi_unlock(spi);
+	rc = efx_spi_unlock(efx, spi);
 	if (rc)
 		return rc;
-	rc = falcon_spi_cmd(spi, SPI_WREN, -1, NULL, NULL, 0);
+	rc = falcon_spi_cmd(efx, spi, SPI_WREN, -1, NULL, NULL, 0);
 	if (rc)
 		return rc;
-	rc = falcon_spi_cmd(spi, spi->erase_command, start, NULL, NULL, 0);
+	rc = falcon_spi_cmd(efx, spi, spi->erase_command, start, NULL,
+			    NULL, 0);
 	if (rc)
 		return rc;
 	rc = efx_spi_slow_wait(efx_mtd, false);
@@ -113,7 +150,8 @@ static int efx_spi_erase(struct efx_mtd *efx_mtd, loff_t start, size_t len)
 	memset(empty, 0xff, sizeof(empty));
 	for (pos = 0; pos < len; pos += block_len) {
 		block_len = min(len - pos, sizeof(buffer));
-		rc = falcon_spi_read(spi, start + pos, block_len, NULL, buffer);
+		rc = falcon_spi_read(efx, spi, start + pos, block_len,
+				     NULL, buffer);
 		if (rc)
 			return rc;
 		if (memcmp(empty, buffer, block_len))
@@ -130,140 +168,228 @@ static int efx_spi_erase(struct efx_mtd *efx_mtd, loff_t start, size_t len)
 
 /* MTD interface */
 
-static int efx_mtd_read(struct mtd_info *mtd, loff_t start, size_t len,
-			size_t *retlen, u8 *buffer)
+static int efx_mtd_erase(struct mtd_info *mtd, struct erase_info *erase)
+{
+	struct efx_mtd *efx_mtd = mtd->priv;
+	int rc;
+
+	rc = efx_mtd->ops->erase(mtd, erase->addr, erase->len);
+	if (rc == 0) {
+		erase->state = MTD_ERASE_DONE;
+	} else {
+		erase->state = MTD_ERASE_FAILED;
+		erase->fail_addr = 0xffffffff;
+	}
+	mtd_erase_callback(erase);
+	return rc;
+}
+
+static void efx_mtd_sync(struct mtd_info *mtd)
+{
+	struct efx_mtd *efx_mtd = mtd->priv;
+	struct efx_nic *efx = efx_mtd->efx;
+	int rc;
+
+	rc = efx_mtd->ops->sync(mtd);
+	if (rc)
+		EFX_ERR(efx, "%s sync failed (%d)\n", efx_mtd->name, rc);
+}
+
+static void efx_mtd_remove_partition(struct efx_mtd_partition *part)
+{
+	int rc;
+
+	for (;;) {
+		rc = del_mtd_device(&part->mtd);
+		if (rc != -EBUSY)
+			break;
+		ssleep(1);
+	}
+	WARN_ON(rc);
+}
+
+static void efx_mtd_remove_device(struct efx_mtd *efx_mtd)
+{
+	struct efx_mtd_partition *part;
+
+	efx_for_each_partition(part, efx_mtd)
+		efx_mtd_remove_partition(part);
+	list_del(&efx_mtd->node);
+	kfree(efx_mtd);
+}
+
+static void efx_mtd_rename_device(struct efx_mtd *efx_mtd)
+{
+	struct efx_mtd_partition *part;
+
+	efx_for_each_partition(part, efx_mtd)
+		snprintf(part->name, sizeof(part->name),
+			 "%s %s", efx_mtd->efx->name,
+			 part->type_name);
+}
+
+static int efx_mtd_probe_device(struct efx_nic *efx, struct efx_mtd *efx_mtd)
+{
+	struct efx_mtd_partition *part;
+
+	efx_mtd->efx = efx;
+
+	efx_mtd_rename_device(efx_mtd);
+
+	efx_for_each_partition(part, efx_mtd) {
+		part->mtd.writesize = 1;
+
+		part->mtd.owner = THIS_MODULE;
+		part->mtd.priv = efx_mtd;
+		part->mtd.name = part->name;
+		part->mtd.erase = efx_mtd_erase;
+		part->mtd.read = efx_mtd->ops->read;
+		part->mtd.write = efx_mtd->ops->write;
+		part->mtd.sync = efx_mtd_sync;
+
+		if (add_mtd_device(&part->mtd))
+			goto fail;
+	}
+
+	list_add(&efx_mtd->node, &efx->mtd_list);
+	return 0;
+
+fail:
+	while (part != &efx_mtd->part[0]) {
+		--part;
+		efx_mtd_remove_partition(part);
+	}
+	/* add_mtd_device() returns 1 if the MTD table is full */
+	return -ENOMEM;
+}
+
+void efx_mtd_remove(struct efx_nic *efx)
 {
+	struct efx_mtd *efx_mtd, *next;
+
+	WARN_ON(efx_dev_registered(efx));
+
+	list_for_each_entry_safe(efx_mtd, next, &efx->mtd_list, node)
+		efx_mtd_remove_device(efx_mtd);
+}
+
+void efx_mtd_rename(struct efx_nic *efx)
+{
+	struct efx_mtd *efx_mtd;
+
+	ASSERT_RTNL();
+
+	list_for_each_entry(efx_mtd, &efx->mtd_list, node)
+		efx_mtd_rename_device(efx_mtd);
+}
+
+int efx_mtd_probe(struct efx_nic *efx)
+{
+	return falcon_mtd_probe(efx);
+}
+
+/* Implementation of MTD operations for Falcon */
+
+static int falcon_mtd_read(struct mtd_info *mtd, loff_t start,
+			   size_t len, size_t *retlen, u8 *buffer)
+{
+	struct efx_mtd_partition *part = to_efx_mtd_partition(mtd);
 	struct efx_mtd *efx_mtd = mtd->priv;
 	const struct efx_spi_device *spi = efx_mtd->spi;
-	struct efx_nic *efx = spi->efx;
+	struct efx_nic *efx = efx_mtd->efx;
 	int rc;
 
 	rc = mutex_lock_interruptible(&efx->spi_lock);
 	if (rc)
 		return rc;
-	rc = falcon_spi_read(spi, FALCON_FLASH_BOOTCODE_START + start,
-			     len, retlen, buffer);
+	rc = falcon_spi_read(efx, spi, part->offset + start, len,
+			     retlen, buffer);
 	mutex_unlock(&efx->spi_lock);
 	return rc;
 }
 
-static int efx_mtd_erase(struct mtd_info *mtd, struct erase_info *erase)
+static int falcon_mtd_erase(struct mtd_info *mtd, loff_t start, size_t len)
 {
+	struct efx_mtd_partition *part = to_efx_mtd_partition(mtd);
 	struct efx_mtd *efx_mtd = mtd->priv;
-	struct efx_nic *efx = efx_mtd->spi->efx;
+	struct efx_nic *efx = efx_mtd->efx;
 	int rc;
 
 	rc = mutex_lock_interruptible(&efx->spi_lock);
 	if (rc)
 		return rc;
-	rc = efx_spi_erase(efx_mtd, FALCON_FLASH_BOOTCODE_START + erase->addr,
-			   erase->len);
+	rc = efx_spi_erase(efx_mtd, part->offset + start, len);
 	mutex_unlock(&efx->spi_lock);
-
-	if (rc == 0) {
-		erase->state = MTD_ERASE_DONE;
-	} else {
-		erase->state = MTD_ERASE_FAILED;
-		erase->fail_addr = 0xffffffff;
-	}
-	mtd_erase_callback(erase);
 	return rc;
 }
 
-static int efx_mtd_write(struct mtd_info *mtd, loff_t start,
-			 size_t len, size_t *retlen, const u8 *buffer)
+static int falcon_mtd_write(struct mtd_info *mtd, loff_t start,
+			    size_t len, size_t *retlen, const u8 *buffer)
 {
+	struct efx_mtd_partition *part = to_efx_mtd_partition(mtd);
 	struct efx_mtd *efx_mtd = mtd->priv;
 	const struct efx_spi_device *spi = efx_mtd->spi;
-	struct efx_nic *efx = spi->efx;
+	struct efx_nic *efx = efx_mtd->efx;
 	int rc;
 
 	rc = mutex_lock_interruptible(&efx->spi_lock);
 	if (rc)
 		return rc;
-	rc = falcon_spi_write(spi, FALCON_FLASH_BOOTCODE_START + start,
-			      len, retlen, buffer);
+	rc = falcon_spi_write(efx, spi, part->offset + start, len,
+			      retlen, buffer);
 	mutex_unlock(&efx->spi_lock);
 	return rc;
 }
 
-static void efx_mtd_sync(struct mtd_info *mtd)
+static int falcon_mtd_sync(struct mtd_info *mtd)
 {
 	struct efx_mtd *efx_mtd = mtd->priv;
-	struct efx_nic *efx = efx_mtd->spi->efx;
+	struct efx_nic *efx = efx_mtd->efx;
 	int rc;
 
 	mutex_lock(&efx->spi_lock);
 	rc = efx_spi_slow_wait(efx_mtd, true);
 	mutex_unlock(&efx->spi_lock);
-
-	if (rc)
-		EFX_ERR(efx, "%s sync failed (%d)\n", efx_mtd->name, rc);
-	return;
-}
-
-void efx_mtd_remove(struct efx_nic *efx)
-{
-	if (efx->spi_flash && efx->spi_flash->mtd) {
-		struct efx_mtd *efx_mtd = efx->spi_flash->mtd;
-		int rc;
-
-		for (;;) {
-			rc = del_mtd_device(&efx_mtd->mtd);
-			if (rc != -EBUSY)
-				break;
-			ssleep(1);
-		}
-		WARN_ON(rc);
-		kfree(efx_mtd);
-	}
+	return rc;
 }
 
-void efx_mtd_rename(struct efx_nic *efx)
-{
-	if (efx->spi_flash && efx->spi_flash->mtd) {
-		struct efx_mtd *efx_mtd = efx->spi_flash->mtd;
-		snprintf(efx_mtd->name, sizeof(efx_mtd->name),
-			 "%s sfc_flash_bootrom", efx->name);
-	}
-}
+static struct efx_mtd_ops falcon_mtd_ops = {
+	.read	= falcon_mtd_read,
+	.erase	= falcon_mtd_erase,
+	.write	= falcon_mtd_write,
+	.sync	= falcon_mtd_sync,
+};
 
-int efx_mtd_probe(struct efx_nic *efx)
+static int falcon_mtd_probe(struct efx_nic *efx)
 {
 	struct efx_spi_device *spi = efx->spi_flash;
 	struct efx_mtd *efx_mtd;
+	int rc;
+
+	ASSERT_RTNL();
 
 	if (!spi || spi->size <= FALCON_FLASH_BOOTCODE_START)
 		return -ENODEV;
 
-	efx_mtd = kzalloc(sizeof(*efx_mtd), GFP_KERNEL);
+	efx_mtd = kzalloc(sizeof(*efx_mtd) + sizeof(efx_mtd->part[0]),
+			  GFP_KERNEL);
 	if (!efx_mtd)
 		return -ENOMEM;
 
 	efx_mtd->spi = spi;
-	spi->mtd = efx_mtd;
-
-	efx_mtd->mtd.type = MTD_NORFLASH;
-	efx_mtd->mtd.flags = MTD_CAP_NORFLASH;
-	efx_mtd->mtd.size = spi->size - FALCON_FLASH_BOOTCODE_START;
-	efx_mtd->mtd.erasesize = spi->erase_size;
-	efx_mtd->mtd.writesize = 1;
-	efx_mtd_rename(efx);
-
-	efx_mtd->mtd.owner = THIS_MODULE;
-	efx_mtd->mtd.priv = efx_mtd;
-	efx_mtd->mtd.name = efx_mtd->name;
-	efx_mtd->mtd.erase = efx_mtd_erase;
-	efx_mtd->mtd.read = efx_mtd_read;
-	efx_mtd->mtd.write = efx_mtd_write;
-	efx_mtd->mtd.sync = efx_mtd_sync;
-
-	if (add_mtd_device(&efx_mtd->mtd)) {
+	efx_mtd->name = "flash";
+	efx_mtd->ops = &falcon_mtd_ops;
+
+	efx_mtd->n_parts = 1;
+	efx_mtd->part[0].mtd.type = MTD_NORFLASH;
+	efx_mtd->part[0].mtd.flags = MTD_CAP_NORFLASH;
+	efx_mtd->part[0].mtd.size = spi->size - FALCON_FLASH_BOOTCODE_START;
+	efx_mtd->part[0].mtd.erasesize = spi->erase_size;
+	efx_mtd->part[0].offset = FALCON_FLASH_BOOTCODE_START;
+	efx_mtd->part[0].type_name = "sfc_flash_bootrom";
+
+	rc = efx_mtd_probe_device(efx, efx_mtd);
+	if (rc)
 		kfree(efx_mtd);
-		spi->mtd = NULL;
-		/* add_mtd_device() returns 1 if the MTD table is full */
-		return -ENOMEM;
-	}
-
-	return 0;
+	return rc;
 }
diff --git a/drivers/net/sfc/net_driver.h b/drivers/net/sfc/net_driver.h
index 452f835..e1534ba 100644
--- a/drivers/net/sfc/net_driver.h
+++ b/drivers/net/sfc/net_driver.h
@@ -674,10 +674,11 @@ union efx_multicast_hash {
  *	interrupt is handled.  It is used by falcon_test_interrupt()
  *	to verify that an interrupt has occurred.
  * @spi_flash: SPI flash device
- *	This field will be %NULL if no flash device is present.
+ *	This field will be %NULL if no flash device is present (or for Siena).
  * @spi_eeprom: SPI EEPROM device
- *	This field will be %NULL if no EEPROM device is present.
+ *	This field will be %NULL if no EEPROM device is present (or for Siena).
  * @spi_lock: SPI bus lock
+ * @mtd_list: List of MTDs attached to the NIC
  * @n_rx_nodesc_drop_cnt: RX no descriptor drop count
  * @nic_data: Hardware dependant state
  * @mac_lock: MAC access lock. Protects @port_enabled, @phy_mode,
@@ -759,6 +760,9 @@ struct efx_nic {
 	struct efx_spi_device *spi_flash;
 	struct efx_spi_device *spi_eeprom;
 	struct mutex spi_lock;
+#ifdef CONFIG_SFC_MTD
+	struct list_head mtd_list;
+#endif
 
 	unsigned n_rx_nodesc_drop_cnt;
 
diff --git a/drivers/net/sfc/spi.h b/drivers/net/sfc/spi.h
index 1b1ceb4..8bf4fce 100644
--- a/drivers/net/sfc/spi.h
+++ b/drivers/net/sfc/spi.h
@@ -36,8 +36,6 @@
 
 /**
  * struct efx_spi_device - an Efx SPI (Serial Peripheral Interface) device
- * @efx:		The Efx controller that owns this device
- * @mtd:		MTD state
  * @device_id:		Controller's id for the device
  * @size:		Size (in bytes)
  * @addr_len:		Number of address bytes in read/write commands
@@ -54,10 +52,6 @@
  *	Write commands are limited to blocks with this size and alignment.
  */
 struct efx_spi_device {
-	struct efx_nic *efx;
-#ifdef CONFIG_SFC_MTD
-	void *mtd;
-#endif
 	int device_id;
 	unsigned int size;
 	unsigned int addr_len;
@@ -67,12 +61,16 @@ struct efx_spi_device {
 	unsigned int block_size;
 };
 
-int falcon_spi_cmd(const struct efx_spi_device *spi, unsigned int command,
+int falcon_spi_cmd(struct efx_nic *efx,
+		   const struct efx_spi_device *spi, unsigned int command,
 		   int address, const void* in, void *out, size_t len);
-int falcon_spi_wait_write(const struct efx_spi_device *spi);
-int falcon_spi_read(const struct efx_spi_device *spi, loff_t start,
+int falcon_spi_wait_write(struct efx_nic *efx,
+			  const struct efx_spi_device *spi);
+int falcon_spi_read(struct efx_nic *efx,
+		    const struct efx_spi_device *spi, loff_t start,
 		    size_t len, size_t *retlen, u8 *buffer);
-int falcon_spi_write(const struct efx_spi_device *spi, loff_t start,
+int falcon_spi_write(struct efx_nic *efx,
+		     const struct efx_spi_device *spi, loff_t start,
 		     size_t len, size_t *retlen, const u8 *buffer);
 
 /*
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 05/12] sfc: Allow for additional checksum offload features
From: Ben Hutchings @ 2009-11-30  1:11 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/efx.c        |    2 +-
 drivers/net/sfc/ethtool.c    |   16 +++++++++++++++-
 drivers/net/sfc/falcon.c     |    2 ++
 drivers/net/sfc/net_driver.h |    3 +++
 4 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/drivers/net/sfc/efx.c b/drivers/net/sfc/efx.c
index 4fe6d63..c49d364 100644
--- a/drivers/net/sfc/efx.c
+++ b/drivers/net/sfc/efx.c
@@ -2179,7 +2179,7 @@ static int __devinit efx_pci_probe(struct pci_dev *pci_dev,
 	net_dev = alloc_etherdev(sizeof(*efx));
 	if (!net_dev)
 		return -ENOMEM;
-	net_dev->features |= (NETIF_F_IP_CSUM | NETIF_F_SG |
+	net_dev->features |= (type->offload_features | NETIF_F_SG |
 			      NETIF_F_HIGHDMA | NETIF_F_TSO |
 			      NETIF_F_GRO);
 	/* Mask for features that also apply to VLAN devices */
diff --git a/drivers/net/sfc/ethtool.c b/drivers/net/sfc/ethtool.c
index 08a9db9..0a79ec7 100644
--- a/drivers/net/sfc/ethtool.c
+++ b/drivers/net/sfc/ethtool.c
@@ -468,6 +468,19 @@ static void efx_ethtool_get_stats(struct net_device *net_dev,
 	}
 }
 
+static int efx_ethtool_set_tx_csum(struct net_device *net_dev, u32 enable)
+{
+	struct efx_nic *efx = netdev_priv(net_dev);
+	unsigned long features = efx->type->offload_features & NETIF_F_ALL_CSUM;
+
+	if (enable)
+		net_dev->features |= features;
+	else
+		net_dev->features &= ~features;
+
+	return 0;
+}
+
 static int efx_ethtool_set_rx_csum(struct net_device *net_dev, u32 enable)
 {
 	struct efx_nic *efx = netdev_priv(net_dev);
@@ -813,7 +826,8 @@ const struct ethtool_ops efx_ethtool_ops = {
 	.get_rx_csum		= efx_ethtool_get_rx_csum,
 	.set_rx_csum		= efx_ethtool_set_rx_csum,
 	.get_tx_csum		= ethtool_op_get_tx_csum,
-	.set_tx_csum		= ethtool_op_set_tx_csum,
+	/* Need to enable/disable IPv6 too */
+	.set_tx_csum		= efx_ethtool_set_tx_csum,
 	.get_sg			= ethtool_op_get_sg,
 	.set_sg			= ethtool_op_set_sg,
 	.get_tso		= ethtool_op_get_tso,
diff --git a/drivers/net/sfc/falcon.c b/drivers/net/sfc/falcon.c
index 950de84..f77bbbe 100644
--- a/drivers/net/sfc/falcon.c
+++ b/drivers/net/sfc/falcon.c
@@ -3304,6 +3304,7 @@ struct efx_nic_type falcon_a1_nic_type = {
 	.phys_addr_channels = 4,
 	.tx_dc_base = 0x130000,
 	.rx_dc_base = 0x100000,
+	.offload_features = NETIF_F_IP_CSUM,
 	.reset_world_flags = ETH_RESET_IRQ,
 };
 
@@ -3351,6 +3352,7 @@ struct efx_nic_type falcon_b0_nic_type = {
 				   * channels */
 	.tx_dc_base = 0x130000,
 	.rx_dc_base = 0x100000,
+	.offload_features = NETIF_F_IP_CSUM,
 	.reset_world_flags = ETH_RESET_IRQ,
 };
 
diff --git a/drivers/net/sfc/net_driver.h b/drivers/net/sfc/net_driver.h
index e1534ba..96d3f00 100644
--- a/drivers/net/sfc/net_driver.h
+++ b/drivers/net/sfc/net_driver.h
@@ -864,6 +864,8 @@ static inline const char *efx_dev_name(struct efx_nic *efx)
  *	descriptors
  * @tx_dc_base: Base address in SRAM of TX queue descriptor caches
  * @rx_dc_base: Base address in SRAM of RX queue descriptor caches
+ * @offload_features: net_device feature flags for protocol offload
+ *	features implemented in hardware
  * @reset_world_flags: Flags for additional components covered by
  *	reset method RESET_TYPE_WORLD
  */
@@ -904,6 +906,7 @@ struct efx_nic_type {
 	unsigned int phys_addr_channels;
 	unsigned int tx_dc_base;
 	unsigned int rx_dc_base;
+	unsigned long offload_features;
 	u32 reset_world_flags;
 };
 
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* Re: linux-next: manual merge of the net tree with the net-current tree
From: David Miller @ 2009-11-30  1:11 UTC (permalink / raw)
  To: sfr; +Cc: netdev, linux-next, linux-kernel, bruce.w.allan,
	jeffrey.t.kirsher
In-Reply-To: <20091124141122.63e81c55.sfr@canb.auug.org.au>

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Tue, 24 Nov 2009 14:11:22 +1100

> Today's linux-next merge of the net tree got a conflict in
> drivers/net/e1000e/ich8lan.c between commit
> 818f33313caab9be2a10458500dbed4a88c1b334 ("e1000e: do not initiate
> autonegotiation during OEM configuration") from the net-current tree and
> commit 94d8186a693284344ee5cb9734086c7a2370241a ("e1000e: cleanup ops
> function pointers") from the net tree.
> 
> Just context changes. I fixed it up (see below) and can carry the fix for
> a while.

I took care of this merge when I pulled net-2.6 into net-next-2.6
last night.

Just FYI...

^ permalink raw reply

* Re: linux-next: manual merge of the net tree with the net-current tree
From: David Miller @ 2009-11-30  1:12 UTC (permalink / raw)
  To: sfr; +Cc: netdev, linux-next, linux-kernel, bruce.w.allan,
	jeffrey.t.kirsher
In-Reply-To: <20091124141131.49b37eec.sfr@canb.auug.org.au>

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Tue, 24 Nov 2009 14:11:31 +1100

> Hi all,
> 
> Today's linux-next merge of the net tree got a conflict in
> drivers/net/e1000e/phy.c between commit
> 189983d469c6d98e64ddfb9f9ce76725cb082ee5 ("e1000e: remove unnecessary
> 82577 workaround causing link issues") from the net-current tree and
> commit 94d8186a693284344ee5cb9734086c7a2370241a ("e1000e: cleanup ops
> function pointers") from the net tree.
> 
> I fixed it up (see below) and can carry the fix for a while.

This conflict also got dealt with last night when I pulled
net-2.6 into net-next-2.6

^ permalink raw reply

* [PATCH 06/12] sfc: Rename falcon.h to nic.h
From: Ben Hutchings @ 2009-11-30  1:12 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

nic.h is no longer specific to Falcon.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/efx.c               |    2 +-
 drivers/net/sfc/ethtool.c           |    2 +-
 drivers/net/sfc/falcon.c            |    2 +-
 drivers/net/sfc/falcon_boards.c     |    2 +-
 drivers/net/sfc/falcon_gmac.c       |    2 +-
 drivers/net/sfc/falcon_xmac.c       |    2 +-
 drivers/net/sfc/mdio_10g.c          |    2 +-
 drivers/net/sfc/mtd.c               |    2 +-
 drivers/net/sfc/{falcon.h => nic.h} |    6 +++---
 drivers/net/sfc/qt202x_phy.c        |    2 +-
 drivers/net/sfc/rx.c                |    2 +-
 drivers/net/sfc/selftest.c          |    2 +-
 drivers/net/sfc/tenxpress.c         |    2 +-
 drivers/net/sfc/tx.c                |    2 +-
 14 files changed, 16 insertions(+), 16 deletions(-)
 rename drivers/net/sfc/{falcon.h => nic.h} (99%)

diff --git a/drivers/net/sfc/efx.c b/drivers/net/sfc/efx.c
index c49d364..97a6ebd 100644
--- a/drivers/net/sfc/efx.c
+++ b/drivers/net/sfc/efx.c
@@ -23,7 +23,7 @@
 #include "net_driver.h"
 #include "efx.h"
 #include "mdio_10g.h"
-#include "falcon.h"
+#include "nic.h"
 
 /**************************************************************************
  *
diff --git a/drivers/net/sfc/ethtool.c b/drivers/net/sfc/ethtool.c
index 0a79ec7..012ee31 100644
--- a/drivers/net/sfc/ethtool.c
+++ b/drivers/net/sfc/ethtool.c
@@ -15,7 +15,7 @@
 #include "workarounds.h"
 #include "selftest.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "spi.h"
 #include "mdio_10g.h"
 
diff --git a/drivers/net/sfc/falcon.c b/drivers/net/sfc/falcon.c
index f77bbbe..64b47da 100644
--- a/drivers/net/sfc/falcon.c
+++ b/drivers/net/sfc/falcon.c
@@ -20,7 +20,7 @@
 #include "efx.h"
 #include "mac.h"
 #include "spi.h"
-#include "falcon.h"
+#include "nic.h"
 #include "regs.h"
 #include "io.h"
 #include "mdio_10g.h"
diff --git a/drivers/net/sfc/falcon_boards.c b/drivers/net/sfc/falcon_boards.c
index b92decc..fa4d4c7 100644
--- a/drivers/net/sfc/falcon_boards.c
+++ b/drivers/net/sfc/falcon_boards.c
@@ -12,7 +12,7 @@
 #include "net_driver.h"
 #include "phy.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "regs.h"
 #include "io.h"
 #include "workarounds.h"
diff --git a/drivers/net/sfc/falcon_gmac.c b/drivers/net/sfc/falcon_gmac.c
index 19dd3ac..aa9b689 100644
--- a/drivers/net/sfc/falcon_gmac.c
+++ b/drivers/net/sfc/falcon_gmac.c
@@ -11,7 +11,7 @@
 #include <linux/delay.h>
 #include "net_driver.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "mac.h"
 #include "regs.h"
 #include "io.h"
diff --git a/drivers/net/sfc/falcon_xmac.c b/drivers/net/sfc/falcon_xmac.c
index 643622d..cd63f24 100644
--- a/drivers/net/sfc/falcon_xmac.c
+++ b/drivers/net/sfc/falcon_xmac.c
@@ -11,7 +11,7 @@
 #include <linux/delay.h>
 #include "net_driver.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "regs.h"
 #include "io.h"
 #include "mac.h"
diff --git a/drivers/net/sfc/mdio_10g.c b/drivers/net/sfc/mdio_10g.c
index 20e6274..19496da 100644
--- a/drivers/net/sfc/mdio_10g.c
+++ b/drivers/net/sfc/mdio_10g.c
@@ -15,7 +15,7 @@
 #include "net_driver.h"
 #include "mdio_10g.h"
 #include "workarounds.h"
-#include "falcon.h"
+#include "nic.h"
 
 unsigned efx_mdio_id_oui(u32 id)
 {
diff --git a/drivers/net/sfc/mtd.c b/drivers/net/sfc/mtd.c
index 3121e24..65a22f1 100644
--- a/drivers/net/sfc/mtd.c
+++ b/drivers/net/sfc/mtd.c
@@ -17,7 +17,7 @@
 #include "net_driver.h"
 #include "spi.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 
 #define EFX_SPI_VERIFY_BUF_LEN 16
 
diff --git a/drivers/net/sfc/falcon.h b/drivers/net/sfc/nic.h
similarity index 99%
rename from drivers/net/sfc/falcon.h
rename to drivers/net/sfc/nic.h
index 3085ecf..e7eb304 100644
--- a/drivers/net/sfc/falcon.h
+++ b/drivers/net/sfc/nic.h
@@ -8,8 +8,8 @@
  * by the Free Software Foundation, incorporated herein by reference.
  */
 
-#ifndef EFX_FALCON_H
-#define EFX_FALCON_H
+#ifndef EFX_NIC_H
+#define EFX_NIC_H
 
 #include <linux/i2c-algo-bit.h>
 #include "net_driver.h"
@@ -235,4 +235,4 @@ extern void efx_nic_generate_event(struct efx_channel *channel,
 
 extern void falcon_poll_xmac(struct efx_nic *efx);
 
-#endif /* EFX_FALCON_H */
+#endif /* EFX_NIC_H */
diff --git a/drivers/net/sfc/qt202x_phy.c b/drivers/net/sfc/qt202x_phy.c
index 22b0e89..957e534 100644
--- a/drivers/net/sfc/qt202x_phy.c
+++ b/drivers/net/sfc/qt202x_phy.c
@@ -15,7 +15,7 @@
 #include "efx.h"
 #include "mdio_10g.h"
 #include "phy.h"
-#include "falcon.h"
+#include "nic.h"
 
 #define QT202X_REQUIRED_DEVS (MDIO_DEVS_PCS |		\
 			      MDIO_DEVS_PMAPMD |	\
diff --git a/drivers/net/sfc/rx.c b/drivers/net/sfc/rx.c
index 8fffd37..03eace3 100644
--- a/drivers/net/sfc/rx.c
+++ b/drivers/net/sfc/rx.c
@@ -17,7 +17,7 @@
 #include <net/checksum.h>
 #include "net_driver.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "selftest.h"
 #include "workarounds.h"
 
diff --git a/drivers/net/sfc/selftest.c b/drivers/net/sfc/selftest.c
index 16258d8..1635f57 100644
--- a/drivers/net/sfc/selftest.c
+++ b/drivers/net/sfc/selftest.c
@@ -21,7 +21,7 @@
 #include <asm/io.h>
 #include "net_driver.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "selftest.h"
 #include "workarounds.h"
 #include "spi.h"
diff --git a/drivers/net/sfc/tenxpress.c b/drivers/net/sfc/tenxpress.c
index 8de97a9..ff97133 100644
--- a/drivers/net/sfc/tenxpress.c
+++ b/drivers/net/sfc/tenxpress.c
@@ -12,7 +12,7 @@
 #include <linux/seq_file.h>
 #include "efx.h"
 #include "mdio_10g.h"
-#include "falcon.h"
+#include "nic.h"
 #include "phy.h"
 #include "regs.h"
 #include "workarounds.h"
diff --git a/drivers/net/sfc/tx.c b/drivers/net/sfc/tx.c
index 2531d02..389ede4 100644
--- a/drivers/net/sfc/tx.c
+++ b/drivers/net/sfc/tx.c
@@ -16,7 +16,7 @@
 #include <linux/highmem.h>
 #include "net_driver.h"
 #include "efx.h"
-#include "falcon.h"
+#include "nic.h"
 #include "workarounds.h"
 
 /*
-- 
1.6.5.3



-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 07/12] sfc: Move shared NIC code from falcon.c to new source file nic.c
From: Ben Hutchings @ 2009-11-30  1:14 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
Looks huge, but it's pure code movement.

Ben.

 drivers/net/sfc/Makefile |    2 +-
 drivers/net/sfc/falcon.c | 1536 ---------------------------------------------
 drivers/net/sfc/nic.c    | 1548 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 1549 insertions(+), 1537 deletions(-)
 create mode 100644 drivers/net/sfc/nic.c

diff --git a/drivers/net/sfc/Makefile b/drivers/net/sfc/Makefile
index 7b52fe1..223106b 100644
--- a/drivers/net/sfc/Makefile
+++ b/drivers/net/sfc/Makefile
@@ -1,4 +1,4 @@
-sfc-y			+= efx.o falcon.o tx.o rx.o falcon_gmac.o \
+sfc-y			+= efx.o nic.o falcon.o tx.o rx.o falcon_gmac.o \
 			   falcon_xmac.o selftest.o ethtool.o qt202x_phy.o \
 			   mdio_10g.o tenxpress.o falcon_boards.o
 sfc-$(CONFIG_SFC_MTD)	+= mtd.o
diff --git a/drivers/net/sfc/falcon.c b/drivers/net/sfc/falcon.c
index 64b47da..48d28d8 100644
--- a/drivers/net/sfc/falcon.c
+++ b/drivers/net/sfc/falcon.c
@@ -29,26 +29,6 @@
 
 /* Hardware control for SFC4000 (aka Falcon). */
 
-/**************************************************************************
- *
- * Configurable values
- *
- **************************************************************************
- */
-
-/* This is set to 16 for a good reason.  In summary, if larger than
- * 16, the descriptor cache holds more than a default socket
- * buffer's worth of packets (for UDP we can only have at most one
- * socket buffer's worth outstanding).  This combined with the fact
- * that we only get 1 TX event per descriptor cache means the NIC
- * goes idle.
- */
-#define TX_DC_ENTRIES 16
-#define TX_DC_ENTRIES_ORDER 1
-
-#define RX_DC_ENTRIES 64
-#define RX_DC_ENTRIES_ORDER 3
-
 static const unsigned int
 /* "Large" EEPROM device: Atmel AT25640 or similar
  * 8 KB, 16-bit address, 32 B write block */
@@ -63,87 +43,6 @@ default_flash_type = ((17 << SPI_DEV_TYPE_SIZE_LBN)
 		      | (15 << SPI_DEV_TYPE_ERASE_SIZE_LBN)
 		      | (8 << SPI_DEV_TYPE_BLOCK_SIZE_LBN));
 
-/* RX FIFO XOFF watermark
- *
- * When the amount of the RX FIFO increases used increases past this
- * watermark send XOFF. Only used if RX flow control is enabled (ethtool -A)
- * This also has an effect on RX/TX arbitration
- */
-int efx_nic_rx_xoff_thresh = -1;
-module_param_named(rx_xoff_thresh_bytes, efx_nic_rx_xoff_thresh, int, 0644);
-MODULE_PARM_DESC(rx_xoff_thresh_bytes, "RX fifo XOFF threshold");
-
-/* RX FIFO XON watermark
- *
- * When the amount of the RX FIFO used decreases below this
- * watermark send XON. Only used if TX flow control is enabled (ethtool -A)
- * This also has an effect on RX/TX arbitration
- */
-int efx_nic_rx_xon_thresh = -1;
-module_param_named(rx_xon_thresh_bytes, efx_nic_rx_xon_thresh, int, 0644);
-MODULE_PARM_DESC(rx_xon_thresh_bytes, "RX fifo XON threshold");
-
-/* If EFX_MAX_INT_ERRORS internal errors occur within
- * EFX_INT_ERROR_EXPIRE seconds, we consider the NIC broken and
- * disable it.
- */
-#define EFX_INT_ERROR_EXPIRE 3600
-#define EFX_MAX_INT_ERRORS 5
-
-/* We poll for events every FLUSH_INTERVAL ms, and check FLUSH_POLL_COUNT times
- */
-#define EFX_FLUSH_INTERVAL 10
-#define EFX_FLUSH_POLL_COUNT 100
-
-/**************************************************************************
- *
- * Falcon constants
- *
- **************************************************************************
- */
-
-/* Size and alignment of special buffers (4KB) */
-#define EFX_BUF_SIZE 4096
-
-/* Depth of RX flush request fifo */
-#define EFX_RX_FLUSH_COUNT 4
-
-/**************************************************************************
- *
- * Solarstorm hardware access
- *
- **************************************************************************/
-
-static inline void efx_write_buf_tbl(struct efx_nic *efx, efx_qword_t *value,
-				     unsigned int index)
-{
-	efx_sram_writeq(efx, efx->membase + efx->type->buf_tbl_base,
-			value, index);
-}
-
-/* Read the current event from the event queue */
-static inline efx_qword_t *efx_event(struct efx_channel *channel,
-				     unsigned int index)
-{
-	return (((efx_qword_t *) (channel->eventq.addr)) + index);
-}
-
-/* See if an event is present
- *
- * We check both the high and low dword of the event for all ones.  We
- * wrote all ones when we cleared the event, and no valid event can
- * have all ones in either its high or low dwords.  This approach is
- * robust against reordering.
- *
- * Note that using a single 64-bit comparison is incorrect; even
- * though the CPU read will be atomic, the DMA write may not be.
- */
-static inline int efx_event_present(efx_qword_t *event)
-{
-	return (!(EFX_DWORD_IS_ALL_ONES(event->dword[0]) |
-		  EFX_DWORD_IS_ALL_ONES(event->dword[1])));
-}
-
 /**************************************************************************
  *
  * I2C bus - this is a bit-bashing interface using GPIO pins
@@ -200,841 +99,6 @@ static struct i2c_algo_bit_data falcon_i2c_bit_operations = {
 	.timeout	= DIV_ROUND_UP(HZ, 20),
 };
 
-/**************************************************************************
- *
- * Special buffer handling
- * Special buffers are used for event queues and the TX and RX
- * descriptor rings.
- *
- *************************************************************************/
-
-/*
- * Initialise a special buffer
- *
- * This will define a buffer (previously allocated via
- * efx_alloc_special_buffer()) in the buffer table, allowing
- * it to be used for event queues, descriptor rings etc.
- */
-static void
-efx_init_special_buffer(struct efx_nic *efx, struct efx_special_buffer *buffer)
-{
-	efx_qword_t buf_desc;
-	int index;
-	dma_addr_t dma_addr;
-	int i;
-
-	EFX_BUG_ON_PARANOID(!buffer->addr);
-
-	/* Write buffer descriptors to NIC */
-	for (i = 0; i < buffer->entries; i++) {
-		index = buffer->index + i;
-		dma_addr = buffer->dma_addr + (i * 4096);
-		EFX_LOG(efx, "mapping special buffer %d at %llx\n",
-			index, (unsigned long long)dma_addr);
-		EFX_POPULATE_QWORD_3(buf_desc,
-				     FRF_AZ_BUF_ADR_REGION, 0,
-				     FRF_AZ_BUF_ADR_FBUF, dma_addr >> 12,
-				     FRF_AZ_BUF_OWNER_ID_FBUF, 0);
-		efx_write_buf_tbl(efx, &buf_desc, index);
-	}
-}
-
-/* Unmaps a buffer and clears the buffer table entries */
-static void
-efx_fini_special_buffer(struct efx_nic *efx, struct efx_special_buffer *buffer)
-{
-	efx_oword_t buf_tbl_upd;
-	unsigned int start = buffer->index;
-	unsigned int end = (buffer->index + buffer->entries - 1);
-
-	if (!buffer->entries)
-		return;
-
-	EFX_LOG(efx, "unmapping special buffers %d-%d\n",
-		buffer->index, buffer->index + buffer->entries - 1);
-
-	EFX_POPULATE_OWORD_4(buf_tbl_upd,
-			     FRF_AZ_BUF_UPD_CMD, 0,
-			     FRF_AZ_BUF_CLR_CMD, 1,
-			     FRF_AZ_BUF_CLR_END_ID, end,
-			     FRF_AZ_BUF_CLR_START_ID, start);
-	efx_writeo(efx, &buf_tbl_upd, FR_AZ_BUF_TBL_UPD);
-}
-
-/*
- * Allocate a new special buffer
- *
- * This allocates memory for a new buffer, clears it and allocates a
- * new buffer ID range.  It does not write into the buffer table.
- *
- * This call will allocate 4KB buffers, since 8KB buffers can't be
- * used for event queues and descriptor rings.
- */
-static int efx_alloc_special_buffer(struct efx_nic *efx,
-				    struct efx_special_buffer *buffer,
-				    unsigned int len)
-{
-	len = ALIGN(len, EFX_BUF_SIZE);
-
-	buffer->addr = pci_alloc_consistent(efx->pci_dev, len,
-					    &buffer->dma_addr);
-	if (!buffer->addr)
-		return -ENOMEM;
-	buffer->len = len;
-	buffer->entries = len / EFX_BUF_SIZE;
-	BUG_ON(buffer->dma_addr & (EFX_BUF_SIZE - 1));
-
-	/* All zeros is a potentially valid event so memset to 0xff */
-	memset(buffer->addr, 0xff, len);
-
-	/* Select new buffer ID */
-	buffer->index = efx->next_buffer_table;
-	efx->next_buffer_table += buffer->entries;
-
-	EFX_LOG(efx, "allocating special buffers %d-%d at %llx+%x "
-		"(virt %p phys %llx)\n", buffer->index,
-		buffer->index + buffer->entries - 1,
-		(u64)buffer->dma_addr, len,
-		buffer->addr, (u64)virt_to_phys(buffer->addr));
-
-	return 0;
-}
-
-static void
-efx_free_special_buffer(struct efx_nic *efx, struct efx_special_buffer *buffer)
-{
-	if (!buffer->addr)
-		return;
-
-	EFX_LOG(efx, "deallocating special buffers %d-%d at %llx+%x "
-		"(virt %p phys %llx)\n", buffer->index,
-		buffer->index + buffer->entries - 1,
-		(u64)buffer->dma_addr, buffer->len,
-		buffer->addr, (u64)virt_to_phys(buffer->addr));
-
-	pci_free_consistent(efx->pci_dev, buffer->len, buffer->addr,
-			    buffer->dma_addr);
-	buffer->addr = NULL;
-	buffer->entries = 0;
-}
-
-/**************************************************************************
- *
- * Generic buffer handling
- * These buffers are used for interrupt status and MAC stats
- *
- **************************************************************************/
-
-int efx_nic_alloc_buffer(struct efx_nic *efx, struct efx_buffer *buffer,
-			 unsigned int len)
-{
-	buffer->addr = pci_alloc_consistent(efx->pci_dev, len,
-					    &buffer->dma_addr);
-	if (!buffer->addr)
-		return -ENOMEM;
-	buffer->len = len;
-	memset(buffer->addr, 0, len);
-	return 0;
-}
-
-void efx_nic_free_buffer(struct efx_nic *efx, struct efx_buffer *buffer)
-{
-	if (buffer->addr) {
-		pci_free_consistent(efx->pci_dev, buffer->len,
-				    buffer->addr, buffer->dma_addr);
-		buffer->addr = NULL;
-	}
-}
-
-/**************************************************************************
- *
- * TX path
- *
- **************************************************************************/
-
-/* Returns a pointer to the specified transmit descriptor in the TX
- * descriptor queue belonging to the specified channel.
- */
-static inline efx_qword_t *
-efx_tx_desc(struct efx_tx_queue *tx_queue, unsigned int index)
-{
-	return (((efx_qword_t *) (tx_queue->txd.addr)) + index);
-}
-
-/* This writes to the TX_DESC_WPTR; write pointer for TX descriptor ring */
-static inline void efx_notify_tx_desc(struct efx_tx_queue *tx_queue)
-{
-	unsigned write_ptr;
-	efx_dword_t reg;
-
-	write_ptr = tx_queue->write_count & EFX_TXQ_MASK;
-	EFX_POPULATE_DWORD_1(reg, FRF_AZ_TX_DESC_WPTR_DWORD, write_ptr);
-	efx_writed_page(tx_queue->efx, &reg,
-			FR_AZ_TX_DESC_UPD_DWORD_P0, tx_queue->queue);
-}
-
-
-/* For each entry inserted into the software descriptor ring, create a
- * descriptor in the hardware TX descriptor ring (in host memory), and
- * write a doorbell.
- */
-void efx_nic_push_buffers(struct efx_tx_queue *tx_queue)
-{
-
-	struct efx_tx_buffer *buffer;
-	efx_qword_t *txd;
-	unsigned write_ptr;
-
-	BUG_ON(tx_queue->write_count == tx_queue->insert_count);
-
-	do {
-		write_ptr = tx_queue->write_count & EFX_TXQ_MASK;
-		buffer = &tx_queue->buffer[write_ptr];
-		txd = efx_tx_desc(tx_queue, write_ptr);
-		++tx_queue->write_count;
-
-		/* Create TX descriptor ring entry */
-		EFX_POPULATE_QWORD_4(*txd,
-				     FSF_AZ_TX_KER_CONT, buffer->continuation,
-				     FSF_AZ_TX_KER_BYTE_COUNT, buffer->len,
-				     FSF_AZ_TX_KER_BUF_REGION, 0,
-				     FSF_AZ_TX_KER_BUF_ADDR, buffer->dma_addr);
-	} while (tx_queue->write_count != tx_queue->insert_count);
-
-	wmb(); /* Ensure descriptors are written before they are fetched */
-	efx_notify_tx_desc(tx_queue);
-}
-
-/* Allocate hardware resources for a TX queue */
-int efx_nic_probe_tx(struct efx_tx_queue *tx_queue)
-{
-	struct efx_nic *efx = tx_queue->efx;
-	BUILD_BUG_ON(EFX_TXQ_SIZE < 512 || EFX_TXQ_SIZE > 4096 ||
-		     EFX_TXQ_SIZE & EFX_TXQ_MASK);
-	return efx_alloc_special_buffer(efx, &tx_queue->txd,
-					EFX_TXQ_SIZE * sizeof(efx_qword_t));
-}
-
-void efx_nic_init_tx(struct efx_tx_queue *tx_queue)
-{
-	efx_oword_t tx_desc_ptr;
-	struct efx_nic *efx = tx_queue->efx;
-
-	tx_queue->flushed = FLUSH_NONE;
-
-	/* Pin TX descriptor ring */
-	efx_init_special_buffer(efx, &tx_queue->txd);
-
-	/* Push TX descriptor ring to card */
-	EFX_POPULATE_OWORD_10(tx_desc_ptr,
-			      FRF_AZ_TX_DESCQ_EN, 1,
-			      FRF_AZ_TX_ISCSI_DDIG_EN, 0,
-			      FRF_AZ_TX_ISCSI_HDIG_EN, 0,
-			      FRF_AZ_TX_DESCQ_BUF_BASE_ID, tx_queue->txd.index,
-			      FRF_AZ_TX_DESCQ_EVQ_ID,
-			      tx_queue->channel->channel,
-			      FRF_AZ_TX_DESCQ_OWNER_ID, 0,
-			      FRF_AZ_TX_DESCQ_LABEL, tx_queue->queue,
-			      FRF_AZ_TX_DESCQ_SIZE,
-			      __ffs(tx_queue->txd.entries),
-			      FRF_AZ_TX_DESCQ_TYPE, 0,
-			      FRF_BZ_TX_NON_IP_DROP_DIS, 1);
-
-	if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
-		int csum = tx_queue->queue == EFX_TX_QUEUE_OFFLOAD_CSUM;
-		EFX_SET_OWORD_FIELD(tx_desc_ptr, FRF_BZ_TX_IP_CHKSM_DIS, !csum);
-		EFX_SET_OWORD_FIELD(tx_desc_ptr, FRF_BZ_TX_TCP_CHKSM_DIS,
-				    !csum);
-	}
-
-	efx_writeo_table(efx, &tx_desc_ptr, efx->type->txd_ptr_tbl_base,
-			 tx_queue->queue);
-
-	if (efx_nic_rev(efx) < EFX_REV_FALCON_B0) {
-		efx_oword_t reg;
-
-		/* Only 128 bits in this register */
-		BUILD_BUG_ON(EFX_TX_QUEUE_COUNT >= 128);
-
-		efx_reado(efx, &reg, FR_AA_TX_CHKSM_CFG);
-		if (tx_queue->queue == EFX_TX_QUEUE_OFFLOAD_CSUM)
-			clear_bit_le(tx_queue->queue, (void *)&reg);
-		else
-			set_bit_le(tx_queue->queue, (void *)&reg);
-		efx_writeo(efx, &reg, FR_AA_TX_CHKSM_CFG);
-	}
-}
-
-static void efx_flush_tx_queue(struct efx_tx_queue *tx_queue)
-{
-	struct efx_nic *efx = tx_queue->efx;
-	efx_oword_t tx_flush_descq;
-
-	tx_queue->flushed = FLUSH_PENDING;
-
-	/* Post a flush command */
-	EFX_POPULATE_OWORD_2(tx_flush_descq,
-			     FRF_AZ_TX_FLUSH_DESCQ_CMD, 1,
-			     FRF_AZ_TX_FLUSH_DESCQ, tx_queue->queue);
-	efx_writeo(efx, &tx_flush_descq, FR_AZ_TX_FLUSH_DESCQ);
-}
-
-void efx_nic_fini_tx(struct efx_tx_queue *tx_queue)
-{
-	struct efx_nic *efx = tx_queue->efx;
-	efx_oword_t tx_desc_ptr;
-
-	/* The queue should have been flushed */
-	WARN_ON(tx_queue->flushed != FLUSH_DONE);
-
-	/* Remove TX descriptor ring from card */
-	EFX_ZERO_OWORD(tx_desc_ptr);
-	efx_writeo_table(efx, &tx_desc_ptr, efx->type->txd_ptr_tbl_base,
-			 tx_queue->queue);
-
-	/* Unpin TX descriptor ring */
-	efx_fini_special_buffer(efx, &tx_queue->txd);
-}
-
-/* Free buffers backing TX queue */
-void efx_nic_remove_tx(struct efx_tx_queue *tx_queue)
-{
-	efx_free_special_buffer(tx_queue->efx, &tx_queue->txd);
-}
-
-/**************************************************************************
- *
- * RX path
- *
- **************************************************************************/
-
-/* Returns a pointer to the specified descriptor in the RX descriptor queue */
-static inline efx_qword_t *
-efx_rx_desc(struct efx_rx_queue *rx_queue, unsigned int index)
-{
-	return (((efx_qword_t *) (rx_queue->rxd.addr)) + index);
-}
-
-/* This creates an entry in the RX descriptor queue */
-static inline void
-efx_build_rx_desc(struct efx_rx_queue *rx_queue, unsigned index)
-{
-	struct efx_rx_buffer *rx_buf;
-	efx_qword_t *rxd;
-
-	rxd = efx_rx_desc(rx_queue, index);
-	rx_buf = efx_rx_buffer(rx_queue, index);
-	EFX_POPULATE_QWORD_3(*rxd,
-			     FSF_AZ_RX_KER_BUF_SIZE,
-			     rx_buf->len -
-			     rx_queue->efx->type->rx_buffer_padding,
-			     FSF_AZ_RX_KER_BUF_REGION, 0,
-			     FSF_AZ_RX_KER_BUF_ADDR, rx_buf->dma_addr);
-}
-
-/* This writes to the RX_DESC_WPTR register for the specified receive
- * descriptor ring.
- */
-void efx_nic_notify_rx_desc(struct efx_rx_queue *rx_queue)
-{
-	efx_dword_t reg;
-	unsigned write_ptr;
-
-	while (rx_queue->notified_count != rx_queue->added_count) {
-		efx_build_rx_desc(rx_queue,
-				  rx_queue->notified_count &
-				  EFX_RXQ_MASK);
-		++rx_queue->notified_count;
-	}
-
-	wmb();
-	write_ptr = rx_queue->added_count & EFX_RXQ_MASK;
-	EFX_POPULATE_DWORD_1(reg, FRF_AZ_RX_DESC_WPTR_DWORD, write_ptr);
-	efx_writed_page(rx_queue->efx, &reg,
-			FR_AZ_RX_DESC_UPD_DWORD_P0, rx_queue->queue);
-}
-
-int efx_nic_probe_rx(struct efx_rx_queue *rx_queue)
-{
-	struct efx_nic *efx = rx_queue->efx;
-	BUILD_BUG_ON(EFX_RXQ_SIZE < 512 || EFX_RXQ_SIZE > 4096 ||
-		     EFX_RXQ_SIZE & EFX_RXQ_MASK);
-	return efx_alloc_special_buffer(efx, &rx_queue->rxd,
-					EFX_RXQ_SIZE * sizeof(efx_qword_t));
-}
-
-void efx_nic_init_rx(struct efx_rx_queue *rx_queue)
-{
-	efx_oword_t rx_desc_ptr;
-	struct efx_nic *efx = rx_queue->efx;
-	bool is_b0 = efx_nic_rev(efx) >= EFX_REV_FALCON_B0;
-	bool iscsi_digest_en = is_b0;
-
-	EFX_LOG(efx, "RX queue %d ring in special buffers %d-%d\n",
-		rx_queue->queue, rx_queue->rxd.index,
-		rx_queue->rxd.index + rx_queue->rxd.entries - 1);
-
-	rx_queue->flushed = FLUSH_NONE;
-
-	/* Pin RX descriptor ring */
-	efx_init_special_buffer(efx, &rx_queue->rxd);
-
-	/* Push RX descriptor ring to card */
-	EFX_POPULATE_OWORD_10(rx_desc_ptr,
-			      FRF_AZ_RX_ISCSI_DDIG_EN, iscsi_digest_en,
-			      FRF_AZ_RX_ISCSI_HDIG_EN, iscsi_digest_en,
-			      FRF_AZ_RX_DESCQ_BUF_BASE_ID, rx_queue->rxd.index,
-			      FRF_AZ_RX_DESCQ_EVQ_ID,
-			      rx_queue->channel->channel,
-			      FRF_AZ_RX_DESCQ_OWNER_ID, 0,
-			      FRF_AZ_RX_DESCQ_LABEL, rx_queue->queue,
-			      FRF_AZ_RX_DESCQ_SIZE,
-			      __ffs(rx_queue->rxd.entries),
-			      FRF_AZ_RX_DESCQ_TYPE, 0 /* kernel queue */ ,
-			      /* For >=B0 this is scatter so disable */
-			      FRF_AZ_RX_DESCQ_JUMBO, !is_b0,
-			      FRF_AZ_RX_DESCQ_EN, 1);
-	efx_writeo_table(efx, &rx_desc_ptr, efx->type->rxd_ptr_tbl_base,
-			 rx_queue->queue);
-}
-
-static void efx_flush_rx_queue(struct efx_rx_queue *rx_queue)
-{
-	struct efx_nic *efx = rx_queue->efx;
-	efx_oword_t rx_flush_descq;
-
-	rx_queue->flushed = FLUSH_PENDING;
-
-	/* Post a flush command */
-	EFX_POPULATE_OWORD_2(rx_flush_descq,
-			     FRF_AZ_RX_FLUSH_DESCQ_CMD, 1,
-			     FRF_AZ_RX_FLUSH_DESCQ, rx_queue->queue);
-	efx_writeo(efx, &rx_flush_descq, FR_AZ_RX_FLUSH_DESCQ);
-}
-
-void efx_nic_fini_rx(struct efx_rx_queue *rx_queue)
-{
-	efx_oword_t rx_desc_ptr;
-	struct efx_nic *efx = rx_queue->efx;
-
-	/* The queue should already have been flushed */
-	WARN_ON(rx_queue->flushed != FLUSH_DONE);
-
-	/* Remove RX descriptor ring from card */
-	EFX_ZERO_OWORD(rx_desc_ptr);
-	efx_writeo_table(efx, &rx_desc_ptr, efx->type->rxd_ptr_tbl_base,
-			 rx_queue->queue);
-
-	/* Unpin RX descriptor ring */
-	efx_fini_special_buffer(efx, &rx_queue->rxd);
-}
-
-/* Free buffers backing RX queue */
-void efx_nic_remove_rx(struct efx_rx_queue *rx_queue)
-{
-	efx_free_special_buffer(rx_queue->efx, &rx_queue->rxd);
-}
-
-/**************************************************************************
- *
- * Event queue processing
- * Event queues are processed by per-channel tasklets.
- *
- **************************************************************************/
-
-/* Update a channel's event queue's read pointer (RPTR) register
- *
- * This writes the EVQ_RPTR_REG register for the specified channel's
- * event queue.
- *
- * Note that EVQ_RPTR_REG contains the index of the "last read" event,
- * whereas channel->eventq_read_ptr contains the index of the "next to
- * read" event.
- */
-void efx_nic_eventq_read_ack(struct efx_channel *channel)
-{
-	efx_dword_t reg;
-	struct efx_nic *efx = channel->efx;
-
-	EFX_POPULATE_DWORD_1(reg, FRF_AZ_EVQ_RPTR, channel->eventq_read_ptr);
-	efx_writed_table(efx, &reg, efx->type->evq_rptr_tbl_base,
-			    channel->channel);
-}
-
-/* Use HW to insert a SW defined event */
-void efx_generate_event(struct efx_channel *channel, efx_qword_t *event)
-{
-	efx_oword_t drv_ev_reg;
-
-	BUILD_BUG_ON(FRF_AZ_DRV_EV_DATA_LBN != 0 ||
-		     FRF_AZ_DRV_EV_DATA_WIDTH != 64);
-	drv_ev_reg.u32[0] = event->u32[0];
-	drv_ev_reg.u32[1] = event->u32[1];
-	drv_ev_reg.u32[2] = 0;
-	drv_ev_reg.u32[3] = 0;
-	EFX_SET_OWORD_FIELD(drv_ev_reg, FRF_AZ_DRV_EV_QID, channel->channel);
-	efx_writeo(channel->efx, &drv_ev_reg, FR_AZ_DRV_EV);
-}
-
-/* Handle a transmit completion event
- *
- * The NIC batches TX completion events; the message we receive is of
- * the form "complete all TX events up to this index".
- */
-static void
-efx_handle_tx_event(struct efx_channel *channel, efx_qword_t *event)
-{
-	unsigned int tx_ev_desc_ptr;
-	unsigned int tx_ev_q_label;
-	struct efx_tx_queue *tx_queue;
-	struct efx_nic *efx = channel->efx;
-
-	if (likely(EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_COMP))) {
-		/* Transmit completion */
-		tx_ev_desc_ptr = EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_DESC_PTR);
-		tx_ev_q_label = EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_Q_LABEL);
-		tx_queue = &efx->tx_queue[tx_ev_q_label];
-		channel->irq_mod_score +=
-			(tx_ev_desc_ptr - tx_queue->read_count) &
-			EFX_TXQ_MASK;
-		efx_xmit_done(tx_queue, tx_ev_desc_ptr);
-	} else if (EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_WQ_FF_FULL)) {
-		/* Rewrite the FIFO write pointer */
-		tx_ev_q_label = EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_Q_LABEL);
-		tx_queue = &efx->tx_queue[tx_ev_q_label];
-
-		if (efx_dev_registered(efx))
-			netif_tx_lock(efx->net_dev);
-		efx_notify_tx_desc(tx_queue);
-		if (efx_dev_registered(efx))
-			netif_tx_unlock(efx->net_dev);
-	} else if (EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_PKT_ERR) &&
-		   EFX_WORKAROUND_10727(efx)) {
-		efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
-	} else {
-		EFX_ERR(efx, "channel %d unexpected TX event "
-			EFX_QWORD_FMT"\n", channel->channel,
-			EFX_QWORD_VAL(*event));
-	}
-}
-
-/* Detect errors included in the rx_evt_pkt_ok bit. */
-static void efx_handle_rx_not_ok(struct efx_rx_queue *rx_queue,
-				 const efx_qword_t *event,
-				 bool *rx_ev_pkt_ok,
-				 bool *discard)
-{
-	struct efx_nic *efx = rx_queue->efx;
-	bool rx_ev_buf_owner_id_err, rx_ev_ip_hdr_chksum_err;
-	bool rx_ev_tcp_udp_chksum_err, rx_ev_eth_crc_err;
-	bool rx_ev_frm_trunc, rx_ev_drib_nib, rx_ev_tobe_disc;
-	bool rx_ev_other_err, rx_ev_pause_frm;
-	bool rx_ev_hdr_type, rx_ev_mcast_pkt;
-	unsigned rx_ev_pkt_type;
-
-	rx_ev_hdr_type = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_HDR_TYPE);
-	rx_ev_mcast_pkt = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_MCAST_PKT);
-	rx_ev_tobe_disc = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_TOBE_DISC);
-	rx_ev_pkt_type = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_PKT_TYPE);
-	rx_ev_buf_owner_id_err = EFX_QWORD_FIELD(*event,
-						 FSF_AZ_RX_EV_BUF_OWNER_ID_ERR);
-	rx_ev_ip_hdr_chksum_err = EFX_QWORD_FIELD(*event,
-						  FSF_AZ_RX_EV_IP_HDR_CHKSUM_ERR);
-	rx_ev_tcp_udp_chksum_err = EFX_QWORD_FIELD(*event,
-						   FSF_AZ_RX_EV_TCP_UDP_CHKSUM_ERR);
-	rx_ev_eth_crc_err = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_ETH_CRC_ERR);
-	rx_ev_frm_trunc = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_FRM_TRUNC);
-	rx_ev_drib_nib = ((efx_nic_rev(efx) >= EFX_REV_FALCON_B0) ?
-			  0 : EFX_QWORD_FIELD(*event, FSF_AA_RX_EV_DRIB_NIB));
-	rx_ev_pause_frm = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_PAUSE_FRM_ERR);
-
-	/* Every error apart from tobe_disc and pause_frm */
-	rx_ev_other_err = (rx_ev_drib_nib | rx_ev_tcp_udp_chksum_err |
-			   rx_ev_buf_owner_id_err | rx_ev_eth_crc_err |
-			   rx_ev_frm_trunc | rx_ev_ip_hdr_chksum_err);
-
-	/* Count errors that are not in MAC stats.  Ignore expected
-	 * checksum errors during self-test. */
-	if (rx_ev_frm_trunc)
-		++rx_queue->channel->n_rx_frm_trunc;
-	else if (rx_ev_tobe_disc)
-		++rx_queue->channel->n_rx_tobe_disc;
-	else if (!efx->loopback_selftest) {
-		if (rx_ev_ip_hdr_chksum_err)
-			++rx_queue->channel->n_rx_ip_hdr_chksum_err;
-		else if (rx_ev_tcp_udp_chksum_err)
-			++rx_queue->channel->n_rx_tcp_udp_chksum_err;
-	}
-
-	/* The frame must be discarded if any of these are true. */
-	*discard = (rx_ev_eth_crc_err | rx_ev_frm_trunc | rx_ev_drib_nib |
-		    rx_ev_tobe_disc | rx_ev_pause_frm);
-
-	/* TOBE_DISC is expected on unicast mismatches; don't print out an
-	 * error message.  FRM_TRUNC indicates RXDP dropped the packet due
-	 * to a FIFO overflow.
-	 */
-#ifdef EFX_ENABLE_DEBUG
-	if (rx_ev_other_err) {
-		EFX_INFO_RL(efx, " RX queue %d unexpected RX event "
-			    EFX_QWORD_FMT "%s%s%s%s%s%s%s%s\n",
-			    rx_queue->queue, EFX_QWORD_VAL(*event),
-			    rx_ev_buf_owner_id_err ? " [OWNER_ID_ERR]" : "",
-			    rx_ev_ip_hdr_chksum_err ?
-			    " [IP_HDR_CHKSUM_ERR]" : "",
-			    rx_ev_tcp_udp_chksum_err ?
-			    " [TCP_UDP_CHKSUM_ERR]" : "",
-			    rx_ev_eth_crc_err ? " [ETH_CRC_ERR]" : "",
-			    rx_ev_frm_trunc ? " [FRM_TRUNC]" : "",
-			    rx_ev_drib_nib ? " [DRIB_NIB]" : "",
-			    rx_ev_tobe_disc ? " [TOBE_DISC]" : "",
-			    rx_ev_pause_frm ? " [PAUSE]" : "");
-	}
-#endif
-}
-
-/* Handle receive events that are not in-order. */
-static void
-efx_handle_rx_bad_index(struct efx_rx_queue *rx_queue, unsigned index)
-{
-	struct efx_nic *efx = rx_queue->efx;
-	unsigned expected, dropped;
-
-	expected = rx_queue->removed_count & EFX_RXQ_MASK;
-	dropped = (index - expected) & EFX_RXQ_MASK;
-	EFX_INFO(efx, "dropped %d events (index=%d expected=%d)\n",
-		dropped, index, expected);
-
-	efx_schedule_reset(efx, EFX_WORKAROUND_5676(efx) ?
-			   RESET_TYPE_RX_RECOVERY : RESET_TYPE_DISABLE);
-}
-
-/* Handle a packet received event
- *
- * The NIC gives a "discard" flag if it's a unicast packet with the
- * wrong destination address
- * Also "is multicast" and "matches multicast filter" flags can be used to
- * discard non-matching multicast packets.
- */
-static void
-efx_handle_rx_event(struct efx_channel *channel, const efx_qword_t *event)
-{
-	unsigned int rx_ev_desc_ptr, rx_ev_byte_cnt;
-	unsigned int rx_ev_hdr_type, rx_ev_mcast_pkt;
-	unsigned expected_ptr;
-	bool rx_ev_pkt_ok, discard = false, checksummed;
-	struct efx_rx_queue *rx_queue;
-	struct efx_nic *efx = channel->efx;
-
-	/* Basic packet information */
-	rx_ev_byte_cnt = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_BYTE_CNT);
-	rx_ev_pkt_ok = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_PKT_OK);
-	rx_ev_hdr_type = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_HDR_TYPE);
-	WARN_ON(EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_JUMBO_CONT));
-	WARN_ON(EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_SOP) != 1);
-	WARN_ON(EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_Q_LABEL) !=
-		channel->channel);
-
-	rx_queue = &efx->rx_queue[channel->channel];
-
-	rx_ev_desc_ptr = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_DESC_PTR);
-	expected_ptr = rx_queue->removed_count & EFX_RXQ_MASK;
-	if (unlikely(rx_ev_desc_ptr != expected_ptr))
-		efx_handle_rx_bad_index(rx_queue, rx_ev_desc_ptr);
-
-	if (likely(rx_ev_pkt_ok)) {
-		/* If packet is marked as OK and packet type is TCP/IP or
-		 * UDP/IP, then we can rely on the hardware checksum.
-		 */
-		checksummed =
-			likely(efx->rx_checksum_enabled) &&
-			(rx_ev_hdr_type == FSE_CZ_RX_EV_HDR_TYPE_IPV4V6_TCP ||
-			 rx_ev_hdr_type == FSE_CZ_RX_EV_HDR_TYPE_IPV4V6_UDP);
-	} else {
-		efx_handle_rx_not_ok(rx_queue, event, &rx_ev_pkt_ok, &discard);
-		checksummed = false;
-	}
-
-	/* Detect multicast packets that didn't match the filter */
-	rx_ev_mcast_pkt = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_MCAST_PKT);
-	if (rx_ev_mcast_pkt) {
-		unsigned int rx_ev_mcast_hash_match =
-			EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_MCAST_HASH_MATCH);
-
-		if (unlikely(!rx_ev_mcast_hash_match)) {
-			++channel->n_rx_mcast_mismatch;
-			discard = true;
-		}
-	}
-
-	channel->irq_mod_score += 2;
-
-	/* Handle received packet */
-	efx_rx_packet(rx_queue, rx_ev_desc_ptr, rx_ev_byte_cnt,
-		      checksummed, discard);
-}
-
-/* Global events are basically PHY events */
-static void
-efx_handle_global_event(struct efx_channel *channel, efx_qword_t *event)
-{
-	struct efx_nic *efx = channel->efx;
-	bool handled = false;
-
-	if (EFX_QWORD_FIELD(*event, FSF_AB_GLB_EV_G_PHY0_INTR) ||
-	    EFX_QWORD_FIELD(*event, FSF_AB_GLB_EV_XG_PHY0_INTR) ||
-	    EFX_QWORD_FIELD(*event, FSF_AB_GLB_EV_XFP_PHY0_INTR)) {
-		/* Ignored */
-		handled = true;
-	}
-
-	if ((efx_nic_rev(efx) >= EFX_REV_FALCON_B0) &&
-	    EFX_QWORD_FIELD(*event, FSF_BB_GLB_EV_XG_MGT_INTR)) {
-		efx->xmac_poll_required = true;
-		handled = true;
-	}
-
-	if (efx_nic_rev(efx) <= EFX_REV_FALCON_A1 ?
-	    EFX_QWORD_FIELD(*event, FSF_AA_GLB_EV_RX_RECOVERY) :
-	    EFX_QWORD_FIELD(*event, FSF_BB_GLB_EV_RX_RECOVERY)) {
-		EFX_ERR(efx, "channel %d seen global RX_RESET "
-			"event. Resetting.\n", channel->channel);
-
-		atomic_inc(&efx->rx_reset);
-		efx_schedule_reset(efx, EFX_WORKAROUND_6555(efx) ?
-				   RESET_TYPE_RX_RECOVERY : RESET_TYPE_DISABLE);
-		handled = true;
-	}
-
-	if (!handled)
-		EFX_ERR(efx, "channel %d unknown global event "
-			EFX_QWORD_FMT "\n", channel->channel,
-			EFX_QWORD_VAL(*event));
-}
-
-static void
-efx_handle_driver_event(struct efx_channel *channel, efx_qword_t *event)
-{
-	struct efx_nic *efx = channel->efx;
-	unsigned int ev_sub_code;
-	unsigned int ev_sub_data;
-
-	ev_sub_code = EFX_QWORD_FIELD(*event, FSF_AZ_DRIVER_EV_SUBCODE);
-	ev_sub_data = EFX_QWORD_FIELD(*event, FSF_AZ_DRIVER_EV_SUBDATA);
-
-	switch (ev_sub_code) {
-	case FSE_AZ_TX_DESCQ_FLS_DONE_EV:
-		EFX_TRACE(efx, "channel %d TXQ %d flushed\n",
-			  channel->channel, ev_sub_data);
-		break;
-	case FSE_AZ_RX_DESCQ_FLS_DONE_EV:
-		EFX_TRACE(efx, "channel %d RXQ %d flushed\n",
-			  channel->channel, ev_sub_data);
-		break;
-	case FSE_AZ_EVQ_INIT_DONE_EV:
-		EFX_LOG(efx, "channel %d EVQ %d initialised\n",
-			channel->channel, ev_sub_data);
-		break;
-	case FSE_AZ_SRM_UPD_DONE_EV:
-		EFX_TRACE(efx, "channel %d SRAM update done\n",
-			  channel->channel);
-		break;
-	case FSE_AZ_WAKE_UP_EV:
-		EFX_TRACE(efx, "channel %d RXQ %d wakeup event\n",
-			  channel->channel, ev_sub_data);
-		break;
-	case FSE_AZ_TIMER_EV:
-		EFX_TRACE(efx, "channel %d RX queue %d timer expired\n",
-			  channel->channel, ev_sub_data);
-		break;
-	case FSE_AA_RX_RECOVER_EV:
-		EFX_ERR(efx, "channel %d seen DRIVER RX_RESET event. "
-			"Resetting.\n", channel->channel);
-		atomic_inc(&efx->rx_reset);
-		efx_schedule_reset(efx,
-				   EFX_WORKAROUND_6555(efx) ?
-				   RESET_TYPE_RX_RECOVERY :
-				   RESET_TYPE_DISABLE);
-		break;
-	case FSE_BZ_RX_DSC_ERROR_EV:
-		EFX_ERR(efx, "RX DMA Q %d reports descriptor fetch error."
-			" RX Q %d is disabled.\n", ev_sub_data, ev_sub_data);
-		efx_schedule_reset(efx, RESET_TYPE_RX_DESC_FETCH);
-		break;
-	case FSE_BZ_TX_DSC_ERROR_EV:
-		EFX_ERR(efx, "TX DMA Q %d reports descriptor fetch error."
-			" TX Q %d is disabled.\n", ev_sub_data, ev_sub_data);
-		efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
-		break;
-	default:
-		EFX_TRACE(efx, "channel %d unknown driver event code %d "
-			  "data %04x\n", channel->channel, ev_sub_code,
-			  ev_sub_data);
-		break;
-	}
-}
-
-int efx_nic_process_eventq(struct efx_channel *channel, int rx_quota)
-{
-	unsigned int read_ptr;
-	efx_qword_t event, *p_event;
-	int ev_code;
-	int rx_packets = 0;
-
-	read_ptr = channel->eventq_read_ptr;
-
-	do {
-		p_event = efx_event(channel, read_ptr);
-		event = *p_event;
-
-		if (!efx_event_present(&event))
-			/* End of events */
-			break;
-
-		EFX_TRACE(channel->efx, "channel %d event is "EFX_QWORD_FMT"\n",
-			  channel->channel, EFX_QWORD_VAL(event));
-
-		/* Clear this event by marking it all ones */
-		EFX_SET_QWORD(*p_event);
-
-		ev_code = EFX_QWORD_FIELD(event, FSF_AZ_EV_CODE);
-
-		switch (ev_code) {
-		case FSE_AZ_EV_CODE_RX_EV:
-			efx_handle_rx_event(channel, &event);
-			++rx_packets;
-			break;
-		case FSE_AZ_EV_CODE_TX_EV:
-			efx_handle_tx_event(channel, &event);
-			break;
-		case FSE_AZ_EV_CODE_DRV_GEN_EV:
-			channel->eventq_magic = EFX_QWORD_FIELD(
-				event, FSF_AZ_DRV_GEN_EV_MAGIC);
-			EFX_LOG(channel->efx, "channel %d received generated "
-				"event "EFX_QWORD_FMT"\n", channel->channel,
-				EFX_QWORD_VAL(event));
-			break;
-		case FSE_AZ_EV_CODE_GLOBAL_EV:
-			efx_handle_global_event(channel, &event);
-			break;
-		case FSE_AZ_EV_CODE_DRIVER_EV:
-			efx_handle_driver_event(channel, &event);
-			break;
-		default:
-			EFX_ERR(channel->efx, "channel %d unknown event type %d"
-				" (data " EFX_QWORD_FMT ")\n", channel->channel,
-				ev_code, EFX_QWORD_VAL(event));
-		}
-
-		/* Increment read pointer */
-		read_ptr = (read_ptr + 1) & EFX_EVQ_MASK;
-
-	} while (rx_packets < rx_quota);
-
-	channel->eventq_read_ptr = read_ptr;
-	return rx_packets;
-}
-
 static void falcon_push_irq_moderation(struct efx_channel *channel)
 {
 	efx_dword_t timer_cmd;
@@ -1056,135 +120,6 @@ static void falcon_push_irq_moderation(struct efx_channel *channel)
 	BUILD_BUG_ON(FR_AA_TIMER_COMMAND_KER != FR_BZ_TIMER_COMMAND_P0);
 	efx_writed_page_locked(efx, &timer_cmd, FR_BZ_TIMER_COMMAND_P0,
 			       channel->channel);
-
-}
-
-/* Allocate buffer table entries for event queue */
-int efx_nic_probe_eventq(struct efx_channel *channel)
-{
-	struct efx_nic *efx = channel->efx;
-	BUILD_BUG_ON(EFX_EVQ_SIZE < 512 || EFX_EVQ_SIZE > 32768 ||
-		     EFX_EVQ_SIZE & EFX_EVQ_MASK);
-	return efx_alloc_special_buffer(efx, &channel->eventq,
-					EFX_EVQ_SIZE * sizeof(efx_qword_t));
-}
-
-void efx_nic_init_eventq(struct efx_channel *channel)
-{
-	efx_oword_t evq_ptr;
-	struct efx_nic *efx = channel->efx;
-
-	EFX_LOG(efx, "channel %d event queue in special buffers %d-%d\n",
-		channel->channel, channel->eventq.index,
-		channel->eventq.index + channel->eventq.entries - 1);
-
-	/* Pin event queue buffer */
-	efx_init_special_buffer(efx, &channel->eventq);
-
-	/* Fill event queue with all ones (i.e. empty events) */
-	memset(channel->eventq.addr, 0xff, channel->eventq.len);
-
-	/* Push event queue to card */
-	EFX_POPULATE_OWORD_3(evq_ptr,
-			     FRF_AZ_EVQ_EN, 1,
-			     FRF_AZ_EVQ_SIZE, __ffs(channel->eventq.entries),
-			     FRF_AZ_EVQ_BUF_BASE_ID, channel->eventq.index);
-	efx_writeo_table(efx, &evq_ptr, efx->type->evq_ptr_tbl_base,
-			 channel->channel);
-
-	efx->type->push_irq_moderation(channel);
-}
-
-void efx_nic_fini_eventq(struct efx_channel *channel)
-{
-	efx_oword_t eventq_ptr;
-	struct efx_nic *efx = channel->efx;
-
-	/* Remove event queue from card */
-	EFX_ZERO_OWORD(eventq_ptr);
-	efx_writeo_table(efx, &eventq_ptr, efx->type->evq_ptr_tbl_base,
-			 channel->channel);
-
-	/* Unpin event queue */
-	efx_fini_special_buffer(efx, &channel->eventq);
-}
-
-/* Free buffers backing event queue */
-void efx_nic_remove_eventq(struct efx_channel *channel)
-{
-	efx_free_special_buffer(channel->efx, &channel->eventq);
-}
-
-
-/* Generates a test event on the event queue.  A subsequent call to
- * process_eventq() should pick up the event and place the value of
- * "magic" into channel->eventq_magic;
- */
-void efx_nic_generate_test_event(struct efx_channel *channel, unsigned int magic)
-{
-	efx_qword_t test_event;
-
-	EFX_POPULATE_QWORD_2(test_event, FSF_AZ_EV_CODE,
-			     FSE_AZ_EV_CODE_DRV_GEN_EV,
-			     FSF_AZ_DRV_GEN_EV_MAGIC, magic);
-	efx_generate_event(channel, &test_event);
-}
-
-/**************************************************************************
- *
- * Flush handling
- *
- **************************************************************************/
-
-
-static void efx_poll_flush_events(struct efx_nic *efx)
-{
-	struct efx_channel *channel = &efx->channel[0];
-	struct efx_tx_queue *tx_queue;
-	struct efx_rx_queue *rx_queue;
-	unsigned int read_ptr = channel->eventq_read_ptr;
-	unsigned int end_ptr = (read_ptr - 1) & EFX_EVQ_MASK;
-
-	do {
-		efx_qword_t *event = efx_event(channel, read_ptr);
-		int ev_code, ev_sub_code, ev_queue;
-		bool ev_failed;
-
-		if (!efx_event_present(event))
-			break;
-
-		ev_code = EFX_QWORD_FIELD(*event, FSF_AZ_EV_CODE);
-		ev_sub_code = EFX_QWORD_FIELD(*event,
-					      FSF_AZ_DRIVER_EV_SUBCODE);
-		if (ev_code == FSE_AZ_EV_CODE_DRIVER_EV &&
-		    ev_sub_code == FSE_AZ_TX_DESCQ_FLS_DONE_EV) {
-			ev_queue = EFX_QWORD_FIELD(*event,
-						   FSF_AZ_DRIVER_EV_SUBDATA);
-			if (ev_queue < EFX_TX_QUEUE_COUNT) {
-				tx_queue = efx->tx_queue + ev_queue;
-				tx_queue->flushed = FLUSH_DONE;
-			}
-		} else if (ev_code == FSE_AZ_EV_CODE_DRIVER_EV &&
-			   ev_sub_code == FSE_AZ_RX_DESCQ_FLS_DONE_EV) {
-			ev_queue = EFX_QWORD_FIELD(
-				*event, FSF_AZ_DRIVER_EV_RX_DESCQ_ID);
-			ev_failed = EFX_QWORD_FIELD(
-				*event, FSF_AZ_DRIVER_EV_RX_FLUSH_FAIL);
-			if (ev_queue < efx->n_rx_queues) {
-				rx_queue = efx->rx_queue + ev_queue;
-				rx_queue->flushed =
-					ev_failed ? FLUSH_FAILED : FLUSH_DONE;
-			}
-		}
-
-		/* We're about to destroy the queue anyway, so
-		 * it's ok to throw away every non-flush event */
-		EFX_SET_QWORD(*event);
-
-		read_ptr = (read_ptr + 1) & EFX_EVQ_MASK;
-	} while (read_ptr != end_ptr);
-
-	channel->eventq_read_ptr = read_ptr;
 }
 
 static void falcon_deconfigure_mac_wrapper(struct efx_nic *efx);
@@ -1199,123 +134,6 @@ static void falcon_prepare_flush(struct efx_nic *efx)
 	msleep(10);
 }
 
-/* Handle tx and rx flushes at the same time, since they run in
- * parallel in the hardware and there's no reason for us to
- * serialise them */
-int efx_nic_flush_queues(struct efx_nic *efx)
-{
-	struct efx_rx_queue *rx_queue;
-	struct efx_tx_queue *tx_queue;
-	int i, tx_pending, rx_pending;
-
-	/* If necessary prepare the hardware for flushing */
-	efx->type->prepare_flush(efx);
-
-	/* Flush all tx queues in parallel */
-	efx_for_each_tx_queue(tx_queue, efx)
-		efx_flush_tx_queue(tx_queue);
-
-	/* The hardware supports four concurrent rx flushes, each of which may
-	 * need to be retried if there is an outstanding descriptor fetch */
-	for (i = 0; i < EFX_FLUSH_POLL_COUNT; ++i) {
-		rx_pending = tx_pending = 0;
-		efx_for_each_rx_queue(rx_queue, efx) {
-			if (rx_queue->flushed == FLUSH_PENDING)
-				++rx_pending;
-		}
-		efx_for_each_rx_queue(rx_queue, efx) {
-			if (rx_pending == EFX_RX_FLUSH_COUNT)
-				break;
-			if (rx_queue->flushed == FLUSH_FAILED ||
-			    rx_queue->flushed == FLUSH_NONE) {
-				efx_flush_rx_queue(rx_queue);
-				++rx_pending;
-			}
-		}
-		efx_for_each_tx_queue(tx_queue, efx) {
-			if (tx_queue->flushed != FLUSH_DONE)
-				++tx_pending;
-		}
-
-		if (rx_pending == 0 && tx_pending == 0)
-			return 0;
-
-		msleep(EFX_FLUSH_INTERVAL);
-		efx_poll_flush_events(efx);
-	}
-
-	/* Mark the queues as all flushed. We're going to return failure
-	 * leading to a reset, or fake up success anyway */
-	efx_for_each_tx_queue(tx_queue, efx) {
-		if (tx_queue->flushed != FLUSH_DONE)
-			EFX_ERR(efx, "tx queue %d flush command timed out\n",
-				tx_queue->queue);
-		tx_queue->flushed = FLUSH_DONE;
-	}
-	efx_for_each_rx_queue(rx_queue, efx) {
-		if (rx_queue->flushed != FLUSH_DONE)
-			EFX_ERR(efx, "rx queue %d flush command timed out\n",
-				rx_queue->queue);
-		rx_queue->flushed = FLUSH_DONE;
-	}
-
-	if (EFX_WORKAROUND_7803(efx))
-		return 0;
-
-	return -ETIMEDOUT;
-}
-
-/**************************************************************************
- *
- * Hardware interrupts
- * The hardware interrupt handler does very little work; all the event
- * queue processing is carried out by per-channel tasklets.
- *
- **************************************************************************/
-
-/* Enable/disable/generate interrupts */
-static inline void efx_nic_interrupts(struct efx_nic *efx,
-				      bool enabled, bool force)
-{
-	efx_oword_t int_en_reg_ker;
-
-	EFX_POPULATE_OWORD_2(int_en_reg_ker,
-			     FRF_AZ_KER_INT_KER, force,
-			     FRF_AZ_DRV_INT_EN_KER, enabled);
-	efx_writeo(efx, &int_en_reg_ker, FR_AZ_INT_EN_KER);
-}
-
-void efx_nic_enable_interrupts(struct efx_nic *efx)
-{
-	struct efx_channel *channel;
-
-	EFX_ZERO_OWORD(*((efx_oword_t *) efx->irq_status.addr));
-	wmb(); /* Ensure interrupt vector is clear before interrupts enabled */
-
-	/* Enable interrupts */
-	efx_nic_interrupts(efx, true, false);
-
-	/* Force processing of all the channels to get the EVQ RPTRs up to
-	   date */
-	efx_for_each_channel(channel, efx)
-		efx_schedule_channel(channel);
-}
-
-void efx_nic_disable_interrupts(struct efx_nic *efx)
-{
-	/* Disable interrupts */
-	efx_nic_interrupts(efx, false, false);
-}
-
-/* Generate a test interrupt
- * Interrupt must already have been enabled, otherwise nasty things
- * may happen.
- */
-void efx_nic_generate_interrupt(struct efx_nic *efx)
-{
-	efx_nic_interrupts(efx, true, true);
-}
-
 /* Acknowledge a legacy interrupt from Falcon
  *
  * This acknowledges a legacy (not MSI) interrupt via INT_ACK_KER_REG.
@@ -1335,102 +153,6 @@ inline void falcon_irq_ack_a1(struct efx_nic *efx)
 	efx_readd(efx, &reg, FR_AA_WORK_AROUND_BROKEN_PCI_READS);
 }
 
-/* Process a fatal interrupt
- * Disable bus mastering ASAP and schedule a reset
- */
-irqreturn_t efx_nic_fatal_interrupt(struct efx_nic *efx)
-{
-	struct falcon_nic_data *nic_data = efx->nic_data;
-	efx_oword_t *int_ker = efx->irq_status.addr;
-	efx_oword_t fatal_intr;
-	int error, mem_perr;
-
-	efx_reado(efx, &fatal_intr, FR_AZ_FATAL_INTR_KER);
-	error = EFX_OWORD_FIELD(fatal_intr, FRF_AZ_FATAL_INTR);
-
-	EFX_ERR(efx, "SYSTEM ERROR " EFX_OWORD_FMT " status "
-		EFX_OWORD_FMT ": %s\n", EFX_OWORD_VAL(*int_ker),
-		EFX_OWORD_VAL(fatal_intr),
-		error ? "disabling bus mastering" : "no recognised error");
-	if (error == 0)
-		goto out;
-
-	/* If this is a memory parity error dump which blocks are offending */
-	mem_perr = EFX_OWORD_FIELD(fatal_intr, FRF_AZ_MEM_PERR_INT_KER);
-	if (mem_perr) {
-		efx_oword_t reg;
-		efx_reado(efx, &reg, FR_AZ_MEM_STAT);
-		EFX_ERR(efx, "SYSTEM ERROR: memory parity error "
-			EFX_OWORD_FMT "\n", EFX_OWORD_VAL(reg));
-	}
-
-	/* Disable both devices */
-	pci_clear_master(efx->pci_dev);
-	if (efx_nic_is_dual_func(efx))
-		pci_clear_master(nic_data->pci_dev2);
-	efx_nic_disable_interrupts(efx);
-
-	/* Count errors and reset or disable the NIC accordingly */
-	if (efx->int_error_count == 0 ||
-	    time_after(jiffies, efx->int_error_expire)) {
-		efx->int_error_count = 0;
-		efx->int_error_expire =
-			jiffies + EFX_INT_ERROR_EXPIRE * HZ;
-	}
-	if (++efx->int_error_count < EFX_MAX_INT_ERRORS) {
-		EFX_ERR(efx, "SYSTEM ERROR - reset scheduled\n");
-		efx_schedule_reset(efx, RESET_TYPE_INT_ERROR);
-	} else {
-		EFX_ERR(efx, "SYSTEM ERROR - max number of errors seen."
-			"NIC will be disabled\n");
-		efx_schedule_reset(efx, RESET_TYPE_DISABLE);
-	}
-out:
-	return IRQ_HANDLED;
-}
-
-/* Handle a legacy interrupt
- * Acknowledges the interrupt and schedule event queue processing.
- */
-static irqreturn_t efx_legacy_interrupt(int irq, void *dev_id)
-{
-	struct efx_nic *efx = dev_id;
-	efx_oword_t *int_ker = efx->irq_status.addr;
-	irqreturn_t result = IRQ_NONE;
-	struct efx_channel *channel;
-	efx_dword_t reg;
-	u32 queues;
-	int syserr;
-
-	/* Read the ISR which also ACKs the interrupts */
-	efx_readd(efx, &reg, FR_BZ_INT_ISR0);
-	queues = EFX_EXTRACT_DWORD(reg, 0, 31);
-
-	/* Check to see if we have a serious error condition */
-	syserr = EFX_OWORD_FIELD(*int_ker, FSF_AZ_NET_IVEC_FATAL_INT);
-	if (unlikely(syserr))
-		return efx_nic_fatal_interrupt(efx);
-
-	/* Schedule processing of any interrupting queues */
-	efx_for_each_channel(channel, efx) {
-		if ((queues & 1) ||
-		    efx_event_present(
-			    efx_event(channel, channel->eventq_read_ptr))) {
-			efx_schedule_channel(channel);
-			result = IRQ_HANDLED;
-		}
-		queues >>= 1;
-	}
-
-	if (result == IRQ_HANDLED) {
-		efx->last_irq_cpu = raw_smp_processor_id();
-		EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_DWORD_FMT "\n",
-			  irq, raw_smp_processor_id(), EFX_DWORD_VAL(reg));
-	}
-
-	return result;
-}
-
 
 irqreturn_t falcon_legacy_interrupt_a1(int irq, void *dev_id)
 {
@@ -1477,126 +199,6 @@ irqreturn_t falcon_legacy_interrupt_a1(int irq, void *dev_id)
 
 	return IRQ_HANDLED;
 }
-
-/* Handle an MSI interrupt
- *
- * Handle an MSI hardware interrupt.  This routine schedules event
- * queue processing.  No interrupt acknowledgement cycle is necessary.
- * Also, we never need to check that the interrupt is for us, since
- * MSI interrupts cannot be shared.
- */
-static irqreturn_t efx_msi_interrupt(int irq, void *dev_id)
-{
-	struct efx_channel *channel = dev_id;
-	struct efx_nic *efx = channel->efx;
-	efx_oword_t *int_ker = efx->irq_status.addr;
-	int syserr;
-
-	efx->last_irq_cpu = raw_smp_processor_id();
-	EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_OWORD_FMT "\n",
-		  irq, raw_smp_processor_id(), EFX_OWORD_VAL(*int_ker));
-
-	/* Check to see if we have a serious error condition */
-	syserr = EFX_OWORD_FIELD(*int_ker, FSF_AZ_NET_IVEC_FATAL_INT);
-	if (unlikely(syserr))
-		return efx_nic_fatal_interrupt(efx);
-
-	/* Schedule processing of the channel */
-	efx_schedule_channel(channel);
-
-	return IRQ_HANDLED;
-}
-
-
-/* Setup RSS indirection table.
- * This maps from the hash value of the packet to RXQ
- */
-static void efx_setup_rss_indir_table(struct efx_nic *efx)
-{
-	int i = 0;
-	unsigned long offset;
-	efx_dword_t dword;
-
-	if (efx_nic_rev(efx) < EFX_REV_FALCON_B0)
-		return;
-
-	for (offset = FR_BZ_RX_INDIRECTION_TBL;
-	     offset < FR_BZ_RX_INDIRECTION_TBL + 0x800;
-	     offset += 0x10) {
-		EFX_POPULATE_DWORD_1(dword, FRF_BZ_IT_QUEUE,
-				     i % efx->n_rx_queues);
-		efx_writed(efx, &dword, offset);
-		i++;
-	}
-}
-
-/* Hook interrupt handler(s)
- * Try MSI and then legacy interrupts.
- */
-int efx_nic_init_interrupt(struct efx_nic *efx)
-{
-	struct efx_channel *channel;
-	int rc;
-
-	if (!EFX_INT_MODE_USE_MSI(efx)) {
-		irq_handler_t handler;
-		if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0)
-			handler = efx_legacy_interrupt;
-		else
-			handler = falcon_legacy_interrupt_a1;
-
-		rc = request_irq(efx->legacy_irq, handler, IRQF_SHARED,
-				 efx->name, efx);
-		if (rc) {
-			EFX_ERR(efx, "failed to hook legacy IRQ %d\n",
-				efx->pci_dev->irq);
-			goto fail1;
-		}
-		return 0;
-	}
-
-	/* Hook MSI or MSI-X interrupt */
-	efx_for_each_channel(channel, efx) {
-		rc = request_irq(channel->irq, efx_msi_interrupt,
-				 IRQF_PROBE_SHARED, /* Not shared */
-				 channel->name, channel);
-		if (rc) {
-			EFX_ERR(efx, "failed to hook IRQ %d\n", channel->irq);
-			goto fail2;
-		}
-	}
-
-	return 0;
-
- fail2:
-	efx_for_each_channel(channel, efx)
-		free_irq(channel->irq, channel);
- fail1:
-	return rc;
-}
-
-void efx_nic_fini_interrupt(struct efx_nic *efx)
-{
-	struct efx_channel *channel;
-	efx_oword_t reg;
-
-	/* Disable MSI/MSI-X interrupts */
-	efx_for_each_channel(channel, efx) {
-		if (channel->irq)
-			free_irq(channel->irq, channel);
-	}
-
-	/* ACK legacy interrupt */
-	if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0)
-		efx_reado(efx, &reg, FR_BZ_INT_ISR0);
-	else
-		falcon_irq_ack_a1(efx);
-
-	/* Disable legacy interrupt */
-	if (efx->legacy_irq)
-		free_irq(efx->legacy_irq, efx);
-}
-
 /**************************************************************************
  *
  * EEPROM/flash
@@ -2440,68 +1042,6 @@ static const struct efx_nic_register_test falcon_b0_register_tests[] = {
 	  EFX_OWORD32(0x0003FF0F, 0x00000000, 0x00000000, 0x00000000) },
 };
 
-static bool efx_masked_compare_oword(const efx_oword_t *a, const efx_oword_t *b,
-				     const efx_oword_t *mask)
-{
-	return ((a->u64[0] ^ b->u64[0]) & mask->u64[0]) ||
-		((a->u64[1] ^ b->u64[1]) & mask->u64[1]);
-}
-
-int efx_nic_test_registers(struct efx_nic *efx,
-			   const struct efx_nic_register_test *regs,
-			   size_t n_regs)
-{
-	unsigned address = 0, i, j;
-	efx_oword_t mask, imask, original, reg, buf;
-
-	/* Falcon should be in loopback to isolate the XMAC from the PHY */
-	WARN_ON(!LOOPBACK_INTERNAL(efx));
-
-	for (i = 0; i < n_regs; ++i) {
-		address = regs[i].address;
-		mask = imask = regs[i].mask;
-		EFX_INVERT_OWORD(imask);
-
-		efx_reado(efx, &original, address);
-
-		/* bit sweep on and off */
-		for (j = 0; j < 128; j++) {
-			if (!EFX_EXTRACT_OWORD32(mask, j, j))
-				continue;
-
-			/* Test this testable bit can be set in isolation */
-			EFX_AND_OWORD(reg, original, mask);
-			EFX_SET_OWORD32(reg, j, j, 1);
-
-			efx_writeo(efx, &reg, address);
-			efx_reado(efx, &buf, address);
-
-			if (efx_masked_compare_oword(&reg, &buf, &mask))
-				goto fail;
-
-			/* Test this testable bit can be cleared in isolation */
-			EFX_OR_OWORD(reg, original, mask);
-			EFX_SET_OWORD32(reg, j, j, 0);
-
-			efx_writeo(efx, &reg, address);
-			efx_reado(efx, &buf, address);
-
-			if (efx_masked_compare_oword(&reg, &buf, &mask))
-				goto fail;
-		}
-
-		efx_writeo(efx, &original, address);
-	}
-
-	return 0;
-
-fail:
-	EFX_ERR(efx, "wrote "EFX_OWORD_FMT" read "EFX_OWORD_FMT
-		" at address 0x%x mask "EFX_OWORD_FMT"\n", EFX_OWORD_VAL(reg),
-		EFX_OWORD_VAL(buf), address, EFX_OWORD_VAL(mask));
-	return -EIO;
-}
-
 static int falcon_b0_test_registers(struct efx_nic *efx)
 {
 	return efx_nic_test_registers(efx, falcon_b0_register_tests,
@@ -2719,7 +1259,6 @@ static int falcon_spi_device_init(struct efx_nic *efx,
 	return 0;
 }
 
-
 static void falcon_remove_spi_devices(struct efx_nic *efx)
 {
 	kfree(efx->spi_eeprom);
@@ -2789,14 +1328,6 @@ static int falcon_probe_nvconfig(struct efx_nic *efx)
 	return rc;
 }
 
-u32 efx_nic_fpga_ver(struct efx_nic *efx)
-{
-	efx_oword_t altera_build;
-
-	efx_reado(efx, &altera_build, FR_AZ_ALTERA_BUILD);
-	return EFX_OWORD_FIELD(altera_build, FRF_AZ_ALTERA_BUILD_VER);
-}
-
 /* Probe all SPI devices on the NIC */
 static void falcon_probe_spi_devices(struct efx_nic *efx)
 {
@@ -3006,73 +1537,6 @@ static void falcon_init_rx_cfg(struct efx_nic *efx)
 	efx_writeo(efx, &reg, FR_AZ_RX_CFG);
 }
 
-void efx_nic_init_common(struct efx_nic *efx)
-{
-	efx_oword_t temp;
-
-	/* Set positions of descriptor caches in SRAM. */
-	EFX_POPULATE_OWORD_1(temp, FRF_AZ_SRM_TX_DC_BASE_ADR,
-			     efx->type->tx_dc_base / 8);
-	efx_writeo(efx, &temp, FR_AZ_SRM_TX_DC_CFG);
-	EFX_POPULATE_OWORD_1(temp, FRF_AZ_SRM_RX_DC_BASE_ADR,
-			     efx->type->rx_dc_base / 8);
-	efx_writeo(efx, &temp, FR_AZ_SRM_RX_DC_CFG);
-
-	/* Set TX descriptor cache size. */
-	BUILD_BUG_ON(TX_DC_ENTRIES != (8 << TX_DC_ENTRIES_ORDER));
-	EFX_POPULATE_OWORD_1(temp, FRF_AZ_TX_DC_SIZE, TX_DC_ENTRIES_ORDER);
-	efx_writeo(efx, &temp, FR_AZ_TX_DC_CFG);
-
-	/* Set RX descriptor cache size.  Set low watermark to size-8, as
-	 * this allows most efficient prefetching.
-	 */
-	BUILD_BUG_ON(RX_DC_ENTRIES != (8 << RX_DC_ENTRIES_ORDER));
-	EFX_POPULATE_OWORD_1(temp, FRF_AZ_RX_DC_SIZE, RX_DC_ENTRIES_ORDER);
-	efx_writeo(efx, &temp, FR_AZ_RX_DC_CFG);
-	EFX_POPULATE_OWORD_1(temp, FRF_AZ_RX_DC_PF_LWM, RX_DC_ENTRIES - 8);
-	efx_writeo(efx, &temp, FR_AZ_RX_DC_PF_WM);
-
-	/* Program INT_KER address */
-	EFX_POPULATE_OWORD_2(temp,
-			     FRF_AZ_NORM_INT_VEC_DIS_KER,
-			     EFX_INT_MODE_USE_MSI(efx),
-			     FRF_AZ_INT_ADR_KER, efx->irq_status.dma_addr);
-	efx_writeo(efx, &temp, FR_AZ_INT_ADR_KER);
-
-	/* Enable all the genuinely fatal interrupts.  (They are still
-	 * masked by the overall interrupt mask, controlled by
-	 * falcon_interrupts()).
-	 *
-	 * Note: All other fatal interrupts are enabled
-	 */
-	EFX_POPULATE_OWORD_3(temp,
-			     FRF_AZ_ILL_ADR_INT_KER_EN, 1,
-			     FRF_AZ_RBUF_OWN_INT_KER_EN, 1,
-			     FRF_AZ_TBUF_OWN_INT_KER_EN, 1);
-	EFX_INVERT_OWORD(temp);
-	efx_writeo(efx, &temp, FR_AZ_FATAL_INTR_KER);
-
-	efx_setup_rss_indir_table(efx);
-
-	/* Disable the ugly timer-based TX DMA backoff and allow TX DMA to be
-	 * controlled by the RX FIFO fill level. Set arbitration to one pkt/Q.
-	 */
-	efx_reado(efx, &temp, FR_AZ_TX_RESERVED);
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_RX_SPACER, 0xfe);
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_RX_SPACER_EN, 1);
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_ONE_PKT_PER_Q, 1);
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_PUSH_EN, 0);
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_DIS_NON_IP_EV, 1);
-	/* Enable SW_EV to inherit in char driver - assume harmless here */
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_SOFT_EVT_EN, 1);
-	/* Prefetch threshold 2 => fetch when descriptor cache half empty */
-	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_PREF_THRESHOLD, 2);
-	/* Squash TX of packets of 16 bytes or less */
-	if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0)
-		EFX_SET_OWORD_FIELD(temp, FRF_BZ_TX_FLUSH_MIN_LEN_EN, 1);
-	efx_writeo(efx, &temp, FR_AZ_TX_RESERVED);
-}
-
 /* This call performs hardware-specific global initialisation, such as
  * defining the descriptor cache sizes and number of RSS channels.
  * It does not set up any buffers, descriptor rings or event queues.
diff --git a/drivers/net/sfc/nic.c b/drivers/net/sfc/nic.c
new file mode 100644
index 0000000..55dbd79
--- /dev/null
+++ b/drivers/net/sfc/nic.c
@@ -0,0 +1,1548 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2005-2006 Fen Systems Ltd.
+ * Copyright 2006-2008 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/module.h>
+#include <linux/seq_file.h>
+#include "net_driver.h"
+#include "bitfield.h"
+#include "efx.h"
+#include "nic.h"
+#include "regs.h"
+#include "io.h"
+#include "workarounds.h"
+
+/**************************************************************************
+ *
+ * Configurable values
+ *
+ **************************************************************************
+ */
+
+/* This is set to 16 for a good reason.  In summary, if larger than
+ * 16, the descriptor cache holds more than a default socket
+ * buffer's worth of packets (for UDP we can only have at most one
+ * socket buffer's worth outstanding).  This combined with the fact
+ * that we only get 1 TX event per descriptor cache means the NIC
+ * goes idle.
+ */
+#define TX_DC_ENTRIES 16
+#define TX_DC_ENTRIES_ORDER 1
+
+#define RX_DC_ENTRIES 64
+#define RX_DC_ENTRIES_ORDER 3
+
+/* RX FIFO XOFF watermark
+ *
+ * When the amount of the RX FIFO increases used increases past this
+ * watermark send XOFF. Only used if RX flow control is enabled (ethtool -A)
+ * This also has an effect on RX/TX arbitration
+ */
+int efx_nic_rx_xoff_thresh = -1;
+module_param_named(rx_xoff_thresh_bytes, efx_nic_rx_xoff_thresh, int, 0644);
+MODULE_PARM_DESC(rx_xoff_thresh_bytes, "RX fifo XOFF threshold");
+
+/* RX FIFO XON watermark
+ *
+ * When the amount of the RX FIFO used decreases below this
+ * watermark send XON. Only used if TX flow control is enabled (ethtool -A)
+ * This also has an effect on RX/TX arbitration
+ */
+int efx_nic_rx_xon_thresh = -1;
+module_param_named(rx_xon_thresh_bytes, efx_nic_rx_xon_thresh, int, 0644);
+MODULE_PARM_DESC(rx_xon_thresh_bytes, "RX fifo XON threshold");
+
+/* If EFX_MAX_INT_ERRORS internal errors occur within
+ * EFX_INT_ERROR_EXPIRE seconds, we consider the NIC broken and
+ * disable it.
+ */
+#define EFX_INT_ERROR_EXPIRE 3600
+#define EFX_MAX_INT_ERRORS 5
+
+/* We poll for events every FLUSH_INTERVAL ms, and check FLUSH_POLL_COUNT times
+ */
+#define EFX_FLUSH_INTERVAL 10
+#define EFX_FLUSH_POLL_COUNT 100
+
+/* Size and alignment of special buffers (4KB) */
+#define EFX_BUF_SIZE 4096
+
+/* Depth of RX flush request fifo */
+#define EFX_RX_FLUSH_COUNT 4
+
+/**************************************************************************
+ *
+ * Solarstorm hardware access
+ *
+ **************************************************************************/
+
+static inline void efx_write_buf_tbl(struct efx_nic *efx, efx_qword_t *value,
+				     unsigned int index)
+{
+	efx_sram_writeq(efx, efx->membase + efx->type->buf_tbl_base,
+			value, index);
+}
+
+/* Read the current event from the event queue */
+static inline efx_qword_t *efx_event(struct efx_channel *channel,
+				     unsigned int index)
+{
+	return (((efx_qword_t *) (channel->eventq.addr)) + index);
+}
+
+/* See if an event is present
+ *
+ * We check both the high and low dword of the event for all ones.  We
+ * wrote all ones when we cleared the event, and no valid event can
+ * have all ones in either its high or low dwords.  This approach is
+ * robust against reordering.
+ *
+ * Note that using a single 64-bit comparison is incorrect; even
+ * though the CPU read will be atomic, the DMA write may not be.
+ */
+static inline int efx_event_present(efx_qword_t *event)
+{
+	return (!(EFX_DWORD_IS_ALL_ONES(event->dword[0]) |
+		  EFX_DWORD_IS_ALL_ONES(event->dword[1])));
+}
+
+static bool efx_masked_compare_oword(const efx_oword_t *a, const efx_oword_t *b,
+				     const efx_oword_t *mask)
+{
+	return ((a->u64[0] ^ b->u64[0]) & mask->u64[0]) ||
+		((a->u64[1] ^ b->u64[1]) & mask->u64[1]);
+}
+
+int efx_nic_test_registers(struct efx_nic *efx,
+			   const struct efx_nic_register_test *regs,
+			   size_t n_regs)
+{
+	unsigned address = 0, i, j;
+	efx_oword_t mask, imask, original, reg, buf;
+
+	/* Falcon should be in loopback to isolate the XMAC from the PHY */
+	WARN_ON(!LOOPBACK_INTERNAL(efx));
+
+	for (i = 0; i < n_regs; ++i) {
+		address = regs[i].address;
+		mask = imask = regs[i].mask;
+		EFX_INVERT_OWORD(imask);
+
+		efx_reado(efx, &original, address);
+
+		/* bit sweep on and off */
+		for (j = 0; j < 128; j++) {
+			if (!EFX_EXTRACT_OWORD32(mask, j, j))
+				continue;
+
+			/* Test this testable bit can be set in isolation */
+			EFX_AND_OWORD(reg, original, mask);
+			EFX_SET_OWORD32(reg, j, j, 1);
+
+			efx_writeo(efx, &reg, address);
+			efx_reado(efx, &buf, address);
+
+			if (efx_masked_compare_oword(&reg, &buf, &mask))
+				goto fail;
+
+			/* Test this testable bit can be cleared in isolation */
+			EFX_OR_OWORD(reg, original, mask);
+			EFX_SET_OWORD32(reg, j, j, 0);
+
+			efx_writeo(efx, &reg, address);
+			efx_reado(efx, &buf, address);
+
+			if (efx_masked_compare_oword(&reg, &buf, &mask))
+				goto fail;
+		}
+
+		efx_writeo(efx, &original, address);
+	}
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "wrote "EFX_OWORD_FMT" read "EFX_OWORD_FMT
+		" at address 0x%x mask "EFX_OWORD_FMT"\n", EFX_OWORD_VAL(reg),
+		EFX_OWORD_VAL(buf), address, EFX_OWORD_VAL(mask));
+	return -EIO;
+}
+
+/**************************************************************************
+ *
+ * Special buffer handling
+ * Special buffers are used for event queues and the TX and RX
+ * descriptor rings.
+ *
+ *************************************************************************/
+
+/*
+ * Initialise a special buffer
+ *
+ * This will define a buffer (previously allocated via
+ * efx_alloc_special_buffer()) in the buffer table, allowing
+ * it to be used for event queues, descriptor rings etc.
+ */
+static void
+efx_init_special_buffer(struct efx_nic *efx, struct efx_special_buffer *buffer)
+{
+	efx_qword_t buf_desc;
+	int index;
+	dma_addr_t dma_addr;
+	int i;
+
+	EFX_BUG_ON_PARANOID(!buffer->addr);
+
+	/* Write buffer descriptors to NIC */
+	for (i = 0; i < buffer->entries; i++) {
+		index = buffer->index + i;
+		dma_addr = buffer->dma_addr + (i * 4096);
+		EFX_LOG(efx, "mapping special buffer %d at %llx\n",
+			index, (unsigned long long)dma_addr);
+		EFX_POPULATE_QWORD_3(buf_desc,
+				     FRF_AZ_BUF_ADR_REGION, 0,
+				     FRF_AZ_BUF_ADR_FBUF, dma_addr >> 12,
+				     FRF_AZ_BUF_OWNER_ID_FBUF, 0);
+		efx_write_buf_tbl(efx, &buf_desc, index);
+	}
+}
+
+/* Unmaps a buffer and clears the buffer table entries */
+static void
+efx_fini_special_buffer(struct efx_nic *efx, struct efx_special_buffer *buffer)
+{
+	efx_oword_t buf_tbl_upd;
+	unsigned int start = buffer->index;
+	unsigned int end = (buffer->index + buffer->entries - 1);
+
+	if (!buffer->entries)
+		return;
+
+	EFX_LOG(efx, "unmapping special buffers %d-%d\n",
+		buffer->index, buffer->index + buffer->entries - 1);
+
+	EFX_POPULATE_OWORD_4(buf_tbl_upd,
+			     FRF_AZ_BUF_UPD_CMD, 0,
+			     FRF_AZ_BUF_CLR_CMD, 1,
+			     FRF_AZ_BUF_CLR_END_ID, end,
+			     FRF_AZ_BUF_CLR_START_ID, start);
+	efx_writeo(efx, &buf_tbl_upd, FR_AZ_BUF_TBL_UPD);
+}
+
+/*
+ * Allocate a new special buffer
+ *
+ * This allocates memory for a new buffer, clears it and allocates a
+ * new buffer ID range.  It does not write into the buffer table.
+ *
+ * This call will allocate 4KB buffers, since 8KB buffers can't be
+ * used for event queues and descriptor rings.
+ */
+static int efx_alloc_special_buffer(struct efx_nic *efx,
+				    struct efx_special_buffer *buffer,
+				    unsigned int len)
+{
+	len = ALIGN(len, EFX_BUF_SIZE);
+
+	buffer->addr = pci_alloc_consistent(efx->pci_dev, len,
+					    &buffer->dma_addr);
+	if (!buffer->addr)
+		return -ENOMEM;
+	buffer->len = len;
+	buffer->entries = len / EFX_BUF_SIZE;
+	BUG_ON(buffer->dma_addr & (EFX_BUF_SIZE - 1));
+
+	/* All zeros is a potentially valid event so memset to 0xff */
+	memset(buffer->addr, 0xff, len);
+
+	/* Select new buffer ID */
+	buffer->index = efx->next_buffer_table;
+	efx->next_buffer_table += buffer->entries;
+
+	EFX_LOG(efx, "allocating special buffers %d-%d at %llx+%x "
+		"(virt %p phys %llx)\n", buffer->index,
+		buffer->index + buffer->entries - 1,
+		(u64)buffer->dma_addr, len,
+		buffer->addr, (u64)virt_to_phys(buffer->addr));
+
+	return 0;
+}
+
+static void
+efx_free_special_buffer(struct efx_nic *efx, struct efx_special_buffer *buffer)
+{
+	if (!buffer->addr)
+		return;
+
+	EFX_LOG(efx, "deallocating special buffers %d-%d at %llx+%x "
+		"(virt %p phys %llx)\n", buffer->index,
+		buffer->index + buffer->entries - 1,
+		(u64)buffer->dma_addr, buffer->len,
+		buffer->addr, (u64)virt_to_phys(buffer->addr));
+
+	pci_free_consistent(efx->pci_dev, buffer->len, buffer->addr,
+			    buffer->dma_addr);
+	buffer->addr = NULL;
+	buffer->entries = 0;
+}
+
+/**************************************************************************
+ *
+ * Generic buffer handling
+ * These buffers are used for interrupt status and MAC stats
+ *
+ **************************************************************************/
+
+int efx_nic_alloc_buffer(struct efx_nic *efx, struct efx_buffer *buffer,
+			 unsigned int len)
+{
+	buffer->addr = pci_alloc_consistent(efx->pci_dev, len,
+					    &buffer->dma_addr);
+	if (!buffer->addr)
+		return -ENOMEM;
+	buffer->len = len;
+	memset(buffer->addr, 0, len);
+	return 0;
+}
+
+void efx_nic_free_buffer(struct efx_nic *efx, struct efx_buffer *buffer)
+{
+	if (buffer->addr) {
+		pci_free_consistent(efx->pci_dev, buffer->len,
+				    buffer->addr, buffer->dma_addr);
+		buffer->addr = NULL;
+	}
+}
+
+/**************************************************************************
+ *
+ * TX path
+ *
+ **************************************************************************/
+
+/* Returns a pointer to the specified transmit descriptor in the TX
+ * descriptor queue belonging to the specified channel.
+ */
+static inline efx_qword_t *
+efx_tx_desc(struct efx_tx_queue *tx_queue, unsigned int index)
+{
+	return (((efx_qword_t *) (tx_queue->txd.addr)) + index);
+}
+
+/* This writes to the TX_DESC_WPTR; write pointer for TX descriptor ring */
+static inline void efx_notify_tx_desc(struct efx_tx_queue *tx_queue)
+{
+	unsigned write_ptr;
+	efx_dword_t reg;
+
+	write_ptr = tx_queue->write_count & EFX_TXQ_MASK;
+	EFX_POPULATE_DWORD_1(reg, FRF_AZ_TX_DESC_WPTR_DWORD, write_ptr);
+	efx_writed_page(tx_queue->efx, &reg,
+			FR_AZ_TX_DESC_UPD_DWORD_P0, tx_queue->queue);
+}
+
+
+/* For each entry inserted into the software descriptor ring, create a
+ * descriptor in the hardware TX descriptor ring (in host memory), and
+ * write a doorbell.
+ */
+void efx_nic_push_buffers(struct efx_tx_queue *tx_queue)
+{
+
+	struct efx_tx_buffer *buffer;
+	efx_qword_t *txd;
+	unsigned write_ptr;
+
+	BUG_ON(tx_queue->write_count == tx_queue->insert_count);
+
+	do {
+		write_ptr = tx_queue->write_count & EFX_TXQ_MASK;
+		buffer = &tx_queue->buffer[write_ptr];
+		txd = efx_tx_desc(tx_queue, write_ptr);
+		++tx_queue->write_count;
+
+		/* Create TX descriptor ring entry */
+		EFX_POPULATE_QWORD_4(*txd,
+				     FSF_AZ_TX_KER_CONT, buffer->continuation,
+				     FSF_AZ_TX_KER_BYTE_COUNT, buffer->len,
+				     FSF_AZ_TX_KER_BUF_REGION, 0,
+				     FSF_AZ_TX_KER_BUF_ADDR, buffer->dma_addr);
+	} while (tx_queue->write_count != tx_queue->insert_count);
+
+	wmb(); /* Ensure descriptors are written before they are fetched */
+	efx_notify_tx_desc(tx_queue);
+}
+
+/* Allocate hardware resources for a TX queue */
+int efx_nic_probe_tx(struct efx_tx_queue *tx_queue)
+{
+	struct efx_nic *efx = tx_queue->efx;
+	BUILD_BUG_ON(EFX_TXQ_SIZE < 512 || EFX_TXQ_SIZE > 4096 ||
+		     EFX_TXQ_SIZE & EFX_TXQ_MASK);
+	return efx_alloc_special_buffer(efx, &tx_queue->txd,
+					EFX_TXQ_SIZE * sizeof(efx_qword_t));
+}
+
+void efx_nic_init_tx(struct efx_tx_queue *tx_queue)
+{
+	efx_oword_t tx_desc_ptr;
+	struct efx_nic *efx = tx_queue->efx;
+
+	tx_queue->flushed = FLUSH_NONE;
+
+	/* Pin TX descriptor ring */
+	efx_init_special_buffer(efx, &tx_queue->txd);
+
+	/* Push TX descriptor ring to card */
+	EFX_POPULATE_OWORD_10(tx_desc_ptr,
+			      FRF_AZ_TX_DESCQ_EN, 1,
+			      FRF_AZ_TX_ISCSI_DDIG_EN, 0,
+			      FRF_AZ_TX_ISCSI_HDIG_EN, 0,
+			      FRF_AZ_TX_DESCQ_BUF_BASE_ID, tx_queue->txd.index,
+			      FRF_AZ_TX_DESCQ_EVQ_ID,
+			      tx_queue->channel->channel,
+			      FRF_AZ_TX_DESCQ_OWNER_ID, 0,
+			      FRF_AZ_TX_DESCQ_LABEL, tx_queue->queue,
+			      FRF_AZ_TX_DESCQ_SIZE,
+			      __ffs(tx_queue->txd.entries),
+			      FRF_AZ_TX_DESCQ_TYPE, 0,
+			      FRF_BZ_TX_NON_IP_DROP_DIS, 1);
+
+	if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
+		int csum = tx_queue->queue == EFX_TX_QUEUE_OFFLOAD_CSUM;
+		EFX_SET_OWORD_FIELD(tx_desc_ptr, FRF_BZ_TX_IP_CHKSM_DIS, !csum);
+		EFX_SET_OWORD_FIELD(tx_desc_ptr, FRF_BZ_TX_TCP_CHKSM_DIS,
+				    !csum);
+	}
+
+	efx_writeo_table(efx, &tx_desc_ptr, efx->type->txd_ptr_tbl_base,
+			 tx_queue->queue);
+
+	if (efx_nic_rev(efx) < EFX_REV_FALCON_B0) {
+		efx_oword_t reg;
+
+		/* Only 128 bits in this register */
+		BUILD_BUG_ON(EFX_TX_QUEUE_COUNT >= 128);
+
+		efx_reado(efx, &reg, FR_AA_TX_CHKSM_CFG);
+		if (tx_queue->queue == EFX_TX_QUEUE_OFFLOAD_CSUM)
+			clear_bit_le(tx_queue->queue, (void *)&reg);
+		else
+			set_bit_le(tx_queue->queue, (void *)&reg);
+		efx_writeo(efx, &reg, FR_AA_TX_CHKSM_CFG);
+	}
+}
+
+static void efx_flush_tx_queue(struct efx_tx_queue *tx_queue)
+{
+	struct efx_nic *efx = tx_queue->efx;
+	efx_oword_t tx_flush_descq;
+
+	tx_queue->flushed = FLUSH_PENDING;
+
+	/* Post a flush command */
+	EFX_POPULATE_OWORD_2(tx_flush_descq,
+			     FRF_AZ_TX_FLUSH_DESCQ_CMD, 1,
+			     FRF_AZ_TX_FLUSH_DESCQ, tx_queue->queue);
+	efx_writeo(efx, &tx_flush_descq, FR_AZ_TX_FLUSH_DESCQ);
+}
+
+void efx_nic_fini_tx(struct efx_tx_queue *tx_queue)
+{
+	struct efx_nic *efx = tx_queue->efx;
+	efx_oword_t tx_desc_ptr;
+
+	/* The queue should have been flushed */
+	WARN_ON(tx_queue->flushed != FLUSH_DONE);
+
+	/* Remove TX descriptor ring from card */
+	EFX_ZERO_OWORD(tx_desc_ptr);
+	efx_writeo_table(efx, &tx_desc_ptr, efx->type->txd_ptr_tbl_base,
+			 tx_queue->queue);
+
+	/* Unpin TX descriptor ring */
+	efx_fini_special_buffer(efx, &tx_queue->txd);
+}
+
+/* Free buffers backing TX queue */
+void efx_nic_remove_tx(struct efx_tx_queue *tx_queue)
+{
+	efx_free_special_buffer(tx_queue->efx, &tx_queue->txd);
+}
+
+/**************************************************************************
+ *
+ * RX path
+ *
+ **************************************************************************/
+
+/* Returns a pointer to the specified descriptor in the RX descriptor queue */
+static inline efx_qword_t *
+efx_rx_desc(struct efx_rx_queue *rx_queue, unsigned int index)
+{
+	return (((efx_qword_t *) (rx_queue->rxd.addr)) + index);
+}
+
+/* This creates an entry in the RX descriptor queue */
+static inline void
+efx_build_rx_desc(struct efx_rx_queue *rx_queue, unsigned index)
+{
+	struct efx_rx_buffer *rx_buf;
+	efx_qword_t *rxd;
+
+	rxd = efx_rx_desc(rx_queue, index);
+	rx_buf = efx_rx_buffer(rx_queue, index);
+	EFX_POPULATE_QWORD_3(*rxd,
+			     FSF_AZ_RX_KER_BUF_SIZE,
+			     rx_buf->len -
+			     rx_queue->efx->type->rx_buffer_padding,
+			     FSF_AZ_RX_KER_BUF_REGION, 0,
+			     FSF_AZ_RX_KER_BUF_ADDR, rx_buf->dma_addr);
+}
+
+/* This writes to the RX_DESC_WPTR register for the specified receive
+ * descriptor ring.
+ */
+void efx_nic_notify_rx_desc(struct efx_rx_queue *rx_queue)
+{
+	efx_dword_t reg;
+	unsigned write_ptr;
+
+	while (rx_queue->notified_count != rx_queue->added_count) {
+		efx_build_rx_desc(rx_queue,
+				  rx_queue->notified_count &
+				  EFX_RXQ_MASK);
+		++rx_queue->notified_count;
+	}
+
+	wmb();
+	write_ptr = rx_queue->added_count & EFX_RXQ_MASK;
+	EFX_POPULATE_DWORD_1(reg, FRF_AZ_RX_DESC_WPTR_DWORD, write_ptr);
+	efx_writed_page(rx_queue->efx, &reg,
+			FR_AZ_RX_DESC_UPD_DWORD_P0, rx_queue->queue);
+}
+
+int efx_nic_probe_rx(struct efx_rx_queue *rx_queue)
+{
+	struct efx_nic *efx = rx_queue->efx;
+	BUILD_BUG_ON(EFX_RXQ_SIZE < 512 || EFX_RXQ_SIZE > 4096 ||
+		     EFX_RXQ_SIZE & EFX_RXQ_MASK);
+	return efx_alloc_special_buffer(efx, &rx_queue->rxd,
+					EFX_RXQ_SIZE * sizeof(efx_qword_t));
+}
+
+void efx_nic_init_rx(struct efx_rx_queue *rx_queue)
+{
+	efx_oword_t rx_desc_ptr;
+	struct efx_nic *efx = rx_queue->efx;
+	bool is_b0 = efx_nic_rev(efx) >= EFX_REV_FALCON_B0;
+	bool iscsi_digest_en = is_b0;
+
+	EFX_LOG(efx, "RX queue %d ring in special buffers %d-%d\n",
+		rx_queue->queue, rx_queue->rxd.index,
+		rx_queue->rxd.index + rx_queue->rxd.entries - 1);
+
+	rx_queue->flushed = FLUSH_NONE;
+
+	/* Pin RX descriptor ring */
+	efx_init_special_buffer(efx, &rx_queue->rxd);
+
+	/* Push RX descriptor ring to card */
+	EFX_POPULATE_OWORD_10(rx_desc_ptr,
+			      FRF_AZ_RX_ISCSI_DDIG_EN, iscsi_digest_en,
+			      FRF_AZ_RX_ISCSI_HDIG_EN, iscsi_digest_en,
+			      FRF_AZ_RX_DESCQ_BUF_BASE_ID, rx_queue->rxd.index,
+			      FRF_AZ_RX_DESCQ_EVQ_ID,
+			      rx_queue->channel->channel,
+			      FRF_AZ_RX_DESCQ_OWNER_ID, 0,
+			      FRF_AZ_RX_DESCQ_LABEL, rx_queue->queue,
+			      FRF_AZ_RX_DESCQ_SIZE,
+			      __ffs(rx_queue->rxd.entries),
+			      FRF_AZ_RX_DESCQ_TYPE, 0 /* kernel queue */ ,
+			      /* For >=B0 this is scatter so disable */
+			      FRF_AZ_RX_DESCQ_JUMBO, !is_b0,
+			      FRF_AZ_RX_DESCQ_EN, 1);
+	efx_writeo_table(efx, &rx_desc_ptr, efx->type->rxd_ptr_tbl_base,
+			 rx_queue->queue);
+}
+
+static void efx_flush_rx_queue(struct efx_rx_queue *rx_queue)
+{
+	struct efx_nic *efx = rx_queue->efx;
+	efx_oword_t rx_flush_descq;
+
+	rx_queue->flushed = FLUSH_PENDING;
+
+	/* Post a flush command */
+	EFX_POPULATE_OWORD_2(rx_flush_descq,
+			     FRF_AZ_RX_FLUSH_DESCQ_CMD, 1,
+			     FRF_AZ_RX_FLUSH_DESCQ, rx_queue->queue);
+	efx_writeo(efx, &rx_flush_descq, FR_AZ_RX_FLUSH_DESCQ);
+}
+
+void efx_nic_fini_rx(struct efx_rx_queue *rx_queue)
+{
+	efx_oword_t rx_desc_ptr;
+	struct efx_nic *efx = rx_queue->efx;
+
+	/* The queue should already have been flushed */
+	WARN_ON(rx_queue->flushed != FLUSH_DONE);
+
+	/* Remove RX descriptor ring from card */
+	EFX_ZERO_OWORD(rx_desc_ptr);
+	efx_writeo_table(efx, &rx_desc_ptr, efx->type->rxd_ptr_tbl_base,
+			 rx_queue->queue);
+
+	/* Unpin RX descriptor ring */
+	efx_fini_special_buffer(efx, &rx_queue->rxd);
+}
+
+/* Free buffers backing RX queue */
+void efx_nic_remove_rx(struct efx_rx_queue *rx_queue)
+{
+	efx_free_special_buffer(rx_queue->efx, &rx_queue->rxd);
+}
+
+/**************************************************************************
+ *
+ * Event queue processing
+ * Event queues are processed by per-channel tasklets.
+ *
+ **************************************************************************/
+
+/* Update a channel's event queue's read pointer (RPTR) register
+ *
+ * This writes the EVQ_RPTR_REG register for the specified channel's
+ * event queue.
+ *
+ * Note that EVQ_RPTR_REG contains the index of the "last read" event,
+ * whereas channel->eventq_read_ptr contains the index of the "next to
+ * read" event.
+ */
+void efx_nic_eventq_read_ack(struct efx_channel *channel)
+{
+	efx_dword_t reg;
+	struct efx_nic *efx = channel->efx;
+
+	EFX_POPULATE_DWORD_1(reg, FRF_AZ_EVQ_RPTR, channel->eventq_read_ptr);
+	efx_writed_table(efx, &reg, efx->type->evq_rptr_tbl_base,
+			 channel->channel);
+}
+
+/* Use HW to insert a SW defined event */
+void efx_generate_event(struct efx_channel *channel, efx_qword_t *event)
+{
+	efx_oword_t drv_ev_reg;
+
+	BUILD_BUG_ON(FRF_AZ_DRV_EV_DATA_LBN != 0 ||
+		     FRF_AZ_DRV_EV_DATA_WIDTH != 64);
+	drv_ev_reg.u32[0] = event->u32[0];
+	drv_ev_reg.u32[1] = event->u32[1];
+	drv_ev_reg.u32[2] = 0;
+	drv_ev_reg.u32[3] = 0;
+	EFX_SET_OWORD_FIELD(drv_ev_reg, FRF_AZ_DRV_EV_QID, channel->channel);
+	efx_writeo(channel->efx, &drv_ev_reg, FR_AZ_DRV_EV);
+}
+
+/* Handle a transmit completion event
+ *
+ * The NIC batches TX completion events; the message we receive is of
+ * the form "complete all TX events up to this index".
+ */
+static void
+efx_handle_tx_event(struct efx_channel *channel, efx_qword_t *event)
+{
+	unsigned int tx_ev_desc_ptr;
+	unsigned int tx_ev_q_label;
+	struct efx_tx_queue *tx_queue;
+	struct efx_nic *efx = channel->efx;
+
+	if (likely(EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_COMP))) {
+		/* Transmit completion */
+		tx_ev_desc_ptr = EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_DESC_PTR);
+		tx_ev_q_label = EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_Q_LABEL);
+		tx_queue = &efx->tx_queue[tx_ev_q_label];
+		channel->irq_mod_score +=
+			(tx_ev_desc_ptr - tx_queue->read_count) &
+			EFX_TXQ_MASK;
+		efx_xmit_done(tx_queue, tx_ev_desc_ptr);
+	} else if (EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_WQ_FF_FULL)) {
+		/* Rewrite the FIFO write pointer */
+		tx_ev_q_label = EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_Q_LABEL);
+		tx_queue = &efx->tx_queue[tx_ev_q_label];
+
+		if (efx_dev_registered(efx))
+			netif_tx_lock(efx->net_dev);
+		efx_notify_tx_desc(tx_queue);
+		if (efx_dev_registered(efx))
+			netif_tx_unlock(efx->net_dev);
+	} else if (EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_PKT_ERR) &&
+		   EFX_WORKAROUND_10727(efx)) {
+		efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
+	} else {
+		EFX_ERR(efx, "channel %d unexpected TX event "
+			EFX_QWORD_FMT"\n", channel->channel,
+			EFX_QWORD_VAL(*event));
+	}
+}
+
+/* Detect errors included in the rx_evt_pkt_ok bit. */
+static void efx_handle_rx_not_ok(struct efx_rx_queue *rx_queue,
+				 const efx_qword_t *event,
+				 bool *rx_ev_pkt_ok,
+				 bool *discard)
+{
+	struct efx_nic *efx = rx_queue->efx;
+	bool rx_ev_buf_owner_id_err, rx_ev_ip_hdr_chksum_err;
+	bool rx_ev_tcp_udp_chksum_err, rx_ev_eth_crc_err;
+	bool rx_ev_frm_trunc, rx_ev_drib_nib, rx_ev_tobe_disc;
+	bool rx_ev_other_err, rx_ev_pause_frm;
+	bool rx_ev_hdr_type, rx_ev_mcast_pkt;
+	unsigned rx_ev_pkt_type;
+
+	rx_ev_hdr_type = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_HDR_TYPE);
+	rx_ev_mcast_pkt = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_MCAST_PKT);
+	rx_ev_tobe_disc = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_TOBE_DISC);
+	rx_ev_pkt_type = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_PKT_TYPE);
+	rx_ev_buf_owner_id_err = EFX_QWORD_FIELD(*event,
+						 FSF_AZ_RX_EV_BUF_OWNER_ID_ERR);
+	rx_ev_ip_hdr_chksum_err = EFX_QWORD_FIELD(*event,
+						  FSF_AZ_RX_EV_IP_HDR_CHKSUM_ERR);
+	rx_ev_tcp_udp_chksum_err = EFX_QWORD_FIELD(*event,
+						   FSF_AZ_RX_EV_TCP_UDP_CHKSUM_ERR);
+	rx_ev_eth_crc_err = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_ETH_CRC_ERR);
+	rx_ev_frm_trunc = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_FRM_TRUNC);
+	rx_ev_drib_nib = ((efx_nic_rev(efx) >= EFX_REV_FALCON_B0) ?
+			  0 : EFX_QWORD_FIELD(*event, FSF_AA_RX_EV_DRIB_NIB));
+	rx_ev_pause_frm = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_PAUSE_FRM_ERR);
+
+	/* Every error apart from tobe_disc and pause_frm */
+	rx_ev_other_err = (rx_ev_drib_nib | rx_ev_tcp_udp_chksum_err |
+			   rx_ev_buf_owner_id_err | rx_ev_eth_crc_err |
+			   rx_ev_frm_trunc | rx_ev_ip_hdr_chksum_err);
+
+	/* Count errors that are not in MAC stats.  Ignore expected
+	 * checksum errors during self-test. */
+	if (rx_ev_frm_trunc)
+		++rx_queue->channel->n_rx_frm_trunc;
+	else if (rx_ev_tobe_disc)
+		++rx_queue->channel->n_rx_tobe_disc;
+	else if (!efx->loopback_selftest) {
+		if (rx_ev_ip_hdr_chksum_err)
+			++rx_queue->channel->n_rx_ip_hdr_chksum_err;
+		else if (rx_ev_tcp_udp_chksum_err)
+			++rx_queue->channel->n_rx_tcp_udp_chksum_err;
+	}
+
+	/* The frame must be discarded if any of these are true. */
+	*discard = (rx_ev_eth_crc_err | rx_ev_frm_trunc | rx_ev_drib_nib |
+		    rx_ev_tobe_disc | rx_ev_pause_frm);
+
+	/* TOBE_DISC is expected on unicast mismatches; don't print out an
+	 * error message.  FRM_TRUNC indicates RXDP dropped the packet due
+	 * to a FIFO overflow.
+	 */
+#ifdef EFX_ENABLE_DEBUG
+	if (rx_ev_other_err) {
+		EFX_INFO_RL(efx, " RX queue %d unexpected RX event "
+			    EFX_QWORD_FMT "%s%s%s%s%s%s%s%s\n",
+			    rx_queue->queue, EFX_QWORD_VAL(*event),
+			    rx_ev_buf_owner_id_err ? " [OWNER_ID_ERR]" : "",
+			    rx_ev_ip_hdr_chksum_err ?
+			    " [IP_HDR_CHKSUM_ERR]" : "",
+			    rx_ev_tcp_udp_chksum_err ?
+			    " [TCP_UDP_CHKSUM_ERR]" : "",
+			    rx_ev_eth_crc_err ? " [ETH_CRC_ERR]" : "",
+			    rx_ev_frm_trunc ? " [FRM_TRUNC]" : "",
+			    rx_ev_drib_nib ? " [DRIB_NIB]" : "",
+			    rx_ev_tobe_disc ? " [TOBE_DISC]" : "",
+			    rx_ev_pause_frm ? " [PAUSE]" : "");
+	}
+#endif
+}
+
+/* Handle receive events that are not in-order. */
+static void
+efx_handle_rx_bad_index(struct efx_rx_queue *rx_queue, unsigned index)
+{
+	struct efx_nic *efx = rx_queue->efx;
+	unsigned expected, dropped;
+
+	expected = rx_queue->removed_count & EFX_RXQ_MASK;
+	dropped = (index - expected) & EFX_RXQ_MASK;
+	EFX_INFO(efx, "dropped %d events (index=%d expected=%d)\n",
+		dropped, index, expected);
+
+	efx_schedule_reset(efx, EFX_WORKAROUND_5676(efx) ?
+			   RESET_TYPE_RX_RECOVERY : RESET_TYPE_DISABLE);
+}
+
+/* Handle a packet received event
+ *
+ * The NIC gives a "discard" flag if it's a unicast packet with the
+ * wrong destination address
+ * Also "is multicast" and "matches multicast filter" flags can be used to
+ * discard non-matching multicast packets.
+ */
+static void
+efx_handle_rx_event(struct efx_channel *channel, const efx_qword_t *event)
+{
+	unsigned int rx_ev_desc_ptr, rx_ev_byte_cnt;
+	unsigned int rx_ev_hdr_type, rx_ev_mcast_pkt;
+	unsigned expected_ptr;
+	bool rx_ev_pkt_ok, discard = false, checksummed;
+	struct efx_rx_queue *rx_queue;
+	struct efx_nic *efx = channel->efx;
+
+	/* Basic packet information */
+	rx_ev_byte_cnt = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_BYTE_CNT);
+	rx_ev_pkt_ok = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_PKT_OK);
+	rx_ev_hdr_type = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_HDR_TYPE);
+	WARN_ON(EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_JUMBO_CONT));
+	WARN_ON(EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_SOP) != 1);
+	WARN_ON(EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_Q_LABEL) !=
+		channel->channel);
+
+	rx_queue = &efx->rx_queue[channel->channel];
+
+	rx_ev_desc_ptr = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_DESC_PTR);
+	expected_ptr = rx_queue->removed_count & EFX_RXQ_MASK;
+	if (unlikely(rx_ev_desc_ptr != expected_ptr))
+		efx_handle_rx_bad_index(rx_queue, rx_ev_desc_ptr);
+
+	if (likely(rx_ev_pkt_ok)) {
+		/* If packet is marked as OK and packet type is TCP/IP or
+		 * UDP/IP, then we can rely on the hardware checksum.
+		 */
+		checksummed =
+			likely(efx->rx_checksum_enabled) &&
+			(rx_ev_hdr_type == FSE_CZ_RX_EV_HDR_TYPE_IPV4V6_TCP ||
+			 rx_ev_hdr_type == FSE_CZ_RX_EV_HDR_TYPE_IPV4V6_UDP);
+	} else {
+		efx_handle_rx_not_ok(rx_queue, event, &rx_ev_pkt_ok, &discard);
+		checksummed = false;
+	}
+
+	/* Detect multicast packets that didn't match the filter */
+	rx_ev_mcast_pkt = EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_MCAST_PKT);
+	if (rx_ev_mcast_pkt) {
+		unsigned int rx_ev_mcast_hash_match =
+			EFX_QWORD_FIELD(*event, FSF_AZ_RX_EV_MCAST_HASH_MATCH);
+
+		if (unlikely(!rx_ev_mcast_hash_match)) {
+			++channel->n_rx_mcast_mismatch;
+			discard = true;
+		}
+	}
+
+	channel->irq_mod_score += 2;
+
+	/* Handle received packet */
+	efx_rx_packet(rx_queue, rx_ev_desc_ptr, rx_ev_byte_cnt,
+		      checksummed, discard);
+}
+
+/* Global events are basically PHY events */
+static void
+efx_handle_global_event(struct efx_channel *channel, efx_qword_t *event)
+{
+	struct efx_nic *efx = channel->efx;
+	bool handled = false;
+
+	if (EFX_QWORD_FIELD(*event, FSF_AB_GLB_EV_G_PHY0_INTR) ||
+	    EFX_QWORD_FIELD(*event, FSF_AB_GLB_EV_XG_PHY0_INTR) ||
+	    EFX_QWORD_FIELD(*event, FSF_AB_GLB_EV_XFP_PHY0_INTR)) {
+		/* Ignored */
+		handled = true;
+	}
+
+	if ((efx_nic_rev(efx) >= EFX_REV_FALCON_B0) &&
+	    EFX_QWORD_FIELD(*event, FSF_BB_GLB_EV_XG_MGT_INTR)) {
+		efx->xmac_poll_required = true;
+		handled = true;
+	}
+
+	if (efx_nic_rev(efx) <= EFX_REV_FALCON_A1 ?
+	    EFX_QWORD_FIELD(*event, FSF_AA_GLB_EV_RX_RECOVERY) :
+	    EFX_QWORD_FIELD(*event, FSF_BB_GLB_EV_RX_RECOVERY)) {
+		EFX_ERR(efx, "channel %d seen global RX_RESET "
+			"event. Resetting.\n", channel->channel);
+
+		atomic_inc(&efx->rx_reset);
+		efx_schedule_reset(efx, EFX_WORKAROUND_6555(efx) ?
+				   RESET_TYPE_RX_RECOVERY : RESET_TYPE_DISABLE);
+		handled = true;
+	}
+
+	if (!handled)
+		EFX_ERR(efx, "channel %d unknown global event "
+			EFX_QWORD_FMT "\n", channel->channel,
+			EFX_QWORD_VAL(*event));
+}
+
+static void
+efx_handle_driver_event(struct efx_channel *channel, efx_qword_t *event)
+{
+	struct efx_nic *efx = channel->efx;
+	unsigned int ev_sub_code;
+	unsigned int ev_sub_data;
+
+	ev_sub_code = EFX_QWORD_FIELD(*event, FSF_AZ_DRIVER_EV_SUBCODE);
+	ev_sub_data = EFX_QWORD_FIELD(*event, FSF_AZ_DRIVER_EV_SUBDATA);
+
+	switch (ev_sub_code) {
+	case FSE_AZ_TX_DESCQ_FLS_DONE_EV:
+		EFX_TRACE(efx, "channel %d TXQ %d flushed\n",
+			  channel->channel, ev_sub_data);
+		break;
+	case FSE_AZ_RX_DESCQ_FLS_DONE_EV:
+		EFX_TRACE(efx, "channel %d RXQ %d flushed\n",
+			  channel->channel, ev_sub_data);
+		break;
+	case FSE_AZ_EVQ_INIT_DONE_EV:
+		EFX_LOG(efx, "channel %d EVQ %d initialised\n",
+			channel->channel, ev_sub_data);
+		break;
+	case FSE_AZ_SRM_UPD_DONE_EV:
+		EFX_TRACE(efx, "channel %d SRAM update done\n",
+			  channel->channel);
+		break;
+	case FSE_AZ_WAKE_UP_EV:
+		EFX_TRACE(efx, "channel %d RXQ %d wakeup event\n",
+			  channel->channel, ev_sub_data);
+		break;
+	case FSE_AZ_TIMER_EV:
+		EFX_TRACE(efx, "channel %d RX queue %d timer expired\n",
+			  channel->channel, ev_sub_data);
+		break;
+	case FSE_AA_RX_RECOVER_EV:
+		EFX_ERR(efx, "channel %d seen DRIVER RX_RESET event. "
+			"Resetting.\n", channel->channel);
+		atomic_inc(&efx->rx_reset);
+		efx_schedule_reset(efx,
+				   EFX_WORKAROUND_6555(efx) ?
+				   RESET_TYPE_RX_RECOVERY :
+				   RESET_TYPE_DISABLE);
+		break;
+	case FSE_BZ_RX_DSC_ERROR_EV:
+		EFX_ERR(efx, "RX DMA Q %d reports descriptor fetch error."
+			" RX Q %d is disabled.\n", ev_sub_data, ev_sub_data);
+		efx_schedule_reset(efx, RESET_TYPE_RX_DESC_FETCH);
+		break;
+	case FSE_BZ_TX_DSC_ERROR_EV:
+		EFX_ERR(efx, "TX DMA Q %d reports descriptor fetch error."
+			" TX Q %d is disabled.\n", ev_sub_data, ev_sub_data);
+		efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
+		break;
+	default:
+		EFX_TRACE(efx, "channel %d unknown driver event code %d "
+			  "data %04x\n", channel->channel, ev_sub_code,
+			  ev_sub_data);
+		break;
+	}
+}
+
+int efx_nic_process_eventq(struct efx_channel *channel, int rx_quota)
+{
+	unsigned int read_ptr;
+	efx_qword_t event, *p_event;
+	int ev_code;
+	int rx_packets = 0;
+
+	read_ptr = channel->eventq_read_ptr;
+
+	do {
+		p_event = efx_event(channel, read_ptr);
+		event = *p_event;
+
+		if (!efx_event_present(&event))
+			/* End of events */
+			break;
+
+		EFX_TRACE(channel->efx, "channel %d event is "EFX_QWORD_FMT"\n",
+			  channel->channel, EFX_QWORD_VAL(event));
+
+		/* Clear this event by marking it all ones */
+		EFX_SET_QWORD(*p_event);
+
+		ev_code = EFX_QWORD_FIELD(event, FSF_AZ_EV_CODE);
+
+		switch (ev_code) {
+		case FSE_AZ_EV_CODE_RX_EV:
+			efx_handle_rx_event(channel, &event);
+			++rx_packets;
+			break;
+		case FSE_AZ_EV_CODE_TX_EV:
+			efx_handle_tx_event(channel, &event);
+			break;
+		case FSE_AZ_EV_CODE_DRV_GEN_EV:
+			channel->eventq_magic = EFX_QWORD_FIELD(
+				event, FSF_AZ_DRV_GEN_EV_MAGIC);
+			EFX_LOG(channel->efx, "channel %d received generated "
+				"event "EFX_QWORD_FMT"\n", channel->channel,
+				EFX_QWORD_VAL(event));
+			break;
+		case FSE_AZ_EV_CODE_GLOBAL_EV:
+			efx_handle_global_event(channel, &event);
+			break;
+		case FSE_AZ_EV_CODE_DRIVER_EV:
+			efx_handle_driver_event(channel, &event);
+			break;
+		default:
+			EFX_ERR(channel->efx, "channel %d unknown event type %d"
+				" (data " EFX_QWORD_FMT ")\n", channel->channel,
+				ev_code, EFX_QWORD_VAL(event));
+		}
+
+		/* Increment read pointer */
+		read_ptr = (read_ptr + 1) & EFX_EVQ_MASK;
+
+	} while (rx_packets < rx_quota);
+
+	channel->eventq_read_ptr = read_ptr;
+	return rx_packets;
+}
+
+
+/* Allocate buffer table entries for event queue */
+int efx_nic_probe_eventq(struct efx_channel *channel)
+{
+	struct efx_nic *efx = channel->efx;
+	BUILD_BUG_ON(EFX_EVQ_SIZE < 512 || EFX_EVQ_SIZE > 32768 ||
+		     EFX_EVQ_SIZE & EFX_EVQ_MASK);
+	return efx_alloc_special_buffer(efx, &channel->eventq,
+					EFX_EVQ_SIZE * sizeof(efx_qword_t));
+}
+
+void efx_nic_init_eventq(struct efx_channel *channel)
+{
+	efx_oword_t evq_ptr;
+	struct efx_nic *efx = channel->efx;
+
+	EFX_LOG(efx, "channel %d event queue in special buffers %d-%d\n",
+		channel->channel, channel->eventq.index,
+		channel->eventq.index + channel->eventq.entries - 1);
+
+	/* Pin event queue buffer */
+	efx_init_special_buffer(efx, &channel->eventq);
+
+	/* Fill event queue with all ones (i.e. empty events) */
+	memset(channel->eventq.addr, 0xff, channel->eventq.len);
+
+	/* Push event queue to card */
+	EFX_POPULATE_OWORD_3(evq_ptr,
+			     FRF_AZ_EVQ_EN, 1,
+			     FRF_AZ_EVQ_SIZE, __ffs(channel->eventq.entries),
+			     FRF_AZ_EVQ_BUF_BASE_ID, channel->eventq.index);
+	efx_writeo_table(efx, &evq_ptr, efx->type->evq_ptr_tbl_base,
+			 channel->channel);
+
+	efx->type->push_irq_moderation(channel);
+}
+
+void efx_nic_fini_eventq(struct efx_channel *channel)
+{
+	efx_oword_t eventq_ptr;
+	struct efx_nic *efx = channel->efx;
+
+	/* Remove event queue from card */
+	EFX_ZERO_OWORD(eventq_ptr);
+	efx_writeo_table(efx, &eventq_ptr, efx->type->evq_ptr_tbl_base,
+			 channel->channel);
+
+	/* Unpin event queue */
+	efx_fini_special_buffer(efx, &channel->eventq);
+}
+
+/* Free buffers backing event queue */
+void efx_nic_remove_eventq(struct efx_channel *channel)
+{
+	efx_free_special_buffer(channel->efx, &channel->eventq);
+}
+
+
+/* Generates a test event on the event queue.  A subsequent call to
+ * process_eventq() should pick up the event and place the value of
+ * "magic" into channel->eventq_magic;
+ */
+void efx_nic_generate_test_event(struct efx_channel *channel, unsigned int magic)
+{
+	efx_qword_t test_event;
+
+	EFX_POPULATE_QWORD_2(test_event, FSF_AZ_EV_CODE,
+			     FSE_AZ_EV_CODE_DRV_GEN_EV,
+			     FSF_AZ_DRV_GEN_EV_MAGIC, magic);
+	efx_generate_event(channel, &test_event);
+}
+
+/**************************************************************************
+ *
+ * Flush handling
+ *
+ **************************************************************************/
+
+
+static void efx_poll_flush_events(struct efx_nic *efx)
+{
+	struct efx_channel *channel = &efx->channel[0];
+	struct efx_tx_queue *tx_queue;
+	struct efx_rx_queue *rx_queue;
+	unsigned int read_ptr = channel->eventq_read_ptr;
+	unsigned int end_ptr = (read_ptr - 1) & EFX_EVQ_MASK;
+
+	do {
+		efx_qword_t *event = efx_event(channel, read_ptr);
+		int ev_code, ev_sub_code, ev_queue;
+		bool ev_failed;
+
+		if (!efx_event_present(event))
+			break;
+
+		ev_code = EFX_QWORD_FIELD(*event, FSF_AZ_EV_CODE);
+		ev_sub_code = EFX_QWORD_FIELD(*event,
+					      FSF_AZ_DRIVER_EV_SUBCODE);
+		if (ev_code == FSE_AZ_EV_CODE_DRIVER_EV &&
+		    ev_sub_code == FSE_AZ_TX_DESCQ_FLS_DONE_EV) {
+			ev_queue = EFX_QWORD_FIELD(*event,
+						   FSF_AZ_DRIVER_EV_SUBDATA);
+			if (ev_queue < EFX_TX_QUEUE_COUNT) {
+				tx_queue = efx->tx_queue + ev_queue;
+				tx_queue->flushed = FLUSH_DONE;
+			}
+		} else if (ev_code == FSE_AZ_EV_CODE_DRIVER_EV &&
+			   ev_sub_code == FSE_AZ_RX_DESCQ_FLS_DONE_EV) {
+			ev_queue = EFX_QWORD_FIELD(
+				*event, FSF_AZ_DRIVER_EV_RX_DESCQ_ID);
+			ev_failed = EFX_QWORD_FIELD(
+				*event, FSF_AZ_DRIVER_EV_RX_FLUSH_FAIL);
+			if (ev_queue < efx->n_rx_queues) {
+				rx_queue = efx->rx_queue + ev_queue;
+				rx_queue->flushed =
+					ev_failed ? FLUSH_FAILED : FLUSH_DONE;
+			}
+		}
+
+		/* We're about to destroy the queue anyway, so
+		 * it's ok to throw away every non-flush event */
+		EFX_SET_QWORD(*event);
+
+		read_ptr = (read_ptr + 1) & EFX_EVQ_MASK;
+	} while (read_ptr != end_ptr);
+
+	channel->eventq_read_ptr = read_ptr;
+}
+
+/* Handle tx and rx flushes at the same time, since they run in
+ * parallel in the hardware and there's no reason for us to
+ * serialise them */
+int efx_nic_flush_queues(struct efx_nic *efx)
+{
+	struct efx_rx_queue *rx_queue;
+	struct efx_tx_queue *tx_queue;
+	int i, tx_pending, rx_pending;
+
+	/* If necessary prepare the hardware for flushing */
+	efx->type->prepare_flush(efx);
+
+	/* Flush all tx queues in parallel */
+	efx_for_each_tx_queue(tx_queue, efx)
+		efx_flush_tx_queue(tx_queue);
+
+	/* The hardware supports four concurrent rx flushes, each of which may
+	 * need to be retried if there is an outstanding descriptor fetch */
+	for (i = 0; i < EFX_FLUSH_POLL_COUNT; ++i) {
+		rx_pending = tx_pending = 0;
+		efx_for_each_rx_queue(rx_queue, efx) {
+			if (rx_queue->flushed == FLUSH_PENDING)
+				++rx_pending;
+		}
+		efx_for_each_rx_queue(rx_queue, efx) {
+			if (rx_pending == EFX_RX_FLUSH_COUNT)
+				break;
+			if (rx_queue->flushed == FLUSH_FAILED ||
+			    rx_queue->flushed == FLUSH_NONE) {
+				efx_flush_rx_queue(rx_queue);
+				++rx_pending;
+			}
+		}
+		efx_for_each_tx_queue(tx_queue, efx) {
+			if (tx_queue->flushed != FLUSH_DONE)
+				++tx_pending;
+		}
+
+		if (rx_pending == 0 && tx_pending == 0)
+			return 0;
+
+		msleep(EFX_FLUSH_INTERVAL);
+		efx_poll_flush_events(efx);
+	}
+
+	/* Mark the queues as all flushed. We're going to return failure
+	 * leading to a reset, or fake up success anyway */
+	efx_for_each_tx_queue(tx_queue, efx) {
+		if (tx_queue->flushed != FLUSH_DONE)
+			EFX_ERR(efx, "tx queue %d flush command timed out\n",
+				tx_queue->queue);
+		tx_queue->flushed = FLUSH_DONE;
+	}
+	efx_for_each_rx_queue(rx_queue, efx) {
+		if (rx_queue->flushed != FLUSH_DONE)
+			EFX_ERR(efx, "rx queue %d flush command timed out\n",
+				rx_queue->queue);
+		rx_queue->flushed = FLUSH_DONE;
+	}
+
+	if (EFX_WORKAROUND_7803(efx))
+		return 0;
+
+	return -ETIMEDOUT;
+}
+
+/**************************************************************************
+ *
+ * Hardware interrupts
+ * The hardware interrupt handler does very little work; all the event
+ * queue processing is carried out by per-channel tasklets.
+ *
+ **************************************************************************/
+
+/* Enable/disable/generate interrupts */
+static inline void efx_nic_interrupts(struct efx_nic *efx,
+				      bool enabled, bool force)
+{
+	efx_oword_t int_en_reg_ker;
+
+	EFX_POPULATE_OWORD_2(int_en_reg_ker,
+			     FRF_AZ_KER_INT_KER, force,
+			     FRF_AZ_DRV_INT_EN_KER, enabled);
+	efx_writeo(efx, &int_en_reg_ker, FR_AZ_INT_EN_KER);
+}
+
+void efx_nic_enable_interrupts(struct efx_nic *efx)
+{
+	struct efx_channel *channel;
+
+	EFX_ZERO_OWORD(*((efx_oword_t *) efx->irq_status.addr));
+	wmb(); /* Ensure interrupt vector is clear before interrupts enabled */
+
+	/* Enable interrupts */
+	efx_nic_interrupts(efx, true, false);
+
+	/* Force processing of all the channels to get the EVQ RPTRs up to
+	   date */
+	efx_for_each_channel(channel, efx)
+		efx_schedule_channel(channel);
+}
+
+void efx_nic_disable_interrupts(struct efx_nic *efx)
+{
+	/* Disable interrupts */
+	efx_nic_interrupts(efx, false, false);
+}
+
+/* Generate a test interrupt
+ * Interrupt must already have been enabled, otherwise nasty things
+ * may happen.
+ */
+void efx_nic_generate_interrupt(struct efx_nic *efx)
+{
+	efx_nic_interrupts(efx, true, true);
+}
+
+/* Process a fatal interrupt
+ * Disable bus mastering ASAP and schedule a reset
+ */
+irqreturn_t efx_nic_fatal_interrupt(struct efx_nic *efx)
+{
+	struct falcon_nic_data *nic_data = efx->nic_data;
+	efx_oword_t *int_ker = efx->irq_status.addr;
+	efx_oword_t fatal_intr;
+	int error, mem_perr;
+
+	efx_reado(efx, &fatal_intr, FR_AZ_FATAL_INTR_KER);
+	error = EFX_OWORD_FIELD(fatal_intr, FRF_AZ_FATAL_INTR);
+
+	EFX_ERR(efx, "SYSTEM ERROR " EFX_OWORD_FMT " status "
+		EFX_OWORD_FMT ": %s\n", EFX_OWORD_VAL(*int_ker),
+		EFX_OWORD_VAL(fatal_intr),
+		error ? "disabling bus mastering" : "no recognised error");
+	if (error == 0)
+		goto out;
+
+	/* If this is a memory parity error dump which blocks are offending */
+	mem_perr = EFX_OWORD_FIELD(fatal_intr, FRF_AZ_MEM_PERR_INT_KER);
+	if (mem_perr) {
+		efx_oword_t reg;
+		efx_reado(efx, &reg, FR_AZ_MEM_STAT);
+		EFX_ERR(efx, "SYSTEM ERROR: memory parity error "
+			EFX_OWORD_FMT "\n", EFX_OWORD_VAL(reg));
+	}
+
+	/* Disable both devices */
+	pci_clear_master(efx->pci_dev);
+	if (efx_nic_is_dual_func(efx))
+		pci_clear_master(nic_data->pci_dev2);
+	efx_nic_disable_interrupts(efx);
+
+	/* Count errors and reset or disable the NIC accordingly */
+	if (efx->int_error_count == 0 ||
+	    time_after(jiffies, efx->int_error_expire)) {
+		efx->int_error_count = 0;
+		efx->int_error_expire =
+			jiffies + EFX_INT_ERROR_EXPIRE * HZ;
+	}
+	if (++efx->int_error_count < EFX_MAX_INT_ERRORS) {
+		EFX_ERR(efx, "SYSTEM ERROR - reset scheduled\n");
+		efx_schedule_reset(efx, RESET_TYPE_INT_ERROR);
+	} else {
+		EFX_ERR(efx, "SYSTEM ERROR - max number of errors seen."
+			"NIC will be disabled\n");
+		efx_schedule_reset(efx, RESET_TYPE_DISABLE);
+	}
+out:
+	return IRQ_HANDLED;
+}
+
+/* Handle a legacy interrupt
+ * Acknowledges the interrupt and schedule event queue processing.
+ */
+static irqreturn_t efx_legacy_interrupt(int irq, void *dev_id)
+{
+	struct efx_nic *efx = dev_id;
+	efx_oword_t *int_ker = efx->irq_status.addr;
+	irqreturn_t result = IRQ_NONE;
+	struct efx_channel *channel;
+	efx_dword_t reg;
+	u32 queues;
+	int syserr;
+
+	/* Read the ISR which also ACKs the interrupts */
+	efx_readd(efx, &reg, FR_BZ_INT_ISR0);
+	queues = EFX_EXTRACT_DWORD(reg, 0, 31);
+
+	/* Check to see if we have a serious error condition */
+	syserr = EFX_OWORD_FIELD(*int_ker, FSF_AZ_NET_IVEC_FATAL_INT);
+	if (unlikely(syserr))
+		return efx_nic_fatal_interrupt(efx);
+
+	/* Schedule processing of any interrupting queues */
+	efx_for_each_channel(channel, efx) {
+		if ((queues & 1) ||
+		    efx_event_present(
+			    efx_event(channel, channel->eventq_read_ptr))) {
+				efx_schedule_channel(channel);
+			result = IRQ_HANDLED;
+		}
+		queues >>= 1;
+	}
+
+	if (result == IRQ_HANDLED) {
+		efx->last_irq_cpu = raw_smp_processor_id();
+		EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_DWORD_FMT "\n",
+			  irq, raw_smp_processor_id(), EFX_DWORD_VAL(reg));
+	}
+
+	return result;
+}
+
+/* Handle an MSI interrupt
+ *
+ * Handle an MSI hardware interrupt.  This routine schedules event
+ * queue processing.  No interrupt acknowledgement cycle is necessary.
+ * Also, we never need to check that the interrupt is for us, since
+ * MSI interrupts cannot be shared.
+ */
+static irqreturn_t efx_msi_interrupt(int irq, void *dev_id)
+{
+	struct efx_channel *channel = dev_id;
+	struct efx_nic *efx = channel->efx;
+	efx_oword_t *int_ker = efx->irq_status.addr;
+	int syserr;
+
+	efx->last_irq_cpu = raw_smp_processor_id();
+	EFX_TRACE(efx, "IRQ %d on CPU %d status " EFX_OWORD_FMT "\n",
+		  irq, raw_smp_processor_id(), EFX_OWORD_VAL(*int_ker));
+
+	/* Check to see if we have a serious error condition */
+	syserr = EFX_OWORD_FIELD(*int_ker, FSF_AZ_NET_IVEC_FATAL_INT);
+	if (unlikely(syserr))
+		return efx_nic_fatal_interrupt(efx);
+
+	/* Schedule processing of the channel */
+	efx_schedule_channel(channel);
+
+	return IRQ_HANDLED;
+}
+
+
+/* Setup RSS indirection table.
+ * This maps from the hash value of the packet to RXQ
+ */
+static void efx_setup_rss_indir_table(struct efx_nic *efx)
+{
+	int i = 0;
+	unsigned long offset;
+	efx_dword_t dword;
+
+	if (efx_nic_rev(efx) < EFX_REV_FALCON_B0)
+		return;
+
+	for (offset = FR_BZ_RX_INDIRECTION_TBL;
+	     offset < FR_BZ_RX_INDIRECTION_TBL + 0x800;
+	     offset += 0x10) {
+		EFX_POPULATE_DWORD_1(dword, FRF_BZ_IT_QUEUE,
+				     i % efx->n_rx_queues);
+		efx_writed(efx, &dword, offset);
+		i++;
+	}
+}
+
+/* Hook interrupt handler(s)
+ * Try MSI and then legacy interrupts.
+ */
+int efx_nic_init_interrupt(struct efx_nic *efx)
+{
+	struct efx_channel *channel;
+	int rc;
+
+	if (!EFX_INT_MODE_USE_MSI(efx)) {
+		irq_handler_t handler;
+		if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0)
+			handler = efx_legacy_interrupt;
+		else
+			handler = falcon_legacy_interrupt_a1;
+
+		rc = request_irq(efx->legacy_irq, handler, IRQF_SHARED,
+				 efx->name, efx);
+		if (rc) {
+			EFX_ERR(efx, "failed to hook legacy IRQ %d\n",
+				efx->pci_dev->irq);
+			goto fail1;
+		}
+		return 0;
+	}
+
+	/* Hook MSI or MSI-X interrupt */
+	efx_for_each_channel(channel, efx) {
+		rc = request_irq(channel->irq, efx_msi_interrupt,
+				 IRQF_PROBE_SHARED, /* Not shared */
+				 channel->name, channel);
+		if (rc) {
+			EFX_ERR(efx, "failed to hook IRQ %d\n", channel->irq);
+			goto fail2;
+		}
+	}
+
+	return 0;
+
+ fail2:
+	efx_for_each_channel(channel, efx)
+		free_irq(channel->irq, channel);
+ fail1:
+	return rc;
+}
+
+void efx_nic_fini_interrupt(struct efx_nic *efx)
+{
+	struct efx_channel *channel;
+	efx_oword_t reg;
+
+	/* Disable MSI/MSI-X interrupts */
+	efx_for_each_channel(channel, efx) {
+		if (channel->irq)
+			free_irq(channel->irq, channel);
+	}
+
+	/* ACK legacy interrupt */
+	if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0)
+		efx_reado(efx, &reg, FR_BZ_INT_ISR0);
+	else
+		falcon_irq_ack_a1(efx);
+
+	/* Disable legacy interrupt */
+	if (efx->legacy_irq)
+		free_irq(efx->legacy_irq, efx);
+}
+
+u32 efx_nic_fpga_ver(struct efx_nic *efx)
+{
+	efx_oword_t altera_build;
+	efx_reado(efx, &altera_build, FR_AZ_ALTERA_BUILD);
+	return EFX_OWORD_FIELD(altera_build, FRF_AZ_ALTERA_BUILD_VER);
+}
+
+void efx_nic_init_common(struct efx_nic *efx)
+{
+	efx_oword_t temp;
+
+	/* Set positions of descriptor caches in SRAM. */
+	EFX_POPULATE_OWORD_1(temp, FRF_AZ_SRM_TX_DC_BASE_ADR,
+			     efx->type->tx_dc_base / 8);
+	efx_writeo(efx, &temp, FR_AZ_SRM_TX_DC_CFG);
+	EFX_POPULATE_OWORD_1(temp, FRF_AZ_SRM_RX_DC_BASE_ADR,
+			     efx->type->rx_dc_base / 8);
+	efx_writeo(efx, &temp, FR_AZ_SRM_RX_DC_CFG);
+
+	/* Set TX descriptor cache size. */
+	BUILD_BUG_ON(TX_DC_ENTRIES != (8 << TX_DC_ENTRIES_ORDER));
+	EFX_POPULATE_OWORD_1(temp, FRF_AZ_TX_DC_SIZE, TX_DC_ENTRIES_ORDER);
+	efx_writeo(efx, &temp, FR_AZ_TX_DC_CFG);
+
+	/* Set RX descriptor cache size.  Set low watermark to size-8, as
+	 * this allows most efficient prefetching.
+	 */
+	BUILD_BUG_ON(RX_DC_ENTRIES != (8 << RX_DC_ENTRIES_ORDER));
+	EFX_POPULATE_OWORD_1(temp, FRF_AZ_RX_DC_SIZE, RX_DC_ENTRIES_ORDER);
+	efx_writeo(efx, &temp, FR_AZ_RX_DC_CFG);
+	EFX_POPULATE_OWORD_1(temp, FRF_AZ_RX_DC_PF_LWM, RX_DC_ENTRIES - 8);
+	efx_writeo(efx, &temp, FR_AZ_RX_DC_PF_WM);
+
+	/* Program INT_KER address */
+	EFX_POPULATE_OWORD_2(temp,
+			     FRF_AZ_NORM_INT_VEC_DIS_KER,
+			     EFX_INT_MODE_USE_MSI(efx),
+			     FRF_AZ_INT_ADR_KER, efx->irq_status.dma_addr);
+	efx_writeo(efx, &temp, FR_AZ_INT_ADR_KER);
+
+	/* Enable all the genuinely fatal interrupts.  (They are still
+	 * masked by the overall interrupt mask, controlled by
+	 * falcon_interrupts()).
+	 *
+	 * Note: All other fatal interrupts are enabled
+	 */
+	EFX_POPULATE_OWORD_3(temp,
+			     FRF_AZ_ILL_ADR_INT_KER_EN, 1,
+			     FRF_AZ_RBUF_OWN_INT_KER_EN, 1,
+			     FRF_AZ_TBUF_OWN_INT_KER_EN, 1);
+	EFX_INVERT_OWORD(temp);
+	efx_writeo(efx, &temp, FR_AZ_FATAL_INTR_KER);
+
+	efx_setup_rss_indir_table(efx);
+
+	/* Disable the ugly timer-based TX DMA backoff and allow TX DMA to be
+	 * controlled by the RX FIFO fill level. Set arbitration to one pkt/Q.
+	 */
+	efx_reado(efx, &temp, FR_AZ_TX_RESERVED);
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_RX_SPACER, 0xfe);
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_RX_SPACER_EN, 1);
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_ONE_PKT_PER_Q, 1);
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_PUSH_EN, 0);
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_DIS_NON_IP_EV, 1);
+	/* Enable SW_EV to inherit in char driver - assume harmless here */
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_SOFT_EVT_EN, 1);
+	/* Prefetch threshold 2 => fetch when descriptor cache half empty */
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_PREF_THRESHOLD, 2);
+	/* Squash TX of packets of 16 bytes or less */
+	if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0)
+		EFX_SET_OWORD_FIELD(temp, FRF_BZ_TX_FLUSH_MIN_LEN_EN, 1);
+	efx_writeo(efx, &temp, FR_AZ_TX_RESERVED);
+}
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 08/12] sfc: Add firmware protocol definitions (MCDI)
From: Ben Hutchings @ 2009-11-30  1:15 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

From: Steve Hodgson <shodgson@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/mcdi_pcol.h | 1578 +++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 1578 insertions(+), 0 deletions(-)
 create mode 100644 drivers/net/sfc/mcdi_pcol.h

diff --git a/drivers/net/sfc/mcdi_pcol.h b/drivers/net/sfc/mcdi_pcol.h
new file mode 100644
index 0000000..2a85360
--- /dev/null
+++ b/drivers/net/sfc/mcdi_pcol.h
@@ -0,0 +1,1578 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2009 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+
+#ifndef MCDI_PCOL_H
+#define MCDI_PCOL_H
+
+/* Values to be written into FMCR_CZ_RESET_STATE_REG to control boot. */
+/* Power-on reset state */
+#define MC_FW_STATE_POR (1)
+/* If this is set in MC_RESET_STATE_REG then it should be
+ * possible to jump into IMEM without loading code from flash. */
+#define MC_FW_WARM_BOOT_OK (2)
+/* The MC main image has started to boot. */
+#define MC_FW_STATE_BOOTING (4)
+/* The Scheduler has started. */
+#define MC_FW_STATE_SCHED (8)
+
+/* Values to be written to the per-port status dword in shared
+ * memory on reboot and assert */
+#define MC_STATUS_DWORD_REBOOT (0xb007b007)
+#define MC_STATUS_DWORD_ASSERT (0xdeaddead)
+
+/* The current version of the MCDI protocol.
+ *
+ * Note that the ROM burnt into the card only talks V0, so at the very
+ * least every driver must support version 0 and MCDI_PCOL_VERSION
+ */
+#define MCDI_PCOL_VERSION 1
+
+/**
+ * MCDI version 1
+ *
+ * Each MCDI request starts with an MCDI_HEADER, which is a 32byte
+ * structure, filled in by the client.
+ *
+ *       0       7  8     16    20     22  23  24    31
+ *      | CODE | R | LEN | SEQ | Rsvd | E | R | XFLAGS |
+ *               |                      |   |
+ *               |                      |   \--- Response
+ *               |                      \------- Error
+ *               \------------------------------ Resync (always set)
+ *
+ * The client writes it's request into MC shared memory, and rings the
+ * doorbell. Each request is completed by either by the MC writting
+ * back into shared memory, or by writting out an event.
+ *
+ * All MCDI commands support completion by shared memory response. Each
+ * request may also contain additional data (accounted for by HEADER.LEN),
+ * and some response's may also contain additional data (again, accounted
+ * for by HEADER.LEN).
+ *
+ * Some MCDI commands support completion by event, in which any associated
+ * response data is included in the event.
+ *
+ * The protocol requires one response to be delivered for every request, a
+ * request should not be sent unless the response for the previous request
+ * has been received (either by polling shared memory, or by receiving
+ * an event).
+ */
+
+/** Request/Response structure */
+#define MCDI_HEADER_OFST 0
+#define MCDI_HEADER_CODE_LBN 0
+#define MCDI_HEADER_CODE_WIDTH 7
+#define MCDI_HEADER_RESYNC_LBN 7
+#define MCDI_HEADER_RESYNC_WIDTH 1
+#define MCDI_HEADER_DATALEN_LBN 8
+#define MCDI_HEADER_DATALEN_WIDTH 8
+#define MCDI_HEADER_SEQ_LBN 16
+#define MCDI_HEADER_RSVD_LBN 20
+#define MCDI_HEADER_RSVD_WIDTH 2
+#define MCDI_HEADER_SEQ_WIDTH 4
+#define MCDI_HEADER_ERROR_LBN 22
+#define MCDI_HEADER_ERROR_WIDTH 1
+#define MCDI_HEADER_RESPONSE_LBN 23
+#define MCDI_HEADER_RESPONSE_WIDTH 1
+#define MCDI_HEADER_XFLAGS_LBN 24
+#define MCDI_HEADER_XFLAGS_WIDTH 8
+/* Request response using event */
+#define MCDI_HEADER_XFLAGS_EVREQ 0x01
+
+/* Maximum number of payload bytes */
+#define MCDI_CTL_SDU_LEN_MAX 0xfc
+
+/* The MC can generate events for two reasons:
+ *   - To complete a shared memory request if XFLAGS_EVREQ was set
+ *   - As a notification (link state, i2c event), controlled
+ *     via MC_CMD_LOG_CTRL
+ *
+ * Both events share a common structure:
+ *
+ *  0      32     33      36    44     52     60
+ * | Data | Cont | Level | Src | Code | Rsvd |
+ *           |
+ *           \ There is another event pending in this notification
+ *
+ * If Code==CMDDONE, then the fields are further interpreted as:
+ *
+ *   - LEVEL==INFO    Command succeded
+ *   - LEVEL==ERR     Command failed
+ *
+ *    0     8         16      24     32
+ *   | Seq | Datalen | Errno | Rsvd |
+ *
+ *   These fields are taken directly out of the standard MCDI header, i.e.,
+ *   LEVEL==ERR, Datalen == 0 => Reboot
+ *
+ * Events can be squirted out of the UART (using LOG_CTRL) without a
+ * MCDI header.  An event can be distinguished from a MCDI response by
+ * examining the first byte which is 0xc0.  This corresponds to the
+ * non-existent MCDI command MC_CMD_DEBUG_LOG.
+ *
+ *      0         7        8
+ *     | command | Resync |     = 0xc0
+ *
+ * Since the event is written in big-endian byte order, this works
+ * providing bits 56-63 of the event are 0xc0.
+ *
+ *      56     60  63
+ *     | Rsvd | Code |    = 0xc0
+ *
+ * Which means for convenience the event code is 0xc for all MC
+ * generated events.
+ */
+#define FSE_AZ_EV_CODE_MCDI_EVRESPONSE 0xc
+
+#define MCDI_EVENT_DATA_LBN 0
+#define MCDI_EVENT_DATA_WIDTH 32
+#define MCDI_EVENT_CONT_LBN 32
+#define MCDI_EVENT_CONT_WIDTH 1
+#define MCDI_EVENT_LEVEL_LBN 33
+#define MCDI_EVENT_LEVEL_WIDTH 3
+#define MCDI_EVENT_LEVEL_INFO (0)
+#define MCDI_EVENT_LEVEL_WARN (1)
+#define MCDI_EVENT_LEVEL_ERR (2)
+#define MCDI_EVENT_LEVEL_FATAL (3)
+#define MCDI_EVENT_SRC_LBN 36
+#define MCDI_EVENT_SRC_WIDTH 8
+#define MCDI_EVENT_CODE_LBN 44
+#define MCDI_EVENT_CODE_WIDTH 8
+#define MCDI_EVENT_CODE_BADSSERT (1)
+#define MCDI_EVENT_CODE_PMNOTICE (2)
+#define MCDI_EVENT_CODE_CMDDONE (3)
+#define  MCDI_EVENT_CMDDONE_SEQ_LBN 0
+#define  MCDI_EVENT_CMDDONE_SEQ_WIDTH 8
+#define  MCDI_EVENT_CMDDONE_DATALEN_LBN 8
+#define  MCDI_EVENT_CMDDONE_DATALEN_WIDTH 8
+#define  MCDI_EVENT_CMDDONE_ERRNO_LBN 16
+#define  MCDI_EVENT_CMDDONE_ERRNO_WIDTH 8
+#define MCDI_EVENT_CODE_LINKCHANGE (4)
+#define  MCDI_EVENT_LINKCHANGE_LP_CAP_LBN 0
+#define  MCDI_EVENT_LINKCHANGE_LP_CAP_WIDTH 16
+#define  MCDI_EVENT_LINKCHANGE_SPEED_LBN 16
+#define  MCDI_EVENT_LINKCHANGE_SPEED_WIDTH 4
+#define  MCDI_EVENT_LINKCHANGE_SPEED_100M 1
+#define  MCDI_EVENT_LINKCHANGE_SPEED_1G 2
+#define  MCDI_EVENT_LINKCHANGE_SPEED_10G 3
+#define  MCDI_EVENT_LINKCHANGE_FCNTL_LBN 20
+#define  MCDI_EVENT_LINKCHANGE_FCNTL_WIDTH 4
+#define  MCDI_EVENT_LINKCHANGE_LINK_FLAGS_LBN 24
+#define  MCDI_EVENT_LINKCHANGE_LINK_FLAGS_WIDTH 8
+#define MCDI_EVENT_CODE_SENSOREVT (5)
+#define  MCDI_EVENT_SENSOREVT_MONITOR_LBN 0
+#define  MCDI_EVENT_SENSOREVT_MONITOR_WIDTH 8
+#define  MCDI_EVENT_SENSOREVT_STATE_LBN 8
+#define  MCDI_EVENT_SENSOREVT_STATE_WIDTH 8
+#define  MCDI_EVENT_SENSOREVT_VALUE_LBN 16
+#define  MCDI_EVENT_SENSOREVT_VALUE_WIDTH 16
+#define MCDI_EVENT_CODE_SCHEDERR (6)
+#define MCDI_EVENT_CODE_REBOOT (7)
+#define MCDI_EVENT_CODE_MAC_STATS_DMA (8)
+#define  MCDI_EVENT_MAC_STATS_DMA_GENERATION_LBN 0
+#define  MCDI_EVENT_MAC_STATS_DMA_GENERATION_WIDTH 32
+
+/* Non-existent command target */
+#define MC_CMD_ERR_ENOENT 2
+/* assert() has killed the MC */
+#define MC_CMD_ERR_EINTR 4
+/* Caller does not hold required locks */
+#define MC_CMD_ERR_EACCES 13
+/* Resource is currently unavailable (e.g. lock contention) */
+#define MC_CMD_ERR_EBUSY 16
+/* Invalid argument to target */
+#define MC_CMD_ERR_EINVAL 22
+/* Non-recursive resource is already acquired */
+#define MC_CMD_ERR_EDEADLK 35
+/* Operation not implemented */
+#define MC_CMD_ERR_ENOSYS 38
+/* Operation timed out */
+#define MC_CMD_ERR_ETIME 62
+
+#define MC_CMD_ERR_CODE_OFST 0
+
+
+/* MC_CMD_READ32: (debug, variadic out)
+ * Read multiple 32byte words from MC memory
+ */
+#define MC_CMD_READ32 0x01
+#define MC_CMD_READ32_IN_LEN 8
+#define MC_CMD_READ32_IN_ADDR_OFST 0
+#define MC_CMD_READ32_IN_NUMWORDS_OFST 4
+#define MC_CMD_READ32_OUT_LEN(_numwords) \
+	(4 * (_numwords))
+#define MC_CMD_READ32_OUT_BUFFER_OFST 0
+
+/* MC_CMD_WRITE32: (debug, variadic in)
+ * Write multiple 32byte words to MC memory
+ */
+#define MC_CMD_WRITE32 0x02
+#define MC_CMD_WRITE32_IN_LEN(_numwords) (((_numwords) * 4) + 4)
+#define MC_CMD_WRITE32_IN_ADDR_OFST 0
+#define MC_CMD_WRITE32_IN_BUFFER_OFST 4
+#define MC_CMD_WRITE32_OUT_LEN 0
+
+/* MC_CMD_COPYCODE: (debug)
+ * Copy MC code between two locations and jump
+ */
+#define MC_CMD_COPYCODE 0x03
+#define MC_CMD_COPYCODE_IN_LEN 16
+#define MC_CMD_COPYCODE_IN_SRC_ADDR_OFST 0
+#define MC_CMD_COPYCODE_IN_DEST_ADDR_OFST 4
+#define MC_CMD_COPYCODE_IN_NUMWORDS_OFST 8
+#define MC_CMD_COPYCODE_IN_JUMP_OFST 12
+/* Control should return to the caller rather than jumping */
+#define MC_CMD_COPYCODE_JUMP_NONE 1
+#define MC_CMD_COPYCODE_OUT_LEN 0
+
+/* MC_CMD_SET_FUNC: (debug)
+ * Select function for function-specific commands.
+ */
+#define MC_CMD_SET_FUNC 0x04
+#define MC_CMD_SET_FUNC_IN_LEN 4
+#define MC_CMD_SET_FUNC_IN_FUNC_OFST 0
+#define MC_CMD_SET_FUNC_OUT_LEN 0
+
+/* MC_CMD_GET_BOOT_STATUS:
+ * Get the instruction address from which the MC booted.
+ */
+#define MC_CMD_GET_BOOT_STATUS 0x05
+#define MC_CMD_GET_BOOT_STATUS_IN_LEN 0
+#define MC_CMD_GET_BOOT_STATUS_OUT_LEN 8
+#define MC_CMD_GET_BOOT_STATUS_OUT_BOOT_OFFSET_OFST 0
+#define MC_CMD_GET_BOOT_STATUS_OUT_FLAGS_OFST 4
+/* Reboot caused by watchdog */
+#define MC_CMD_GET_BOOT_STATUS_FLAGS_WATCHDOG_LBN   (0)
+#define MC_CMD_GET_BOOT_STATUS_FLAGS_WATCHDOG_WIDTH (1)
+/* MC booted from primary flash partition */
+#define MC_CMD_GET_BOOT_STATUS_FLAGS_PRIMARY_LBN    (1)
+#define MC_CMD_GET_BOOT_STATUS_FLAGS_PRIMARY_WIDTH  (1)
+/* MC booted from backup flash partition */
+#define MC_CMD_GET_BOOT_STATUS_FLAGS_BACKUP_LBN     (2)
+#define MC_CMD_GET_BOOT_STATUS_FLAGS_BACKUP_WIDTH   (1)
+
+/* MC_CMD_GET_ASSERTS: (debug, variadic out)
+ * Get (and optionally clear) the current assertion status.
+ *
+ * Only OUT.GLOBAL_FLAGS is guaranteed to exist in the completion
+ * payload. The other fields will only be present if
+ * OUT.GLOBAL_FLAGS != NO_FAILS
+ */
+#define MC_CMD_GET_ASSERTS 0x06
+#define MC_CMD_GET_ASSERTS_IN_LEN 4
+#define MC_CMD_GET_ASSERTS_IN_CLEAR_OFST 0
+#define MC_CMD_GET_ASSERTS_OUT_LEN 140
+/* Assertion status flag */
+#define MC_CMD_GET_ASSERTS_OUT_GLOBAL_FLAGS_OFST 0
+/*! No assertions have failed. */
+#define MC_CMD_GET_ASSERTS_FLAGS_NO_FAILS 1
+/*! A system-level assertion has failed. */
+#define MC_CMD_GET_ASSERTS_FLAGS_SYS_FAIL 2
+/*! A thread-level assertion has failed. */
+#define MC_CMD_GET_ASSERTS_FLAGS_THR_FAIL 3
+/*! The system was reset by the watchdog. */
+#define MC_CMD_GET_ASSERTS_FLAGS_WDOG_FIRED 4
+/* Failing PC value */
+#define MC_CMD_GET_ASSERTS_OUT_SAVED_PC_OFFS_OFST 4
+/* Saved GP regs */
+#define MC_CMD_GET_ASSERTS_OUT_GP_REGS_OFFS_OFST 8
+#define MC_CMD_GET_ASSERTS_OUT_GP_REGS_LEN 124
+/* Failing thread address */
+#define MC_CMD_GET_ASSERTS_OUT_THREAD_OFFS_OFST 132
+
+/* MC_CMD_LOG_CTRL:
+ * Determine the output stream for various events and messages
+ */
+#define MC_CMD_LOG_CTRL 0x07
+#define MC_CMD_LOG_CTRL_IN_LEN 8
+#define MC_CMD_LOG_CTRL_IN_LOG_DEST_OFST 0
+#define MC_CMD_LOG_CTRL_IN_LOG_DEST_UART (1)
+#define MC_CMD_LOG_CTRL_IN_LOG_DEST_EVQ (2)
+#define MC_CMD_LOG_CTRL_IN_LOG_DEST_EVQ_OFST 4
+#define MC_CMD_LOG_CTRL_OUT_LEN 0
+
+/* MC_CMD_GET_VERSION:
+ * Get version information about the MC firmware
+ */
+#define MC_CMD_GET_VERSION 0x08
+#define MC_CMD_GET_VERSION_IN_LEN 0
+#define MC_CMD_GET_VERSION_V0_OUT_LEN 4
+#define MC_CMD_GET_VERSION_V1_OUT_LEN 32
+#define MC_CMD_GET_VERSION_OUT_FIRMWARE_OFST 0
+/* Reserved version number to indicate "any" version. */
+#define MC_CMD_GET_VERSION_OUT_FIRMWARE_ANY 0xffffffff
+/* The version response of a boot ROM awaiting rescue */
+#define MC_CMD_GET_VERSION_OUT_FIRMWARE_BOOTROM 0xb0070000
+#define MC_CMD_GET_VERSION_V1_OUT_PCOL_OFST 4
+/* 128bit mask of functions supported by the current firmware */
+#define MC_CMD_GET_VERSION_V1_OUT_SUPPORTED_FUNCS_OFST 8
+/* The command set exported by the boot ROM (MCDI v0) */
+#define MC_CMD_GET_VERSION_V0_SUPPORTED_FUNCS {		\
+	(1 << MC_CMD_READ32)	|			\
+	(1 << MC_CMD_WRITE32)	|			\
+	(1 << MC_CMD_COPYCODE)	|			\
+	(1 << MC_CMD_GET_VERSION),			\
+	0, 0, 0 }
+#define MC_CMD_GET_VERSION_OUT_VERSION_OFST 24
+
+/* Vectors in the boot ROM */
+/* Point to the copycode entry point. */
+#define MC_BOOTROM_COPYCODE_VEC (0x7f4)
+/* Points to the recovery mode entry point. */
+#define MC_BOOTROM_NOFLASH_VEC (0x7f8)
+
+/* Test execution limits */
+#define MC_TESTEXEC_VARIANT_COUNT 16
+#define MC_TESTEXEC_RESULT_COUNT 7
+
+/* MC_CMD_SET_TESTVARS: (debug, variadic in)
+ * Write variant words for test.
+ *
+ * The user supplies a bitmap of the variants they wish to set.
+ * They must ensure that IN.LEN >= 4 + 4 * ffs(BITMAP)
+ */
+#define MC_CMD_SET_TESTVARS 0x09
+#define MC_CMD_SET_TESTVARS_IN_LEN(_numwords)	\
+  (4 + 4*(_numwords))
+#define MC_CMD_SET_TESTVARS_IN_ARGS_BITMAP_OFST 0
+/* Up to MC_TESTEXEC_VARIANT_COUNT of 32byte words start here */
+#define MC_CMD_SET_TESTVARS_IN_ARGS_BUFFER_OFST 4
+#define MC_CMD_SET_TESTVARS_OUT_LEN 0
+
+/* MC_CMD_GET_TESTRCS: (debug, variadic out)
+ * Return result words from test.
+ */
+#define MC_CMD_GET_TESTRCS 0x0a
+#define MC_CMD_GET_TESTRCS_IN_LEN 4
+#define MC_CMD_GET_TESTRCS_IN_NUMWORDS_OFST 0
+#define MC_CMD_GET_TESTRCS_OUT_LEN(_numwords) \
+	(4 * (_numwords))
+#define MC_CMD_GET_TESTRCS_OUT_BUFFER_OFST 0
+
+/* MC_CMD_RUN_TEST: (debug)
+ * Run the test exported by this firmware image
+ */
+#define MC_CMD_RUN_TEST 0x0b
+#define MC_CMD_RUN_TEST_IN_LEN 0
+#define MC_CMD_RUN_TEST_OUT_LEN 0
+
+/* MC_CMD_CSR_READ32: (debug, variadic out)
+ * Read 32bit words from the indirect memory map
+ */
+#define MC_CMD_CSR_READ32 0x0c
+#define MC_CMD_CSR_READ32_IN_LEN 12
+#define MC_CMD_CSR_READ32_IN_ADDR_OFST 0
+#define MC_CMD_CSR_READ32_IN_STEP_OFST 4
+#define MC_CMD_CSR_READ32_IN_NUMWORDS_OFST 8
+#define MC_CMD_CSR_READ32_OUT_LEN(_numwords)	\
+	(((_numwords) * 4) + 4)
+/* IN.NUMWORDS of 32bit words start here */
+#define MC_CMD_CSR_READ32_OUT_BUFFER_OFST 0
+#define MC_CMD_CSR_READ32_OUT_IREG_STATUS_OFST(_numwords)	\
+	((_numwords) * 4)
+
+/* MC_CMD_CSR_WRITE32: (debug, variadic in)
+ * Write 32bit dwords to the indirect memory map
+ */
+#define MC_CMD_CSR_WRITE32 0x0d
+#define MC_CMD_CSR_WRITE32_IN_LEN(_numwords)	\
+	(((_numwords) * 4) + 8)
+#define MC_CMD_CSR_WRITE32_IN_ADDR_OFST 0
+#define MC_CMD_CSR_WRITE32_IN_STEP_OFST 4
+/* Multiple 32bit words of data to write start here */
+#define MC_CMD_CSR_WRITE32_IN_BUFFER_OFST 8
+#define MC_CMD_CSR_WRITE32_OUT_LEN 4
+#define MC_CMD_CSR_WRITE32_OUT_STATUS_OFST 0
+
+/* MC_CMD_JTAG_WORK: (debug, fpga only)
+ * Process JTAG work buffer for RBF acceleration.
+ *
+ *  Host: bit count, (up to) 32 words of data to clock out to JTAG
+ *   (bits 1,0=TMS,TDO for first bit; bits 3,2=TMS,TDO for second bit, etc.)
+ *  MC: bit count, (up to) 32 words of data clocked in from JTAG
+ *   (bit 0=TDI for first bit, bit 1=TDI for second bit, etc.; [31:16] unused)
+ */
+#define MC_CMD_JTAG_WORK 0x0e
+
+/* MC_CMD_STACKINFO: (debug, variadic out)
+ * Get stack information
+ *
+ * Host: nothing
+ * MC: (thread ptr, stack size, free space) for each thread in system
+ */
+#define MC_CMD_STACKINFO 0x0f
+
+/* MC_CMD_MDIO_READ:
+ * MDIO register read
+ */
+#define MC_CMD_MDIO_READ 0x10
+#define MC_CMD_MDIO_READ_IN_LEN 16
+#define MC_CMD_MDIO_READ_IN_BUS_OFST 0
+#define MC_CMD_MDIO_READ_IN_PRTAD_OFST 4
+#define MC_CMD_MDIO_READ_IN_DEVAD_OFST 8
+#define MC_CMD_MDIO_READ_IN_ADDR_OFST 12
+#define MC_CMD_MDIO_READ_OUT_LEN 8
+#define MC_CMD_MDIO_READ_OUT_VALUE_OFST 0
+#define MC_CMD_MDIO_READ_OUT_STATUS_OFST 4
+
+/* MC_CMD_MDIO_WRITE:
+ * MDIO register write
+ */
+#define MC_CMD_MDIO_WRITE 0x11
+#define MC_CMD_MDIO_WRITE_IN_LEN 20
+#define MC_CMD_MDIO_WRITE_IN_BUS_OFST 0
+#define MC_CMD_MDIO_WRITE_IN_PRTAD_OFST 4
+#define MC_CMD_MDIO_WRITE_IN_DEVAD_OFST 8
+#define MC_CMD_MDIO_WRITE_IN_ADDR_OFST 12
+#define MC_CMD_MDIO_WRITE_IN_VALUE_OFST 16
+#define MC_CMD_MDIO_WRITE_OUT_LEN 4
+#define MC_CMD_MDIO_WRITE_OUT_STATUS_OFST 0
+
+/* By default all the MCDI MDIO operations perform clause45 mode.
+ * If you want to use clause22 then set DEVAD = MC_CMD_MDIO_CLAUSE22.
+ */
+#define MC_CMD_MDIO_CLAUSE22 32
+
+/* There are two MDIO buses: one for the internal PHY, and one for external
+ * devices.
+ */
+#define MC_CMD_MDIO_BUS_INTERNAL 0
+#define MC_CMD_MDIO_BUS_EXTERNAL 1
+
+/* The MDIO commands return the raw status bits from the MDIO block.  A "good"
+ * transaction should have the DONE bit set and all other bits clear.
+ */
+#define MC_CMD_MDIO_STATUS_GOOD 0x08
+
+
+/* MC_CMD_DBI_WRITE: (debug)
+ * Write DBI register(s)
+ *
+ * Host: address, byte-enables (and VF selection, and cs2 flag),
+ *       value [,address ...]
+ * MC: nothing
+ */
+#define MC_CMD_DBI_WRITE 0x12
+#define MC_CMD_DBI_WRITE_IN_LEN(_numwords)		\
+	(12 * (_numwords))
+#define MC_CMD_DBI_WRITE_IN_ADDRESS_OFST(_word)		\
+	(((_word) * 12) + 0)
+#define MC_CMD_DBI_WRITE_IN_BYTE_MASK_OFST(_word)	\
+	(((_word) * 12) + 4)
+#define MC_CMD_DBI_WRITE_IN_VALUE_OFST(_word)		\
+	(((_word) * 12) + 8)
+#define MC_CMD_DBI_WRITE_OUT_LEN 0
+
+/* MC_CMD_DBI_READ: (debug)
+ * Read DBI register(s)
+ *
+ * Host: address, [,address ...]
+ * MC: value [,value ...]
+ * (note: this does not support reading from VFs, but is retained for backwards
+ * compatibility; see MC_CMD_DBI_READX below)
+ */
+#define MC_CMD_DBI_READ 0x13
+#define MC_CMD_DBI_READ_IN_LEN(_numwords)		\
+	(4 * (_numwords))
+#define MC_CMD_DBI_READ_OUT_LEN(_numwords)		\
+	(4 * (_numwords))
+
+/* MC_CMD_PORT_READ32: (debug)
+ * Read a 32-bit register from the indirect port register map.
+ *
+ * The port to access is implied by the Shared memory channel used.
+ */
+#define MC_CMD_PORT_READ32 0x14
+#define MC_CMD_PORT_READ32_IN_LEN 4
+#define MC_CMD_PORT_READ32_IN_ADDR_OFST 0
+#define MC_CMD_PORT_READ32_OUT_LEN 8
+#define MC_CMD_PORT_READ32_OUT_VALUE_OFST 0
+#define MC_CMD_PORT_READ32_OUT_STATUS_OFST 4
+
+/* MC_CMD_PORT_WRITE32: (debug)
+ * Write a 32-bit register to the indirect port register map.
+ *
+ * The port to access is implied by the Shared memory channel used.
+ */
+#define MC_CMD_PORT_WRITE32 0x15
+#define MC_CMD_PORT_WRITE32_IN_LEN 8
+#define MC_CMD_PORT_WRITE32_IN_ADDR_OFST 0
+#define MC_CMD_PORT_WRITE32_IN_VALUE_OFST 4
+#define MC_CMD_PORT_WRITE32_OUT_LEN 4
+#define MC_CMD_PORT_WRITE32_OUT_STATUS_OFST 0
+
+/* MC_CMD_PORT_READ128: (debug)
+ * Read a 128-bit register from indirect port register map
+ *
+ * The port to access is implied by the Shared memory channel used.
+ */
+#define MC_CMD_PORT_READ128 0x16
+#define MC_CMD_PORT_READ128_IN_LEN 4
+#define MC_CMD_PORT_READ128_IN_ADDR_OFST 0
+#define MC_CMD_PORT_READ128_OUT_LEN 20
+#define MC_CMD_PORT_READ128_OUT_VALUE_OFST 0
+#define MC_CMD_PORT_READ128_OUT_STATUS_OFST 16
+
+/* MC_CMD_PORT_WRITE128: (debug)
+ * Write a 128-bit register to indirect port register map.
+ *
+ * The port to access is implied by the Shared memory channel used.
+ */
+#define MC_CMD_PORT_WRITE128 0x17
+#define MC_CMD_PORT_WRITE128_IN_LEN 20
+#define MC_CMD_PORT_WRITE128_IN_ADDR_OFST 0
+#define MC_CMD_PORT_WRITE128_IN_VALUE_OFST 4
+#define MC_CMD_PORT_WRITE128_OUT_LEN 4
+#define MC_CMD_PORT_WRITE128_OUT_STATUS_OFST 0
+
+/* MC_CMD_GET_BOARD_CFG:
+ * Returns the MC firmware configuration structure
+ *
+ * The FW_SUBTYPE_LIST contains a 16-bit value for each of the 12 types of
+ * NVRAM area.  The values are defined in the firmware/mc/platform/<xxx>.c file
+ * for a specific board type, but otherwise have no meaning to the MC; they
+ * are used by the driver to manage selection of appropriate firmware updates.
+ */
+#define MC_CMD_GET_BOARD_CFG 0x18
+#define MC_CMD_GET_BOARD_CFG_IN_LEN 0
+#define MC_CMD_GET_BOARD_CFG_OUT_LEN 96
+#define MC_CMD_GET_BOARD_CFG_OUT_BOARD_TYPE_OFST 0
+#define MC_CMD_GET_BOARD_CFG_OUT_BOARD_NAME_OFST 4
+#define MC_CMD_GET_BOARD_CFG_OUT_BOARD_NAME_LEN 32
+#define MC_CMD_GET_BOARD_CFG_OUT_CAPABILITIES_PORT0_OFST 36
+#define MC_CMD_GET_BOARD_CFG_OUT_CAPABILITIES_PORT1_OFST 40
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_ADDR_BASE_PORT0_OFST 44
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_ADDR_BASE_PORT0_LEN 6
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_ADDR_BASE_PORT1_OFST 50
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_ADDR_BASE_PORT1_LEN 6
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_COUNT_PORT0_OFST 56
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_COUNT_PORT1_OFST 60
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_STRIDE_PORT0_OFST 64
+#define MC_CMD_GET_BOARD_CFG_OUT_MAC_STRIDE_PORT1_OFST 68
+#define MC_CMD_GET_BOARD_CFG_OUT_FW_SUBTYPE_LIST_OFST 72
+#define MC_CMD_GET_BOARD_CFG_OUT_FW_SUBTYPE_LIST_LEN 24
+
+/* MC_CMD_DBI_READX: (debug)
+ * Read DBI register(s) -- extended functionality
+ *
+ * Host: vf selection, address, [,vf selection ...]
+ * MC: value [,value ...]
+ */
+#define MC_CMD_DBI_READX 0x19
+#define MC_CMD_DBI_READX_IN_LEN(_numwords)	\
+  (8*(_numwords))
+#define MC_CMD_DBI_READX_OUT_LEN(_numwords)	\
+  (4*(_numwords))
+
+/* MC_CMD_SET_RAND_SEED:
+ * Set the 16byte seed for the MC psuedo-random generator
+ */
+#define MC_CMD_SET_RAND_SEED 0x1a
+#define MC_CMD_SET_RAND_SEED_IN_LEN 16
+#define MC_CMD_SET_RAND_SEED_IN_SEED_OFST 0
+#define MC_CMD_SET_RAND_SEED_OUT_LEN 0
+
+/* MC_CMD_LTSSM_HIST: (debug)
+ * Retrieve the history of the LTSSM, if the build supports it.
+ *
+ * Host: nothing
+ * MC: variable number of LTSSM values, as bytes
+ * The history is read-to-clear.
+ */
+#define MC_CMD_LTSSM_HIST 0x1b
+
+/* MC_CMD_DRV_ATTACH:
+ * Inform MCPU that this port is managed on the host (i.e. driver active)
+ */
+#define MC_CMD_DRV_ATTACH 0x1c
+#define MC_CMD_DRV_ATTACH_IN_LEN 8
+#define MC_CMD_DRV_ATTACH_IN_NEW_STATE_OFST 0
+#define MC_CMD_DRV_ATTACH_IN_UPDATE_OFST 4
+#define MC_CMD_DRV_ATTACH_OUT_LEN 4
+#define MC_CMD_DRV_ATTACH_OUT_OLD_STATE_OFST 0
+
+/* MC_CMD_NCSI_PROD: (debug)
+ * Trigger an NC-SI event (and possibly an AEN in response)
+ */
+#define MC_CMD_NCSI_PROD 0x1d
+#define MC_CMD_NCSI_PROD_IN_LEN 4
+#define MC_CMD_NCSI_PROD_IN_EVENTS_OFST 0
+#define MC_CMD_NCSI_PROD_LINKCHANGE_LBN 0
+#define MC_CMD_NCSI_PROD_LINKCHANGE_WIDTH 1
+#define MC_CMD_NCSI_PROD_RESET_LBN 1
+#define MC_CMD_NCSI_PROD_RESET_WIDTH 1
+#define MC_CMD_NCSI_PROD_DRVATTACH_LBN 2
+#define MC_CMD_NCSI_PROD_DRVATTACH_WIDTH 1
+#define MC_CMD_NCSI_PROD_OUT_LEN 0
+
+/* Enumeration */
+#define MC_CMD_NCSI_PROD_LINKCHANGE 0
+#define MC_CMD_NCSI_PROD_RESET 1
+#define MC_CMD_NCSI_PROD_DRVATTACH 2
+
+/* MC_CMD_DEVEL: (debug)
+ * Reserved for development
+ */
+#define MC_CMD_DEVEL 0x1e
+
+/* MC_CMD_SHMUART: (debug)
+ * Route UART output to circular buffer in shared memory instead.
+ */
+#define MC_CMD_SHMUART 0x1f
+#define MC_CMD_SHMUART_IN_FLAG_OFST 0
+#define MC_CMD_SHMUART_IN_LEN 4
+#define MC_CMD_SHMUART_OUT_LEN 0
+
+/* MC_CMD_PORT_RESET:
+ * Generic per-port reset. There is no equivalent for per-board reset.
+ *
+ * Locks required: None
+ * Return code: 0, ETIME
+ */
+#define MC_CMD_PORT_RESET 0x20
+#define MC_CMD_PORT_RESET_IN_LEN 0
+#define MC_CMD_PORT_RESET_OUT_LEN 0
+
+/* MC_CMD_RESOURCE_LOCK:
+ * Generic resource lock/unlock interface.
+ *
+ * Locks required: None
+ * Return code: 0,
+ *              EBUSY (if trylock is contended by other port),
+ *              EDEADLK (if trylock is already acquired by this port)
+ *              EINVAL (if unlock doesn't own the lock)
+ */
+#define MC_CMD_RESOURCE_LOCK 0x21
+#define MC_CMD_RESOURCE_LOCK_IN_LEN 8
+#define MC_CMD_RESOURCE_LOCK_IN_ACTION_OFST 0
+#define MC_CMD_RESOURCE_LOCK_ACTION_TRYLOCK 1
+#define MC_CMD_RESOURCE_LOCK_ACTION_UNLOCK 0
+#define MC_CMD_RESOURCE_LOCK_IN_RESOURCE_OFST 4
+#define MC_CMD_RESOURCE_LOCK_I2C 2
+#define MC_CMD_RESOURCE_LOCK_PHY 3
+#define MC_CMD_RESOURCE_LOCK_OUT_LEN 0
+
+/* MC_CMD_SPI_COMMAND: (variadic in, variadic out)
+ * Read/Write to/from the SPI device.
+ *
+ * Locks required: SPI_LOCK
+ * Return code: 0, ETIME, EINVAL, EACCES (if SPI_LOCK is not held)
+ */
+#define MC_CMD_SPI_COMMAND 0x22
+#define MC_CMD_SPI_COMMAND_IN_LEN(_write_bytes)	(12 + (_write_bytes))
+#define MC_CMD_SPI_COMMAND_IN_ARGS_OFST 0
+#define MC_CMD_SPI_COMMAND_IN_ARGS_ADDRESS_OFST 0
+#define MC_CMD_SPI_COMMAND_IN_ARGS_READ_BYTES_OFST 4
+#define MC_CMD_SPI_COMMAND_IN_ARGS_CHIP_SELECT_OFST 8
+/* Data to write here */
+#define MC_CMD_SPI_COMMAND_IN_WRITE_BUFFER_OFST 12
+#define MC_CMD_SPI_COMMAND_OUT_LEN(_read_bytes) (_read_bytes)
+/* Data read here */
+#define MC_CMD_SPI_COMMAND_OUT_READ_BUFFER_OFST 0
+
+/* MC_CMD_I2C_READ_WRITE: (variadic in, variadic out)
+ * Read/Write to/from the I2C bus.
+ *
+ * Locks required: I2C_LOCK
+ * Return code: 0, ETIME, EINVAL, EACCES (if I2C_LOCK is not held)
+ */
+#define MC_CMD_I2C_RW 0x23
+#define MC_CMD_I2C_RW_IN_LEN(_write_bytes) (8 + (_write_bytes))
+#define MC_CMD_I2C_RW_IN_ARGS_OFST 0
+#define MC_CMD_I2C_RW_IN_ARGS_ADDR_OFST 0
+#define MC_CMD_I2C_RW_IN_ARGS_READ_BYTES_OFST 4
+/* Data to write here */
+#define MC_CMD_I2C_RW_IN_WRITE_BUFFER_OFSET 8
+#define MC_CMD_I2C_RW_OUT_LEN(_read_bytes) (_read_bytes)
+/* Data read here */
+#define MC_CMD_I2C_RW_OUT_READ_BUFFER_OFST 0
+
+/* Generic phy capability bitmask */
+#define MC_CMD_PHY_CAP_10HDX_LBN 1
+#define MC_CMD_PHY_CAP_10HDX_WIDTH 1
+#define MC_CMD_PHY_CAP_10FDX_LBN 2
+#define MC_CMD_PHY_CAP_10FDX_WIDTH 1
+#define MC_CMD_PHY_CAP_100HDX_LBN 3
+#define MC_CMD_PHY_CAP_100HDX_WIDTH 1
+#define MC_CMD_PHY_CAP_100FDX_LBN 4
+#define MC_CMD_PHY_CAP_100FDX_WIDTH 1
+#define MC_CMD_PHY_CAP_1000HDX_LBN 5
+#define MC_CMD_PHY_CAP_1000HDX_WIDTH 1
+#define MC_CMD_PHY_CAP_1000FDX_LBN 6
+#define MC_CMD_PHY_CAP_1000FDX_WIDTH 1
+#define MC_CMD_PHY_CAP_10000FDX_LBN 7
+#define MC_CMD_PHY_CAP_10000FDX_WIDTH 1
+#define MC_CMD_PHY_CAP_PAUSE_LBN 8
+#define MC_CMD_PHY_CAP_PAUSE_WIDTH 1
+#define MC_CMD_PHY_CAP_ASYM_LBN 9
+#define MC_CMD_PHY_CAP_ASYM_WIDTH 1
+#define MC_CMD_PHY_CAP_AN_LBN 10
+#define MC_CMD_PHY_CAP_AN_WIDTH 1
+
+/* Generic loopback enumeration */
+#define MC_CMD_LOOPBACK_NONE 0
+#define MC_CMD_LOOPBACK_DATA 1
+#define MC_CMD_LOOPBACK_GMAC 2
+#define MC_CMD_LOOPBACK_XGMII 3
+#define MC_CMD_LOOPBACK_XGXS 4
+#define MC_CMD_LOOPBACK_XAUI 5
+#define MC_CMD_LOOPBACK_GMII 6
+#define MC_CMD_LOOPBACK_SGMII 7
+#define MC_CMD_LOOPBACK_XGBR 8
+#define MC_CMD_LOOPBACK_XFI 9
+#define MC_CMD_LOOPBACK_XAUI_FAR 10
+#define MC_CMD_LOOPBACK_GMII_FAR 11
+#define MC_CMD_LOOPBACK_SGMII_FAR 12
+#define MC_CMD_LOOPBACK_XFI_FAR 13
+#define MC_CMD_LOOPBACK_GPHY 14
+#define MC_CMD_LOOPBACK_PHYXS 15
+#define MC_CMD_LOOPBACK_PCS 16
+#define MC_CMD_LOOPBACK_PMAPMD 17
+#define MC_CMD_LOOPBACK_XPORT 18
+#define MC_CMD_LOOPBACK_XGMII_WS 19
+#define MC_CMD_LOOPBACK_XAUI_WS 20
+#define MC_CMD_LOOPBACK_XAUI_WS_FAR 21
+#define MC_CMD_LOOPBACK_XAUI_WS_NEAR 22
+#define MC_CMD_LOOPBACK_GMII_WS 23
+#define MC_CMD_LOOPBACK_XFI_WS 24
+#define MC_CMD_LOOPBACK_XFI_WS_FAR 25
+#define MC_CMD_LOOPBACK_PHYXS_WS 26
+
+/* Generic PHY statistics enumeration */
+#define MC_CMD_OUI 0
+#define MC_CMD_PMA_PMD_LINK_UP 1
+#define MC_CMD_PMA_PMD_RX_FAULT 2
+#define MC_CMD_PMA_PMD_TX_FAULT 3
+#define MC_CMD_PMA_PMD_SIGNAL 4
+#define MC_CMD_PMA_PMD_SNR_A 5
+#define MC_CMD_PMA_PMD_SNR_B 6
+#define MC_CMD_PMA_PMD_SNR_C 7
+#define MC_CMD_PMA_PMD_SNR_D 8
+#define MC_CMD_PCS_LINK_UP 9
+#define MC_CMD_PCS_RX_FAULT 10
+#define MC_CMD_PCS_TX_FAULT 11
+#define MC_CMD_PCS_BER 12
+#define MC_CMD_PCS_BLOCK_ERRORS 13
+#define MC_CMD_PHYXS_LINK_UP 14
+#define MC_CMD_PHYXS_RX_FAULT 15
+#define MC_CMD_PHYXS_TX_FAULT 16
+#define MC_CMD_PHYXS_ALIGN 17
+#define MC_CMD_PHYXS_SYNC 18
+#define MC_CMD_AN_LINK_UP 19
+#define MC_CMD_AN_COMPLETE 20
+#define MC_CMD_AN_10GBT_STATUS 21
+#define MC_CMD_CL22_LINK_UP 22
+#define MC_CMD_PHY_NSTATS 23
+
+/* MC_CMD_GET_PHY_CFG:
+ * Report PHY configuration.  This guarantees to succeed even if the PHY is in
+ * a "zombie" state.
+ *
+ * Locks required: None
+ * Return code: 0
+ */
+#define MC_CMD_GET_PHY_CFG 0x24
+
+#define MC_CMD_GET_PHY_CFG_IN_LEN 0
+#define MC_CMD_GET_PHY_CFG_OUT_LEN 72
+
+#define MC_CMD_GET_PHY_CFG_OUT_FLAGS_OFST 0
+#define MC_CMD_GET_PHY_CFG_PRESENT_LBN 0
+#define MC_CMD_GET_PHY_CFG_PRESENT_WIDTH 1
+#define MC_CMD_GET_PHY_CFG_SHORTBIST_LBN 1
+#define MC_CMD_GET_PHY_CFG_SHORTBIST_WIDTH 1
+#define MC_CMD_GET_PHY_CFG_LONGBIST_LBN 2
+#define MC_CMD_GET_PHY_CFG_LONGBIST_WIDTH 1
+#define MC_CMD_GET_PHY_CFG_LOWPOWER_LBN 3
+#define MC_CMD_GET_PHY_CFG_LOWPOWER_WIDTH 1
+#define MC_CMD_GET_PHY_CFG_POWEROFF_LBN 4
+#define MC_CMD_GET_PHY_CFG_POWEROFF_WIDTH 1
+#define MC_CMD_GET_PHY_CFG_TXDIS_LBN 5
+#define MC_CMD_GET_PHY_CFG_TXDIS_WIDTH 1
+#define MC_CMD_GET_PHY_CFG_OUT_TYPE_OFST 4
+/* Bitmask of supported capabilities */
+#define MC_CMD_GET_PHY_CFG_OUT_SUPPORTED_CAP_OFST 8
+#define MC_CMD_GET_PHY_CFG_OUT_CHANNEL_OFST 12
+#define MC_CMD_GET_PHY_CFG_OUT_PRT_OFST 16
+/* PHY statistics bitmap */
+#define MC_CMD_GET_PHY_CFG_OUT_STATS_MASK_OFST 20
+/* PHY type/name string */
+#define MC_CMD_GET_PHY_CFG_OUT_NAME_OFST 24
+#define MC_CMD_GET_PHY_CFG_OUT_NAME_LEN 20
+#define MC_CMD_GET_PHY_CFG_OUT_MEDIA_TYPE_OFST 44
+#define MC_CMD_MEDIA_XAUI 1
+#define MC_CMD_MEDIA_CX4 2
+#define MC_CMD_MEDIA_KX4 3
+#define MC_CMD_MEDIA_XFP 4
+#define MC_CMD_MEDIA_SFP_PLUS 5
+#define MC_CMD_MEDIA_BASE_T 6
+/* MDIO "MMDS" supported */
+#define MC_CMD_GET_PHY_CFG_OUT_MMD_MASK_OFST 48
+/* Native clause 22 */
+#define MC_CMD_MMD_CLAUSE22  0
+#define MC_CMD_MMD_CLAUSE45_PMAPMD 1
+#define MC_CMD_MMD_CLAUSE45_WIS 2
+#define MC_CMD_MMD_CLAUSE45_PCS 3
+#define MC_CMD_MMD_CLAUSE45_PHYXS 4
+#define MC_CMD_MMD_CLAUSE45_DTEXS 5
+#define MC_CMD_MMD_CLAUSE45_TC 6
+#define MC_CMD_MMD_CLAUSE45_AN 7
+/* Clause22 proxied over clause45 by PHY */
+#define MC_CMD_MMD_CLAUSE45_C22EXT 29
+#define MC_CMD_MMD_CLAUSE45_VEND1 30
+#define MC_CMD_MMD_CLAUSE45_VEND2 31
+/* PHY stepping version */
+#define MC_CMD_GET_PHY_CFG_OUT_REVISION_OFST 52
+#define MC_CMD_GET_PHY_CFG_OUT_REVISION_LEN 20
+
+/* MC_CMD_START_PHY_BIST:
+ * Start a BIST test on the PHY.
+ *
+ * Locks required: PHY_LOCK if doing a  PHY BIST
+ * Return code: 0, EINVAL, EACCES (if PHY_LOCK is not held)
+ */
+#define MC_CMD_START_BIST 0x25
+#define MC_CMD_START_BIST_IN_LEN 4
+#define MC_CMD_START_BIST_TYPE_OFST 0
+
+/* Run the PHY's short BIST */
+#define MC_CMD_PHY_BIST_SHORT  1
+/* Run the PHY's long BIST */
+#define MC_CMD_PHY_BIST_LONG   2
+/* Run BIST on the currently selected BPX Serdes (XAUI or XFI) */
+#define MC_CMD_BPX_SERDES_BIST 3
+
+/* MC_CMD_POLL_PHY_BIST: (variadic output)
+ * Poll for BIST completion
+ *
+ * Returns a single status code, and a binary blob of phy-specific
+ * bist output. If the driver can't succesfully parse the BIST output,
+ * it should still respect the Pass/Fail in OUT.RESULT.
+ *
+ * Locks required: PHY_LOCK  if doing a  PHY BIST
+ * Return code: 0, EACCES (if PHY_LOCK is not held)
+ */
+#define MC_CMD_POLL_BIST 0x26
+#define MC_CMD_POLL_BIST_IN_LEN 0
+#define MC_CMD_POLL_BIST_OUT_LEN UNKNOWN
+#define MC_CMD_POLL_BIST_OUT_RESULT_OFST 0
+#define MC_CMD_POLL_BIST_RUNNING 1
+#define MC_CMD_POLL_BIST_PASSED 2
+#define MC_CMD_POLL_BIST_FAILED 3
+#define MC_CMD_POLL_BIST_TIMEOUT 4
+#define MC_CMD_POLL_BIST_OUT_PRIVATE_OFST 4
+
+/* MC_CMD_PHY_SPI: (variadic in, variadic out)
+ * Read/Write/Erase the PHY SPI device
+ *
+ * Locks required: PHY_LOCK
+ * Return code: 0, ETIME, EINVAL, EACCES (if PHY_LOCK is not held)
+ */
+#define MC_CMD_PHY_SPI 0x27
+#define MC_CMD_PHY_SPI_IN_LEN(_write_bytes) (12 + (_write_bytes))
+#define MC_CMD_PHY_SPI_IN_ARGS_OFST 0
+#define MC_CMD_PHY_SPI_IN_ARGS_ADDR_OFST 0
+#define MC_CMD_PHY_SPI_IN_ARGS_READ_BYTES_OFST 4
+#define MC_CMD_PHY_SPI_IN_ARGS_ERASE_ALL_OFST 8
+/* Data to write here */
+#define MC_CMD_PHY_SPI_IN_WRITE_BUFFER_OFSET 12
+#define MC_CMD_PHY_SPI_OUT_LEN(_read_bytes) (_read_bytes)
+/* Data read here */
+#define MC_CMD_PHY_SPI_OUT_READ_BUFFER_OFST 0
+
+
+/* MC_CMD_GET_LOOPBACK_MODES:
+ * Returns a bitmask of loopback modes evailable at each speed.
+ *
+ * Locks required: None
+ * Return code: 0
+ */
+#define MC_CMD_GET_LOOPBACK_MODES 0x28
+#define MC_CMD_GET_LOOPBACK_MODES_IN_LEN 0
+#define MC_CMD_GET_LOOPBACK_MODES_OUT_LEN 32
+#define MC_CMD_GET_LOOPBACK_MODES_100M_OFST 0
+#define MC_CMD_GET_LOOPBACK_MODES_1G_OFST 8
+#define MC_CMD_GET_LOOPBACK_MODES_10G_OFST 16
+#define MC_CMD_GET_LOOPBACK_MODES_SUGGESTED_OFST 24
+
+/* Flow control enumeration */
+#define MC_CMD_FCNTL_OFF 0
+#define MC_CMD_FCNTL_RESPOND 1
+#define MC_CMD_FCNTL_BIDIR 2
+/* Auto - Use what the link has autonegotiated
+ *      - The driver should modify the advertised capabilities via SET_LINK.CAP
+ *        to control the negotiated flow control mode.
+ *      - Can only be set if the PHY supports PAUSE+ASYM capabilities
+ *      - Never returned by GET_LINK as the value programmed into the MAC
+ */
+#define MC_CMD_FCNTL_AUTO 3
+
+/* Generic mac fault bitmask */
+#define MC_CMD_MAC_FAULT_XGMII_LOCAL_LBN 0
+#define MC_CMD_MAC_FAULT_XGMII_LOCAL_WIDTH 1
+#define MC_CMD_MAC_FAULT_XGMII_REMOTE_LBN 1
+#define MC_CMD_MAC_FAULT_XGMII_REMOTE_WIDTH 1
+#define MC_CMD_MAC_FAULT_SGMII_REMOTE_LBN 2
+#define MC_CMD_MAC_FAULT_SGMII_REMOTE_WIDTH 1
+
+/* MC_CMD_GET_LINK:
+ * Read the unified MAC/PHY link state
+ *
+ * Locks required: None
+ * Return code: 0, ETIME
+ */
+#define MC_CMD_GET_LINK 0x29
+#define MC_CMD_GET_LINK_IN_LEN 0
+#define MC_CMD_GET_LINK_OUT_LEN 28
+/* near-side and link-partner advertised capabilities */
+#define MC_CMD_GET_LINK_OUT_CAP_OFST 0
+#define MC_CMD_GET_LINK_OUT_LP_CAP_OFST 4
+/* Autonegotiated speed in mbit/s. The link may still be down
+ * even if this reads non-zero */
+#define MC_CMD_GET_LINK_OUT_LINK_SPEED_OFST 8
+#define MC_CMD_GET_LINK_OUT_LOOPBACK_MODE_OFST 12
+#define MC_CMD_GET_LINK_OUT_FLAGS_OFST 16
+/* Whether we have overall link up */
+#define MC_CMD_GET_LINK_LINK_UP_LBN 0
+#define MC_CMD_GET_LINK_LINK_UP_WIDTH 1
+#define MC_CMD_GET_LINK_FULL_DUPLEX_LBN 1
+#define MC_CMD_GET_LINK_FULL_DUPLEX_WIDTH 1
+/* Whether we have link at the layers provided by the BPX */
+#define MC_CMD_GET_LINK_BPX_LINK_LBN 2
+#define MC_CMD_GET_LINK_BPX_LINK_WIDTH 1
+/* Whether the PHY has external link */
+#define MC_CMD_GET_LINK_PHY_LINK_LBN 3
+#define MC_CMD_GET_LINK_PHY_LINK_WIDTH 1
+#define MC_CMD_GET_LINK_OUT_FCNTL_OFST 20
+#define MC_CMD_GET_LINK_OUT_MAC_FAULT_OFST 24
+
+/* MC_CMD_SET_LINK:
+ * Write the unified MAC/PHY link configuration
+ *
+ * A loopback speed of "0" is supported, and means
+ * (choose any available speed)
+ *
+ * Locks required: None
+ * Return code: 0, EINVAL, ETIME
+ */
+#define MC_CMD_SET_LINK 0x2a
+#define MC_CMD_SET_LINK_IN_LEN 16
+#define MC_CMD_SET_LINK_IN_CAP_OFST 0
+#define MC_CMD_SET_LINK_IN_FLAGS_OFST 4
+#define MC_CMD_SET_LINK_LOWPOWER_LBN 0
+#define MC_CMD_SET_LINK_LOWPOWER_WIDTH 1
+#define MC_CMD_SET_LINK_POWEROFF_LBN 1
+#define MC_CMD_SET_LINK_POWEROFF_WIDTH 1
+#define MC_CMD_SET_LINK_TXDIS_LBN 2
+#define MC_CMD_SET_LINK_TXDIS_WIDTH 1
+#define MC_CMD_SET_LINK_IN_LOOPBACK_MODE_OFST 8
+#define MC_CMD_SET_LINK_IN_LOOPBACK_SPEED_OFST 12
+#define MC_CMD_SET_LINK_OUT_LEN 0
+
+/* MC_CMD_SET_ID_LED:
+ * Set indentification LED state
+ *
+ * Locks required: None
+ * Return code: 0, EINVAL
+ */
+#define MC_CMD_SET_ID_LED 0x2b
+#define MC_CMD_SET_ID_LED_IN_LEN 4
+#define MC_CMD_SET_ID_LED_IN_STATE_OFST 0
+#define  MC_CMD_LED_OFF 0
+#define  MC_CMD_LED_ON 1
+#define  MC_CMD_LED_DEFAULT 2
+#define MC_CMD_SET_ID_LED_OUT_LEN 0
+
+/* MC_CMD_SET_MAC:
+ * Set MAC configuration
+ *
+ * The MTU is the MTU programmed directly into the XMAC/GMAC
+ * (inclusive of EtherII, VLAN, bug16011 padding)
+ *
+ * Locks required: None
+ * Return code: 0, EINVAL
+ */
+#define MC_CMD_SET_MAC 0x2c
+#define MC_CMD_SET_MAC_IN_LEN 24
+#define MC_CMD_SET_MAC_IN_MTU_OFST 0
+#define MC_CMD_SET_MAC_IN_DRAIN_OFST 4
+#define MC_CMD_SET_MAC_IN_ADDR_OFST 8
+#define MC_CMD_SET_MAC_IN_REJECT_OFST 16
+#define MC_CMD_SET_MAC_IN_REJECT_UNCST_LBN 0
+#define MC_CMD_SET_MAC_IN_REJECT_UNCST_WIDTH 1
+#define MC_CMD_SET_MAC_IN_REJECT_BRDCST_LBN 1
+#define MC_CMD_SET_MAC_IN_REJECT_BRDCST_WIDTH 1
+#define MC_CMD_SET_MAC_IN_FCNTL_OFST 20
+#define MC_CMD_SET_MAC_OUT_LEN 0
+
+/* MC_CMD_PHY_STATS:
+ * Get generic PHY statistics
+ *
+ * This call returns the statistics for a generic PHY, by direct DMA
+ * into host memory, in a sparse array (indexed by the enumerate).
+ * Each value is represented by a 32bit number.
+ *
+ * Locks required: None
+ * Returns: 0, ETIME
+ * Response methods: shared memory, event
+ */
+#define MC_CMD_PHY_STATS 0x2d
+#define MC_CMD_PHY_STATS_IN_LEN 8
+#define MC_CMD_PHY_STATS_IN_DMA_ADDR_LO_OFST 0
+#define MC_CMD_PHY_STATS_IN_DMA_ADDR_HI_OFST 4
+#define MC_CMD_PHY_STATS_OUT_LEN 0
+
+/* Unified MAC statistics enumeration */
+#define MC_CMD_MAC_GENERATION_START 0
+#define MC_CMD_MAC_TX_PKTS 1
+#define MC_CMD_MAC_TX_PAUSE_PKTS 2
+#define MC_CMD_MAC_TX_CONTROL_PKTS 3
+#define MC_CMD_MAC_TX_UNICAST_PKTS 4
+#define MC_CMD_MAC_TX_MULTICAST_PKTS 5
+#define MC_CMD_MAC_TX_BROADCAST_PKTS 6
+#define MC_CMD_MAC_TX_BYTES 7
+#define MC_CMD_MAC_TX_BAD_BYTES 8
+#define MC_CMD_MAC_TX_LT64_PKTS 9
+#define MC_CMD_MAC_TX_64_PKTS 10
+#define MC_CMD_MAC_TX_65_TO_127_PKTS 11
+#define MC_CMD_MAC_TX_128_TO_255_PKTS 12
+#define MC_CMD_MAC_TX_256_TO_511_PKTS 13
+#define MC_CMD_MAC_TX_512_TO_1023_PKTS 14
+#define MC_CMD_MAC_TX_1024_TO_15XX_PKTS 15
+#define MC_CMD_MAC_TX_15XX_TO_JUMBO_PKTS 16
+#define MC_CMD_MAC_TX_GTJUMBO_PKTS 17
+#define MC_CMD_MAC_TX_BAD_FCS_PKTS 18
+#define MC_CMD_MAC_TX_SINGLE_COLLISION_PKTS 19
+#define MC_CMD_MAC_TX_MULTIPLE_COLLISION_PKTS 20
+#define MC_CMD_MAC_TX_EXCESSIVE_COLLISION_PKTS 21
+#define MC_CMD_MAC_TX_LATE_COLLISION_PKTS 22
+#define MC_CMD_MAC_TX_DEFERRED_PKTS 23
+#define MC_CMD_MAC_TX_EXCESSIVE_DEFERRED_PKTS 24
+#define MC_CMD_MAC_TX_NON_TCPUDP_PKTS 25
+#define MC_CMD_MAC_TX_MAC_SRC_ERR_PKTS 26
+#define MC_CMD_MAC_TX_IP_SRC_ERR_PKTS 27
+#define MC_CMD_MAC_RX_PKTS 28
+#define MC_CMD_MAC_RX_PAUSE_PKTS 29
+#define MC_CMD_MAC_RX_GOOD_PKTS 30
+#define MC_CMD_MAC_RX_CONTROL_PKTS 31
+#define MC_CMD_MAC_RX_UNICAST_PKTS 32
+#define MC_CMD_MAC_RX_MULTICAST_PKTS 33
+#define MC_CMD_MAC_RX_BROADCAST_PKTS 34
+#define MC_CMD_MAC_RX_BYTES 35
+#define MC_CMD_MAC_RX_BAD_BYTES 36
+#define MC_CMD_MAC_RX_64_PKTS 37
+#define MC_CMD_MAC_RX_65_TO_127_PKTS 38
+#define MC_CMD_MAC_RX_128_TO_255_PKTS 39
+#define MC_CMD_MAC_RX_256_TO_511_PKTS 40
+#define MC_CMD_MAC_RX_512_TO_1023_PKTS 41
+#define MC_CMD_MAC_RX_1024_TO_15XX_PKTS 42
+#define MC_CMD_MAC_RX_15XX_TO_JUMBO_PKTS 43
+#define MC_CMD_MAC_RX_GTJUMBO_PKTS 44
+#define MC_CMD_MAC_RX_UNDERSIZE_PKTS 45
+#define MC_CMD_MAC_RX_BAD_FCS_PKTS 46
+#define MC_CMD_MAC_RX_OVERFLOW_PKTS 47
+#define MC_CMD_MAC_RX_FALSE_CARRIER_PKTS 48
+#define MC_CMD_MAC_RX_SYMBOL_ERROR_PKTS 49
+#define MC_CMD_MAC_RX_ALIGN_ERROR_PKTS 50
+#define MC_CMD_MAC_RX_LENGTH_ERROR_PKTS 51
+#define MC_CMD_MAC_RX_INTERNAL_ERROR_PKTS 52
+#define MC_CMD_MAC_RX_JABBER_PKTS 53
+#define MC_CMD_MAC_RX_NODESC_DROPS 54
+#define MC_CMD_MAC_RX_LANES01_CHAR_ERR 55
+#define MC_CMD_MAC_RX_LANES23_CHAR_ERR 56
+#define MC_CMD_MAC_RX_LANES01_DISP_ERR 57
+#define MC_CMD_MAC_RX_LANES23_DISP_ERR 58
+#define MC_CMD_MAC_RX_MATCH_FAULT 59
+/* Insert new members here. */
+#define MC_CMD_MAC_GENERATION_END 60
+#define MC_CMD_MAC_NSTATS (MC_CMD_MAC_GENERATION_END+1)
+
+/* MC_CMD_MAC_STATS:
+ * Get unified GMAC/XMAC statistics
+ *
+ * This call returns unified statistics maintained by the MC as it
+ * switches between the GMAC and XMAC. The MC will write out all
+ * supported stats.  The driver should zero initialise the buffer to
+ * guarantee consistent results.
+ *
+ * Locks required: None
+ * Returns: 0
+ * Response methods: shared memory, event
+ */
+#define MC_CMD_MAC_STATS 0x2e
+#define MC_CMD_MAC_STATS_IN_LEN 16
+#define MC_CMD_MAC_STATS_IN_DMA_ADDR_LO_OFST 0
+#define MC_CMD_MAC_STATS_IN_DMA_ADDR_HI_OFST 4
+#define MC_CMD_MAC_STATS_IN_CMD_OFST 8
+#define MC_CMD_MAC_STATS_CMD_DMA_LBN 0
+#define MC_CMD_MAC_STATS_CMD_DMA_WIDTH 1
+#define MC_CMD_MAC_STATS_CMD_CLEAR_LBN 1
+#define MC_CMD_MAC_STATS_CMD_CLEAR_WIDTH 1
+#define MC_CMD_MAC_STATS_CMD_PERIODIC_CHANGE_LBN 2
+#define MC_CMD_MAC_STATS_CMD_PERIODIC_CHANGE_WIDTH 1
+/* Fields only relevent when PERIODIC_CHANGE is set */
+#define MC_CMD_MAC_STATS_CMD_PERIODIC_ENABLE_LBN 3
+#define MC_CMD_MAC_STATS_CMD_PERIODIC_ENABLE_WIDTH 1
+#define MC_CMD_MAC_STATS_CMD_PERIODIC_CLEAR_LBN 4
+#define MC_CMD_MAC_STATS_CMD_PERIODIC_CLEAR_WIDTH 1
+#define MC_CMD_MAC_STATS_CMD_PERIOD_MS_LBN 16
+#define MC_CMD_MAC_STATS_CMD_PERIOD_MS_WIDTH 16
+#define MC_CMD_MAC_STATS_IN_DMA_LEN_OFST 12
+
+#define MC_CMD_MAC_STATS_OUT_LEN 0
+
+/* Callisto flags */
+#define MC_CMD_SFT9001_ROBUST_LBN 0
+#define MC_CMD_SFT9001_ROBUST_WIDTH 1
+#define MC_CMD_SFT9001_SHORT_REACH_LBN 1
+#define MC_CMD_SFT9001_SHORT_REACH_WIDTH 1
+
+/* MC_CMD_SFT9001_GET:
+ * Read current callisto specific setting
+ *
+ * Locks required: None
+ * Returns: 0, ETIME
+ */
+#define MC_CMD_SFT9001_GET 0x30
+#define MC_CMD_SFT9001_GET_IN_LEN 0
+#define MC_CMD_SFT9001_GET_OUT_LEN 4
+#define MC_CMD_SFT9001_GET_OUT_FLAGS_OFST 0
+
+/* MC_CMD_SFT9001_SET:
+ * Write current callisto specific setting
+ *
+ * Locks required: None
+ * Returns: 0, ETIME, EINVAL
+ */
+#define MC_CMD_SFT9001_SET 0x31
+#define MC_CMD_SFT9001_SET_IN_LEN 4
+#define MC_CMD_SFT9001_SET_IN_FLAGS_OFST 0
+#define MC_CMD_SFT9001_SET_OUT_LEN 0
+
+
+/* MC_CMD_WOL_FILTER_SET:
+ * Set a WoL filter
+ *
+ * Locks required: None
+ * Returns: 0, EBUSY, EINVAL, ENOSYS
+ */
+#define MC_CMD_WOL_FILTER_SET 0x32
+#define MC_CMD_WOL_FILTER_SET_IN_LEN 192 /* 190 rounded up to a word */
+#define MC_CMD_WOL_FILTER_SET_IN_FILTER_MODE_OFST 0
+#define MC_CMD_WOL_FILTER_SET_IN_WOL_TYPE_OFST 4
+
+/* There is a union at offset 8, following defines overlap due to
+ * this */
+#define MC_CMD_WOL_FILTER_SET_IN_DATA_OFST 8
+
+#define MC_CMD_WOL_FILTER_SET_IN_MAGIC_MAC_OFST		\
+	MC_CMD_WOL_FILTER_SET_IN_DATA_OFST
+
+#define MC_CMD_WOL_FILTER_SET_IN_IPV4_SYN_SRC_IP_OFST   \
+	MC_CMD_WOL_FILTER_SET_IN_DATA_OFST
+#define MC_CMD_WOL_FILTER_SET_IN_IPV4_SYN_DST_IP_OFST   \
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 4)
+#define MC_CMD_WOL_FILTER_SET_IN_IPV4_SYN_SRC_PORT_OFST \
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 8)
+#define MC_CMD_WOL_FILTER_SET_IN_IPV4_SYN_DST_PORT_OFST \
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 10)
+
+#define MC_CMD_WOL_FILTER_SET_IN_IPV6_SYN_SRC_IP_OFST   \
+	MC_CMD_WOL_FILTER_SET_IN_DATA_OFST
+#define MC_CMD_WOL_FILTER_SET_IN_IPV6_SYN_DST_IP_OFST   \
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 16)
+#define MC_CMD_WOL_FILTER_SET_IN_IPV6_SYN_SRC_PORT_OFST \
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 32)
+#define MC_CMD_WOL_FILTER_SET_IN_IPV6_SYN_DST_PORT_OFST \
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 34)
+
+#define MC_CMD_WOL_FILTER_SET_IN_BITMAP_MASK_OFST	\
+	MC_CMD_WOL_FILTER_SET_IN_DATA_OFST
+#define MC_CMD_WOL_FILTER_SET_IN_BITMAP_OFST		\
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 48)
+#define MC_CMD_WOL_FILTER_SET_IN_BITMAP_LEN_OFST	\
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 176)
+#define MC_CMD_WOL_FILTER_SET_IN_BITMAP_LAYER3_OFST	\
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 177)
+#define MC_CMD_WOL_FILTER_SET_IN_BITMAP_LAYER4_OFST	\
+	(MC_CMD_WOL_FILTER_SET_IN_DATA_OFST + 178)
+
+#define MC_CMD_WOL_FILTER_SET_OUT_LEN 4
+#define MC_CMD_WOL_FILTER_SET_OUT_FILTER_ID_OFST 0
+
+/* WOL Filter types enumeration */
+#define MC_CMD_WOL_TYPE_MAGIC      0x0
+			 /* unused 0x1 */
+#define MC_CMD_WOL_TYPE_WIN_MAGIC  0x2
+#define MC_CMD_WOL_TYPE_IPV4_SYN   0x3
+#define MC_CMD_WOL_TYPE_IPV6_SYN   0x4
+#define MC_CMD_WOL_TYPE_BITMAP     0x5
+#define MC_CMD_WOL_TYPE_MAX        0x6
+
+#define MC_CMD_FILTER_MODE_SIMPLE     0x0
+#define MC_CMD_FILTER_MODE_STRUCTURED 0xffffffff
+
+/* MC_CMD_WOL_FILTER_REMOVE:
+ * Remove a WoL filter
+ *
+ * Locks required: None
+ * Returns: 0, EINVAL, ENOSYS
+ */
+#define MC_CMD_WOL_FILTER_REMOVE 0x33
+#define MC_CMD_WOL_FILTER_REMOVE_IN_LEN 4
+#define MC_CMD_WOL_FILTER_REMOVE_IN_FILTER_ID_OFST 0
+#define MC_CMD_WOL_FILTER_REMOVE_OUT_LEN 0
+
+
+/* MC_CMD_WOL_FILTER_RESET:
+ * Reset (i.e. remove all) WoL filters
+ *
+ * Locks required: None
+ * Returns: 0, ENOSYS
+ */
+#define MC_CMD_WOL_FILTER_RESET 0x34
+#define MC_CMD_WOL_FILTER_RESET_IN_LEN 0
+#define MC_CMD_WOL_FILTER_RESET_OUT_LEN 0
+
+/* MC_CMD_SET_MCAST_HASH:
+ * Set the MCASH hash value without otherwise
+ * reconfiguring the MAC
+ */
+#define MC_CMD_SET_MCAST_HASH 0x35
+#define MC_CMD_SET_MCAST_HASH_IN_LEN 32
+#define MC_CMD_SET_MCAST_HASH_IN_HASH0_OFST 0
+#define MC_CMD_SET_MCAST_HASH_IN_HASH1_OFST 16
+#define MC_CMD_SET_MCAST_HASH_OUT_LEN 0
+
+/* MC_CMD_NVRAM_TYPES:
+ * Return bitfield indicating available types of virtual NVRAM partitions
+ *
+ * Locks required: none
+ * Returns: 0
+ */
+#define MC_CMD_NVRAM_TYPES 0x36
+#define MC_CMD_NVRAM_TYPES_IN_LEN 0
+#define MC_CMD_NVRAM_TYPES_OUT_LEN 4
+#define MC_CMD_NVRAM_TYPES_OUT_TYPES_OFST 0
+
+/* Supported NVRAM types */
+#define MC_CMD_NVRAM_TYPE_DISABLED_CALLISTO 0
+#define MC_CMD_NVRAM_TYPE_MC_FW 1
+#define MC_CMD_NVRAM_TYPE_MC_FW_BACKUP 2
+#define MC_CMD_NVRAM_TYPE_STATIC_CFG_PORT0 3
+#define MC_CMD_NVRAM_TYPE_STATIC_CFG_PORT1 4
+#define MC_CMD_NVRAM_TYPE_DYNAMIC_CFG_PORT0 5
+#define MC_CMD_NVRAM_TYPE_DYNAMIC_CFG_PORT1 6
+#define MC_CMD_NVRAM_TYPE_EXP_ROM 7
+#define MC_CMD_NVRAM_TYPE_EXP_ROM_CFG_PORT0 8
+#define MC_CMD_NVRAM_TYPE_EXP_ROM_CFG_PORT1 9
+#define MC_CMD_NVRAM_TYPE_PHY_PORT0 10
+#define MC_CMD_NVRAM_TYPE_PHY_PORT1 11
+#define MC_CMD_NVRAM_TYPE_LOG 12
+
+/* MC_CMD_NVRAM_INFO:
+ * Read info about a virtual NVRAM partition
+ *
+ * Locks required: none
+ * Returns: 0, EINVAL (bad type)
+ */
+#define MC_CMD_NVRAM_INFO 0x37
+#define MC_CMD_NVRAM_INFO_IN_LEN 4
+#define MC_CMD_NVRAM_INFO_IN_TYPE_OFST 0
+#define MC_CMD_NVRAM_INFO_OUT_LEN 24
+#define MC_CMD_NVRAM_INFO_OUT_TYPE_OFST 0
+#define MC_CMD_NVRAM_INFO_OUT_SIZE_OFST 4
+#define MC_CMD_NVRAM_INFO_OUT_ERASESIZE_OFST 8
+#define MC_CMD_NVRAM_INFO_OUT_FLAGS_OFST 12
+#define   MC_CMD_NVRAM_PROTECTED_LBN 0
+#define   MC_CMD_NVRAM_PROTECTED_WIDTH 1
+#define MC_CMD_NVRAM_INFO_OUT_PHYSDEV_OFST 16
+#define MC_CMD_NVRAM_INFO_OUT_PHYSADDR_OFST 20
+
+/* MC_CMD_NVRAM_UPDATE_START:
+ * Start a group of update operations on a virtual NVRAM partition
+ *
+ * Locks required: PHY_LOCK if type==*PHY*
+ * Returns: 0, EINVAL (bad type), EACCES (if PHY_LOCK required and not held)
+ */
+#define MC_CMD_NVRAM_UPDATE_START 0x38
+#define MC_CMD_NVRAM_UPDATE_START_IN_LEN 4
+#define MC_CMD_NVRAM_UPDATE_START_IN_TYPE_OFST 0
+#define MC_CMD_NVRAM_UPDATE_START_OUT_LEN 0
+
+/* MC_CMD_NVRAM_READ:
+ * Read data from a virtual NVRAM partition
+ *
+ * Locks required: PHY_LOCK if type==*PHY*
+ * Returns: 0, EINVAL (bad type/offset/length), EACCES (if PHY_LOCK required and not held)
+ */
+#define MC_CMD_NVRAM_READ 0x39
+#define MC_CMD_NVRAM_READ_IN_LEN 12
+#define MC_CMD_NVRAM_READ_IN_TYPE_OFST 0
+#define MC_CMD_NVRAM_READ_IN_OFFSET_OFST 4
+#define MC_CMD_NVRAM_READ_IN_LENGTH_OFST 8
+#define MC_CMD_NVRAM_READ_OUT_LEN(_read_bytes) (_read_bytes)
+#define MC_CMD_NVRAM_READ_OUT_READ_BUFFER_OFST 0
+
+/* MC_CMD_NVRAM_WRITE:
+ * Write data to a virtual NVRAM partition
+ *
+ * Locks required: PHY_LOCK if type==*PHY*
+ * Returns: 0, EINVAL (bad type/offset/length), EACCES (if PHY_LOCK required and not held)
+ */
+#define MC_CMD_NVRAM_WRITE 0x3a
+#define MC_CMD_NVRAM_WRITE_IN_TYPE_OFST 0
+#define MC_CMD_NVRAM_WRITE_IN_OFFSET_OFST 4
+#define MC_CMD_NVRAM_WRITE_IN_LENGTH_OFST 8
+#define MC_CMD_NVRAM_WRITE_IN_WRITE_BUFFER_OFST 12
+#define MC_CMD_NVRAM_WRITE_IN_LEN(_write_bytes) (12 + _write_bytes)
+#define MC_CMD_NVRAM_WRITE_OUT_LEN 0
+
+/* MC_CMD_NVRAM_ERASE:
+ * Erase sector(s) from a virtual NVRAM partition
+ *
+ * Locks required: PHY_LOCK if type==*PHY*
+ * Returns: 0, EINVAL (bad type/offset/length), EACCES (if PHY_LOCK required and not held)
+ */
+#define MC_CMD_NVRAM_ERASE 0x3b
+#define MC_CMD_NVRAM_ERASE_IN_LEN 12
+#define MC_CMD_NVRAM_ERASE_IN_TYPE_OFST 0
+#define MC_CMD_NVRAM_ERASE_IN_OFFSET_OFST 4
+#define MC_CMD_NVRAM_ERASE_IN_LENGTH_OFST 8
+#define MC_CMD_NVRAM_ERASE_OUT_LEN 0
+
+/* MC_CMD_NVRAM_UPDATE_FINISH:
+ * Finish a group of update operations on a virtual NVRAM partition
+ *
+ * Locks required: PHY_LOCK if type==*PHY*
+ * Returns: 0, EINVAL (bad type/offset/length), EACCES (if PHY_LOCK required and not held)
+ */
+#define MC_CMD_NVRAM_UPDATE_FINISH 0x3c
+#define MC_CMD_NVRAM_UPDATE_FINISH_IN_LEN 4
+#define MC_CMD_NVRAM_UPDATE_FINISH_IN_TYPE_OFST 0
+#define MC_CMD_NVRAM_UPDATE_FINISH_OUT_LEN 0
+
+/* MC_CMD_REBOOT:
+ * Reboot the MC. The AFTER_ASSERTION flag is intended to be used
+ * when the driver notices an assertion failure, to allow two ports to
+ * both recover (semi-)gracefully.
+ *
+ * Locks required: NONE
+ * Returns: Nothing. You get back a response with ERR=1, DATALEN=0
+ */
+#define MC_CMD_REBOOT 0x3d
+#define MC_CMD_REBOOT_IN_LEN 4
+#define MC_CMD_REBOOT_IN_FLAGS_OFST 0
+#define MC_CMD_REBOOT_FLAGS_AFTER_ASSERTION 1
+#define MC_CMD_REBOOT_OUT_LEN 0
+
+/* MC_CMD_SCHEDINFO:
+ * Request scheduler info. from the MC.
+ *
+ * Locks required: NONE
+ * Returns: An array of (timeslice,maximum overrun), one for each thread,
+ * in ascending order of thread address.s
+ */
+#define MC_CMD_SCHEDINFO 0x3e
+#define MC_CMD_SCHEDINFO_IN_LEN 0
+
+
+/* MC_CMD_SET_REBOOT_MODE: (debug)
+ * Set the mode for the next MC reboot.
+ *
+ * Locks required: NONE
+ *
+ * Sets the reboot mode to the specified value.  Returns the old mode.
+ */
+#define MC_CMD_REBOOT_MODE 0x3f
+#define MC_CMD_REBOOT_MODE_IN_LEN 4
+#define MC_CMD_REBOOT_MODE_IN_VALUE_OFST 0
+#define MC_CMD_REBOOT_MODE_OUT_LEN 4
+#define MC_CMD_REBOOT_MODE_OUT_VALUE_OFST 0
+#define   MC_CMD_REBOOT_MODE_NORMAL 0
+#define   MC_CMD_REBOOT_MODE_SNAPPER 3
+
+/* MC_CMD_DEBUG_LOG:
+ * Null request/response command (debug)
+ * - sequence number is always zero
+ * - only supported on the UART interface
+ * (the same set of bytes is delivered as an
+ * event over PCI)
+ */
+#define MC_CMD_DEBUG_LOG 0x40
+#define MC_CMD_DEBUG_LOG_IN_LEN 0
+#define MC_CMD_DEBUG_LOG_OUT_LEN 0
+
+/* Generic sensor enumeration. Note that a dual port NIC
+ * will EITHER expose PHY_COMMON_TEMP OR PHY0_TEMP and
+ * PHY1_TEMP depending on whether there is a single sensor
+ * in the vicinity of the two port, or one per port.
+ */
+#define MC_CMD_SENSOR_CONTROLLER_TEMP 0		/* degC */
+#define MC_CMD_SENSOR_PHY_COMMON_TEMP 1		/* degC */
+#define MC_CMD_SENSOR_CONTROLLER_COOLING 2	/* bool */
+#define MC_CMD_SENSOR_PHY0_TEMP 3		/* degC */
+#define MC_CMD_SENSOR_PHY0_COOLING 4		/* bool */
+#define MC_CMD_SENSOR_PHY1_TEMP 5		/* degC */
+#define MC_CMD_SENSOR_PHY1_COOLING 6		/* bool */
+#define MC_CMD_SENSOR_IN_1V0 7			/* mV */
+#define MC_CMD_SENSOR_IN_1V2 8			/* mV */
+#define MC_CMD_SENSOR_IN_1V8 9			/* mV */
+#define MC_CMD_SENSOR_IN_2V5 10			/* mV */
+#define MC_CMD_SENSOR_IN_3V3 11			/* mV */
+#define MC_CMD_SENSOR_IN_12V0 12		/* mV */
+
+
+/* Sensor state */
+#define MC_CMD_SENSOR_STATE_OK 0
+#define MC_CMD_SENSOR_STATE_WARNING 1
+#define MC_CMD_SENSOR_STATE_FATAL 2
+#define MC_CMD_SENSOR_STATE_BROKEN 3
+
+/* MC_CMD_SENSOR_INFO:
+ * Returns information about every available sensor.
+ *
+ * Each sensor has a single (16bit) value, and a corresponding state.
+ * The mapping between value and sensor is nominally determined by the
+ * MC, but in practise is implemented as zero (BROKEN), one (TEMPERATURE),
+ * or two (VOLTAGE) ranges per sensor per state.
+ *
+ * This call returns a mask (32bit) of the sensors that are supported
+ * by this platform, then an array (indexed by MC_CMD_SENSOR) of byte
+ * offsets to the per-sensor arrays. Each sensor array has four 16bit
+ * numbers, min1, max1, min2, max2.
+ *
+ * Locks required: None
+ * Returns: 0
+ */
+#define MC_CMD_SENSOR_INFO 0x41
+#define MC_CMD_SENSOR_INFO_IN_LEN 0
+#define MC_CMD_SENSOR_INFO_OUT_MASK_OFST 0
+#define MC_CMD_SENSOR_INFO_OUT_OFFSET_OFST(_x) \
+	(4 + (_x))
+#define MC_CMD_SENSOR_INFO_OUT_MIN1_OFST(_ofst) \
+	((_ofst) + 0)
+#define MC_CMD_SENSOR_INFO_OUT_MAX1_OFST(_ofst) \
+	((_ofst) + 2)
+#define MC_CMD_SENSOR_INFO_OUT_MIN2_OFST(_ofst) \
+	((_ofst) + 4)
+#define MC_CMD_SENSOR_INFO_OUT_MAX2_OFST(_ofst) \
+	((_ofst) + 6)
+
+/* MC_CMD_READ_SENSORS
+ * Returns the current (value, state) for each sensor
+ *
+ * Returns the current (value, state) [each 16bit] of each sensor supported by
+ * this board, by DMA'ing a sparse array (indexed by the sensor type) into host
+ * memory.
+ *
+ * The MC will send a SENSOREVT event every time any sensor changes state. The
+ * driver is responsible for ensuring that it doesn't miss any events. The board
+ * will function normally if all sensors are in STATE_OK or state_WARNING.
+ * Otherwise the board should not be expected to function.
+ */
+#define MC_CMD_READ_SENSORS 0x42
+#define MC_CMD_READ_SENSORS_IN_LEN 8
+#define MC_CMD_READ_SENSORS_IN_DMA_ADDR_LO_OFST 0
+#define MC_CMD_READ_SENSORS_IN_DMA_ADDR_HI_OFST 4
+#define MC_CMD_READ_SENSORS_OUT_LEN 0
+
+
+/* MC_CMD_GET_PHY_STATE:
+ * Report current state of PHY.  A "zombie" PHY is a PHY that has failed to
+ * boot (e.g. due to missing or corrupted firmware).
+ *
+ * Locks required: None
+ * Return code: 0
+ */
+#define MC_CMD_GET_PHY_STATE 0x43
+
+#define MC_CMD_GET_PHY_STATE_IN_LEN 0
+#define MC_CMD_GET_PHY_STATE_OUT_LEN 4
+#define MC_CMD_GET_PHY_STATE_STATE_OFST 0
+/* PHY state enumeration: */
+#define MC_CMD_PHY_STATE_OK 1
+#define MC_CMD_PHY_STATE_ZOMBIE 2
+
+
+/* 802.1Qbb control. 8 Tx queues that map to priorities 0 - 7. Use all 1s to
+ * disable 802.Qbb for a given priority. */
+#define MC_CMD_SETUP_8021QBB 0x44
+#define MC_CMD_SETUP_8021QBB_IN_LEN 32
+#define MC_CMD_SETUP_8021QBB_OUT_LEN 0
+#define MC_CMD_SETUP_8021QBB_IN_TXQS_OFFST 0
+
+
+/* MC_CMD_WOL_FILTER_GET:
+ * Retrieve ID of any WoL filters
+ *
+ * Locks required: None
+ * Returns: 0, ENOSYS
+ */
+#define MC_CMD_WOL_FILTER_GET 0x45
+#define MC_CMD_WOL_FILTER_GET_IN_LEN 0
+#define MC_CMD_WOL_FILTER_GET_OUT_LEN 4
+#define MC_CMD_WOL_FILTER_GET_OUT_FILTER_ID_OFST 0
+
+
+/* MC_CMD_ADD_LIGHTSOUT_OFFLOAD:
+ * Offload a protocol to NIC for lights-out state
+ *
+ * Locks required: None
+ * Returns: 0, ENOSYS
+ */
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD 0x46
+
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_LEN 16
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_PROTOCOL_OFST 0
+
+/* There is a union at offset 4, following defines overlap due to
+ * this */
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_DATA_OFST 4
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_ARPMAC_OFST 4
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_ARPIP_OFST 10
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_NSMAC_OFST 4
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_NSSNIPV6_OFST 10
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_IN_NSIPV6_OFST 26
+
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_OUT_LEN 4
+#define MC_CMD_ADD_LIGHTSOUT_OFFLOAD_OUT_FILTER_ID_OFST 0
+
+
+/* MC_CMD_REMOVE_LIGHTSOUT_PROTOCOL_OFFLOAD:
+ * Offload a protocol to NIC for lights-out state
+ *
+ * Locks required: None
+ * Returns: 0, ENOSYS
+ */
+#define MC_CMD_REMOVE_LIGHTSOUT_OFFLOAD 0x47
+#define MC_CMD_REMOVE_LIGHTSOUT_OFFLOAD_IN_LEN 8
+#define MC_CMD_REMOVE_LIGHTSOUT_OFFLOAD_OUT_LEN 0
+
+#define MC_CMD_REMOVE_LIGHTSOUT_OFFLOAD_IN_PROTOCOL_OFST 0
+#define MC_CMD_REMOVE_LIGHTSOUT_OFFLOAD_IN_FILTER_ID_OFST 4
+
+/* Lights-out offload protocols enumeration */
+#define MC_CMD_LIGHTSOUT_OFFLOAD_PROTOCOL_ARP 0x1
+#define MC_CMD_LIGHTSOUT_OFFLOAD_PROTOCOL_NS  0x2
+
+
+/* MC_CMD_MAC_RESET_RESTORE:
+ * Restore MAC after block reset
+ *
+ * Locks required: None
+ * Returns: 0
+ */
+
+#define MC_CMD_MAC_RESET_RESTORE 0x48
+#define MC_CMD_MAC_RESET_RESTORE_IN_LEN 0
+#define MC_CMD_MAC_RESET_RESTORE_OUT_LEN 0
+
+#endif /* MCDI_PCOL_H */
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related

* [PATCH 09/12] sfc: Add support for SFC9000 family (1)
From: Ben Hutchings @ 2009-11-30  1:15 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, linux-net-drivers

This adds support for the SFC9000 family of 10G Ethernet controllers
and LAN-on-motherboard chips, starting with the SFL9021 'Siena' and
SFC9020 'Bethpage'.

The SFC9000 family is based on the SFC4000 'Falcon' architecture, but
with some significant changes:

- Two ports are associated with two independent PCI functions
  (except SFC9010)
- Integrated 10GBASE-T PHY(s) (SFL9021/9022)
- MAC, PHY and board peripherals are managed by firmware
  - Driver does not require board-specific code
  - Firmware supports wake-on-LAN and lights-out management through NC-SI
- IPv6 checksum offload and RSS
- Filtering by MAC address and VLAN (not included in this code)
- PCI SR-IOV (not included in this code)

Credit for this code is largely due to my colleagues at Solarflare:

   Guido Barzini
   Steve Hodgson
   Kieran Mansley
   Matthew Slattery
   Neil Turton

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
 drivers/net/sfc/mcdi.c     | 1112 ++++++++++++++++++++++++++++++++++++++++++++
 drivers/net/sfc/mcdi.h     |  130 +++++
 drivers/net/sfc/mcdi_mac.c |  152 ++++++
 drivers/net/sfc/mcdi_phy.c |  597 ++++++++++++++++++++++++
 drivers/net/sfc/siena.c    |  604 ++++++++++++++++++++++++
 5 files changed, 2595 insertions(+), 0 deletions(-)
 create mode 100644 drivers/net/sfc/mcdi.c
 create mode 100644 drivers/net/sfc/mcdi.h
 create mode 100644 drivers/net/sfc/mcdi_mac.c
 create mode 100644 drivers/net/sfc/mcdi_phy.c
 create mode 100644 drivers/net/sfc/siena.c

diff --git a/drivers/net/sfc/mcdi.c b/drivers/net/sfc/mcdi.c
new file mode 100644
index 0000000..683353b
--- /dev/null
+++ b/drivers/net/sfc/mcdi.c
@@ -0,0 +1,1112 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2008-2009 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#include <linux/delay.h>
+#include "net_driver.h"
+#include "nic.h"
+#include "io.h"
+#include "regs.h"
+#include "mcdi_pcol.h"
+#include "phy.h"
+
+/**************************************************************************
+ *
+ * Management-Controller-to-Driver Interface
+ *
+ **************************************************************************
+ */
+
+/* Software-defined structure to the shared-memory */
+#define CMD_NOTIFY_PORT0 0
+#define CMD_NOTIFY_PORT1 4
+#define CMD_PDU_PORT0    0x008
+#define CMD_PDU_PORT1    0x108
+#define REBOOT_FLAG_PORT0 0x3f8
+#define REBOOT_FLAG_PORT1 0x3fc
+
+#define MCDI_RPC_TIMEOUT       10 /*seconds */
+
+#define MCDI_PDU(efx)							\
+	(efx_port_num(efx) ? CMD_PDU_PORT1 : CMD_PDU_PORT0)
+#define MCDI_DOORBELL(efx)						\
+	(efx_port_num(efx) ? CMD_NOTIFY_PORT1 : CMD_NOTIFY_PORT0)
+#define MCDI_REBOOT_FLAG(efx)						\
+	(efx_port_num(efx) ? REBOOT_FLAG_PORT1 : REBOOT_FLAG_PORT0)
+
+#define SEQ_MASK							\
+	EFX_MASK32(EFX_WIDTH(MCDI_HEADER_SEQ))
+
+static inline struct efx_mcdi_iface *efx_mcdi(struct efx_nic *efx)
+{
+	struct siena_nic_data *nic_data;
+	EFX_BUG_ON_PARANOID(efx_nic_rev(efx) < EFX_REV_SIENA_A0);
+	nic_data = efx->nic_data;
+	return &nic_data->mcdi;
+}
+
+void efx_mcdi_init(struct efx_nic *efx)
+{
+	struct efx_mcdi_iface *mcdi;
+
+	if (efx_nic_rev(efx) < EFX_REV_SIENA_A0)
+		return;
+
+	mcdi = efx_mcdi(efx);
+	init_waitqueue_head(&mcdi->wq);
+	spin_lock_init(&mcdi->iface_lock);
+	atomic_set(&mcdi->state, MCDI_STATE_QUIESCENT);
+	mcdi->mode = MCDI_MODE_POLL;
+
+	(void) efx_mcdi_poll_reboot(efx);
+}
+
+static void efx_mcdi_copyin(struct efx_nic *efx, unsigned cmd,
+			    const u8 *inbuf, size_t inlen)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+	unsigned pdu = FR_CZ_MC_TREG_SMEM + MCDI_PDU(efx);
+	unsigned doorbell = FR_CZ_MC_TREG_SMEM + MCDI_DOORBELL(efx);
+	unsigned int i;
+	efx_dword_t hdr;
+	u32 xflags, seqno;
+
+	BUG_ON(atomic_read(&mcdi->state) == MCDI_STATE_QUIESCENT);
+	BUG_ON(inlen & 3 || inlen >= 0x100);
+
+	seqno = mcdi->seqno & SEQ_MASK;
+	xflags = 0;
+	if (mcdi->mode == MCDI_MODE_EVENTS)
+		xflags |= MCDI_HEADER_XFLAGS_EVREQ;
+
+	EFX_POPULATE_DWORD_6(hdr,
+			     MCDI_HEADER_RESPONSE, 0,
+			     MCDI_HEADER_RESYNC, 1,
+			     MCDI_HEADER_CODE, cmd,
+			     MCDI_HEADER_DATALEN, inlen,
+			     MCDI_HEADER_SEQ, seqno,
+			     MCDI_HEADER_XFLAGS, xflags);
+
+	efx_writed(efx, &hdr, pdu);
+
+	for (i = 0; i < inlen; i += 4)
+		_efx_writed(efx, *((__le32 *)(inbuf + i)), pdu + 4 + i);
+
+	/* Ensure the payload is written out before the header */
+	wmb();
+
+	/* ring the doorbell with a distinctive value */
+	_efx_writed(efx, (__force __le32) 0x45789abc, doorbell);
+}
+
+static void efx_mcdi_copyout(struct efx_nic *efx, u8 *outbuf, size_t outlen)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+	unsigned int pdu = FR_CZ_MC_TREG_SMEM + MCDI_PDU(efx);
+	int i;
+
+	BUG_ON(atomic_read(&mcdi->state) == MCDI_STATE_QUIESCENT);
+	BUG_ON(outlen & 3 || outlen >= 0x100);
+
+	for (i = 0; i < outlen; i += 4)
+		*((__le32 *)(outbuf + i)) = _efx_readd(efx, pdu + 4 + i);
+}
+
+static int efx_mcdi_poll(struct efx_nic *efx)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+	unsigned int time, finish;
+	unsigned int respseq, respcmd, error;
+	unsigned int pdu = FR_CZ_MC_TREG_SMEM + MCDI_PDU(efx);
+	unsigned int rc, spins;
+	efx_dword_t reg;
+
+	/* Check for a reboot atomically with respect to efx_mcdi_copyout() */
+	rc = efx_mcdi_poll_reboot(efx);
+	if (rc)
+		goto out;
+
+	/* Poll for completion. Poll quickly (once a us) for the 1st jiffy,
+	 * because generally mcdi responses are fast. After that, back off
+	 * and poll once a jiffy (approximately)
+	 */
+	spins = TICK_USEC;
+	finish = get_seconds() + MCDI_RPC_TIMEOUT;
+
+	while (1) {
+		if (spins != 0) {
+			--spins;
+			udelay(1);
+		} else
+			schedule();
+
+		time = get_seconds();
+
+		rmb();
+		efx_readd(efx, &reg, pdu);
+
+		/* All 1's indicates that shared memory is in reset (and is
+		 * not a valid header). Wait for it to come out reset before
+		 * completing the command */
+		if (EFX_DWORD_FIELD(reg, EFX_DWORD_0) != 0xffffffff &&
+		    EFX_DWORD_FIELD(reg, MCDI_HEADER_RESPONSE))
+			break;
+
+		if (time >= finish)
+			return -ETIMEDOUT;
+	}
+
+	mcdi->resplen = EFX_DWORD_FIELD(reg, MCDI_HEADER_DATALEN);
+	respseq = EFX_DWORD_FIELD(reg, MCDI_HEADER_SEQ);
+	respcmd = EFX_DWORD_FIELD(reg, MCDI_HEADER_CODE);
+	error = EFX_DWORD_FIELD(reg, MCDI_HEADER_ERROR);
+
+	if (error && mcdi->resplen == 0) {
+		EFX_ERR(efx, "MC rebooted\n");
+		rc = EIO;
+	} else if ((respseq ^ mcdi->seqno) & SEQ_MASK) {
+		EFX_ERR(efx, "MC response mismatch tx seq 0x%x rx seq 0x%x\n",
+			respseq, mcdi->seqno);
+		rc = EIO;
+	} else if (error) {
+		efx_readd(efx, &reg, pdu + 4);
+		switch (EFX_DWORD_FIELD(reg, EFX_DWORD_0)) {
+#define TRANSLATE_ERROR(name)					\
+		case MC_CMD_ERR_ ## name:			\
+			rc = name;				\
+			break
+			TRANSLATE_ERROR(ENOENT);
+			TRANSLATE_ERROR(EINTR);
+			TRANSLATE_ERROR(EACCES);
+			TRANSLATE_ERROR(EBUSY);
+			TRANSLATE_ERROR(EINVAL);
+			TRANSLATE_ERROR(EDEADLK);
+			TRANSLATE_ERROR(ENOSYS);
+			TRANSLATE_ERROR(ETIME);
+#undef TRANSLATE_ERROR
+		default:
+			rc = EIO;
+			break;
+		}
+	} else
+		rc = 0;
+
+out:
+	mcdi->resprc = rc;
+	if (rc)
+		mcdi->resplen = 0;
+
+	/* Return rc=0 like wait_event_timeout() */
+	return 0;
+}
+
+/* Test and clear MC-rebooted flag for this port/function */
+int efx_mcdi_poll_reboot(struct efx_nic *efx)
+{
+	unsigned int addr = FR_CZ_MC_TREG_SMEM + MCDI_REBOOT_FLAG(efx);
+	efx_dword_t reg;
+	uint32_t value;
+
+	if (efx_nic_rev(efx) < EFX_REV_SIENA_A0)
+		return false;
+
+	efx_readd(efx, &reg, addr);
+	value = EFX_DWORD_FIELD(reg, EFX_DWORD_0);
+
+	if (value == 0)
+		return 0;
+
+	EFX_ZERO_DWORD(reg);
+	efx_writed(efx, &reg, addr);
+
+	if (value == MC_STATUS_DWORD_ASSERT)
+		return -EINTR;
+	else
+		return -EIO;
+}
+
+static void efx_mcdi_acquire(struct efx_mcdi_iface *mcdi)
+{
+	/* Wait until the interface becomes QUIESCENT and we win the race
+	 * to mark it RUNNING. */
+	wait_event(mcdi->wq,
+		   atomic_cmpxchg(&mcdi->state,
+				  MCDI_STATE_QUIESCENT,
+				  MCDI_STATE_RUNNING)
+		   == MCDI_STATE_QUIESCENT);
+}
+
+static int efx_mcdi_await_completion(struct efx_nic *efx)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+
+	if (wait_event_timeout(
+		    mcdi->wq,
+		    atomic_read(&mcdi->state) == MCDI_STATE_COMPLETED,
+		    msecs_to_jiffies(MCDI_RPC_TIMEOUT * 1000)) == 0)
+		return -ETIMEDOUT;
+
+	/* Check if efx_mcdi_set_mode() switched us back to polled completions.
+	 * In which case, poll for completions directly. If efx_mcdi_ev_cpl()
+	 * completed the request first, then we'll just end up completing the
+	 * request again, which is safe.
+	 *
+	 * We need an smp_rmb() to synchronise with efx_mcdi_mode_poll(), which
+	 * wait_event_timeout() implicitly provides.
+	 */
+	if (mcdi->mode == MCDI_MODE_POLL)
+		return efx_mcdi_poll(efx);
+
+	return 0;
+}
+
+static bool efx_mcdi_complete(struct efx_mcdi_iface *mcdi)
+{
+	/* If the interface is RUNNING, then move to COMPLETED and wake any
+	 * waiters. If the interface isn't in RUNNING then we've received a
+	 * duplicate completion after we've already transitioned back to
+	 * QUIESCENT. [A subsequent invocation would increment seqno, so would
+	 * have failed the seqno check].
+	 */
+	if (atomic_cmpxchg(&mcdi->state,
+			   MCDI_STATE_RUNNING,
+			   MCDI_STATE_COMPLETED) == MCDI_STATE_RUNNING) {
+		wake_up(&mcdi->wq);
+		return true;
+	}
+
+	return false;
+}
+
+static void efx_mcdi_release(struct efx_mcdi_iface *mcdi)
+{
+	atomic_set(&mcdi->state, MCDI_STATE_QUIESCENT);
+	wake_up(&mcdi->wq);
+}
+
+static void efx_mcdi_ev_cpl(struct efx_nic *efx, unsigned int seqno,
+			    unsigned int datalen, unsigned int errno)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+	bool wake = false;
+
+	spin_lock(&mcdi->iface_lock);
+
+	if ((seqno ^ mcdi->seqno) & SEQ_MASK) {
+		if (mcdi->credits)
+			/* The request has been cancelled */
+			--mcdi->credits;
+		else
+			EFX_ERR(efx, "MC response mismatch tx seq 0x%x rx "
+				"seq 0x%x\n", seqno, mcdi->seqno);
+	} else {
+		mcdi->resprc = errno;
+		mcdi->resplen = datalen;
+
+		wake = true;
+	}
+
+	spin_unlock(&mcdi->iface_lock);
+
+	if (wake)
+		efx_mcdi_complete(mcdi);
+}
+
+/* Issue the given command by writing the data into the shared memory PDU,
+ * ring the doorbell and wait for completion. Copyout the result. */
+int efx_mcdi_rpc(struct efx_nic *efx, unsigned cmd,
+		 const u8 *inbuf, size_t inlen, u8 *outbuf, size_t outlen,
+		 size_t *outlen_actual)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+	int rc;
+	BUG_ON(efx_nic_rev(efx) < EFX_REV_SIENA_A0);
+
+	efx_mcdi_acquire(mcdi);
+
+	/* Serialise with efx_mcdi_ev_cpl() and efx_mcdi_ev_death() */
+	spin_lock_bh(&mcdi->iface_lock);
+	++mcdi->seqno;
+	spin_unlock_bh(&mcdi->iface_lock);
+
+	efx_mcdi_copyin(efx, cmd, inbuf, inlen);
+
+	if (mcdi->mode == MCDI_MODE_POLL)
+		rc = efx_mcdi_poll(efx);
+	else
+		rc = efx_mcdi_await_completion(efx);
+
+	if (rc != 0) {
+		/* Close the race with efx_mcdi_ev_cpl() executing just too late
+		 * and completing a request we've just cancelled, by ensuring
+		 * that the seqno check therein fails.
+		 */
+		spin_lock_bh(&mcdi->iface_lock);
+		++mcdi->seqno;
+		++mcdi->credits;
+		spin_unlock_bh(&mcdi->iface_lock);
+
+		EFX_ERR(efx, "MC command 0x%x inlen %d mode %d timed out\n",
+			cmd, (int)inlen, mcdi->mode);
+	} else {
+		size_t resplen;
+
+		/* At the very least we need a memory barrier here to ensure
+		 * we pick up changes from efx_mcdi_ev_cpl(). Protect against
+		 * a spurious efx_mcdi_ev_cpl() running concurrently by
+		 * acquiring the iface_lock. */
+		spin_lock_bh(&mcdi->iface_lock);
+		rc = -mcdi->resprc;
+		resplen = mcdi->resplen;
+		spin_unlock_bh(&mcdi->iface_lock);
+
+		if (rc == 0) {
+			efx_mcdi_copyout(efx, outbuf,
+					 min(outlen, mcdi->resplen + 3) & ~0x3);
+			if (outlen_actual != NULL)
+				*outlen_actual = resplen;
+		} else if (cmd == MC_CMD_REBOOT && rc == -EIO)
+			; /* Don't reset if MC_CMD_REBOOT returns EIO */
+		else if (rc == -EIO || rc == -EINTR) {
+			EFX_ERR(efx, "MC fatal error %d\n", -rc);
+			efx_schedule_reset(efx, RESET_TYPE_MC_FAILURE);
+		} else
+			EFX_ERR(efx, "MC command 0x%x inlen %d failed rc=%d\n",
+				cmd, (int)inlen, -rc);
+	}
+
+	efx_mcdi_release(mcdi);
+	return rc;
+}
+
+void efx_mcdi_mode_poll(struct efx_nic *efx)
+{
+	struct efx_mcdi_iface *mcdi;
+
+	if (efx_nic_rev(efx) < EFX_REV_SIENA_A0)
+		return;
+
+	mcdi = efx_mcdi(efx);
+	if (mcdi->mode == MCDI_MODE_POLL)
+		return;
+
+	/* We can switch from event completion to polled completion, because
+	 * mcdi requests are always completed in shared memory. We do this by
+	 * switching the mode to POLL'd then completing the request.
+	 * efx_mcdi_await_completion() will then call efx_mcdi_poll().
+	 *
+	 * We need an smp_wmb() to synchronise with efx_mcdi_await_completion(),
+	 * which efx_mcdi_complete() provides for us.
+	 */
+	mcdi->mode = MCDI_MODE_POLL;
+
+	efx_mcdi_complete(mcdi);
+}
+
+void efx_mcdi_mode_event(struct efx_nic *efx)
+{
+	struct efx_mcdi_iface *mcdi;
+
+	if (efx_nic_rev(efx) < EFX_REV_SIENA_A0)
+		return;
+
+	mcdi = efx_mcdi(efx);
+
+	if (mcdi->mode == MCDI_MODE_EVENTS)
+		return;
+
+	/* We can't switch from polled to event completion in the middle of a
+	 * request, because the completion method is specified in the request.
+	 * So acquire the interface to serialise the requestors. We don't need
+	 * to acquire the iface_lock to change the mode here, but we do need a
+	 * write memory barrier ensure that efx_mcdi_rpc() sees it, which
+	 * efx_mcdi_acquire() provides.
+	 */
+	efx_mcdi_acquire(mcdi);
+	mcdi->mode = MCDI_MODE_EVENTS;
+	efx_mcdi_release(mcdi);
+}
+
+static void efx_mcdi_ev_death(struct efx_nic *efx, int rc)
+{
+	struct efx_mcdi_iface *mcdi = efx_mcdi(efx);
+
+	/* If there is an outstanding MCDI request, it has been terminated
+	 * either by a BADASSERT or REBOOT event. If the mcdi interface is
+	 * in polled mode, then do nothing because the MC reboot handler will
+	 * set the header correctly. However, if the mcdi interface is waiting
+	 * for a CMDDONE event it won't receive it [and since all MCDI events
+	 * are sent to the same queue, we can't be racing with
+	 * efx_mcdi_ev_cpl()]
+	 *
+	 * There's a race here with efx_mcdi_rpc(), because we might receive
+	 * a REBOOT event *before* the request has been copied out. In polled
+	 * mode (during startup) this is irrelevent, because efx_mcdi_complete()
+	 * is ignored. In event mode, this condition is just an edge-case of
+	 * receiving a REBOOT event after posting the MCDI request. Did the mc
+	 * reboot before or after the copyout? The best we can do always is
+	 * just return failure.
+	 */
+	spin_lock(&mcdi->iface_lock);
+	if (efx_mcdi_complete(mcdi)) {
+		if (mcdi->mode == MCDI_MODE_EVENTS) {
+			mcdi->resprc = rc;
+			mcdi->resplen = 0;
+		}
+	} else
+		/* Nobody was waiting for an MCDI request, so trigger a reset */
+		efx_schedule_reset(efx, RESET_TYPE_MC_FAILURE);
+
+	spin_unlock(&mcdi->iface_lock);
+}
+
+static unsigned int efx_mcdi_event_link_speed[] = {
+	[MCDI_EVENT_LINKCHANGE_SPEED_100M] = 100,
+	[MCDI_EVENT_LINKCHANGE_SPEED_1G] = 1000,
+	[MCDI_EVENT_LINKCHANGE_SPEED_10G] = 10000,
+};
+
+
+static void efx_mcdi_process_link_change(struct efx_nic *efx, efx_qword_t *ev)
+{
+	u32 flags, fcntl, speed, lpa;
+
+	speed = EFX_QWORD_FIELD(*ev, MCDI_EVENT_LINKCHANGE_SPEED);
+	EFX_BUG_ON_PARANOID(speed >= ARRAY_SIZE(efx_mcdi_event_link_speed));
+	speed = efx_mcdi_event_link_speed[speed];
+
+	flags = EFX_QWORD_FIELD(*ev, MCDI_EVENT_LINKCHANGE_LINK_FLAGS);
+	fcntl = EFX_QWORD_FIELD(*ev, MCDI_EVENT_LINKCHANGE_FCNTL);
+	lpa = EFX_QWORD_FIELD(*ev, MCDI_EVENT_LINKCHANGE_LP_CAP);
+
+	/* efx->link_state is only modified by efx_mcdi_phy_get_link(),
+	 * which is only run after flushing the event queues. Therefore, it
+	 * is safe to modify the link state outside of the mac_lock here.
+	 */
+	efx_mcdi_phy_decode_link(efx, &efx->link_state, speed, flags, fcntl);
+
+	efx_mcdi_phy_check_fcntl(efx, lpa);
+
+	efx_link_status_changed(efx);
+}
+
+static const char *sensor_names[] = {
+	[MC_CMD_SENSOR_CONTROLLER_TEMP] = "Controller temp. sensor",
+	[MC_CMD_SENSOR_PHY_COMMON_TEMP] = "PHY shared temp. sensor",
+	[MC_CMD_SENSOR_CONTROLLER_COOLING] = "Controller cooling",
+	[MC_CMD_SENSOR_PHY0_TEMP] = "PHY 0 temp. sensor",
+	[MC_CMD_SENSOR_PHY0_COOLING] = "PHY 0 cooling",
+	[MC_CMD_SENSOR_PHY1_TEMP] = "PHY 1 temp. sensor",
+	[MC_CMD_SENSOR_PHY1_COOLING] = "PHY 1 cooling",
+	[MC_CMD_SENSOR_IN_1V0] = "1.0V supply sensor",
+	[MC_CMD_SENSOR_IN_1V2] = "1.2V supply sensor",
+	[MC_CMD_SENSOR_IN_1V8] = "1.8V supply sensor",
+	[MC_CMD_SENSOR_IN_2V5] = "2.5V supply sensor",
+	[MC_CMD_SENSOR_IN_3V3] = "3.3V supply sensor",
+	[MC_CMD_SENSOR_IN_12V0] = "12V supply sensor"
+};
+
+static const char *sensor_status_names[] = {
+	[MC_CMD_SENSOR_STATE_OK] = "OK",
+	[MC_CMD_SENSOR_STATE_WARNING] = "Warning",
+	[MC_CMD_SENSOR_STATE_FATAL] = "Fatal",
+	[MC_CMD_SENSOR_STATE_BROKEN] = "Device failure",
+};
+
+static void efx_mcdi_sensor_event(struct efx_nic *efx, efx_qword_t *ev)
+{
+	unsigned int monitor, state, value;
+	const char *name, *state_txt;
+	monitor = EFX_QWORD_FIELD(*ev, MCDI_EVENT_SENSOREVT_MONITOR);
+	state = EFX_QWORD_FIELD(*ev, MCDI_EVENT_SENSOREVT_STATE);
+	value = EFX_QWORD_FIELD(*ev, MCDI_EVENT_SENSOREVT_VALUE);
+	/* Deal gracefully with the board having more drivers than we
+	 * know about, but do not expect new sensor states. */
+	name = (monitor >= ARRAY_SIZE(sensor_names))
+				    ? "No sensor name available" :
+				    sensor_names[monitor];
+	EFX_BUG_ON_PARANOID(state >= ARRAY_SIZE(sensor_status_names));
+	state_txt = sensor_status_names[state];
+
+	EFX_ERR(efx, "Sensor %d (%s) reports condition '%s' for raw value %d\n",
+		monitor, name, state_txt, value);
+}
+
+/* Called from  falcon_process_eventq for MCDI events */
+void efx_mcdi_process_event(struct efx_channel *channel,
+			    efx_qword_t *event)
+{
+	struct efx_nic *efx = channel->efx;
+	int code = EFX_QWORD_FIELD(*event, MCDI_EVENT_CODE);
+	u32 data = EFX_QWORD_FIELD(*event, MCDI_EVENT_DATA);
+
+	switch (code) {
+	case MCDI_EVENT_CODE_BADSSERT:
+		EFX_ERR(efx, "MC watchdog or assertion failure at 0x%x\n", data);
+		efx_mcdi_ev_death(efx, EINTR);
+		break;
+
+	case MCDI_EVENT_CODE_PMNOTICE:
+		EFX_INFO(efx, "MCDI PM event.\n");
+		break;
+
+	case MCDI_EVENT_CODE_CMDDONE:
+		efx_mcdi_ev_cpl(efx,
+				MCDI_EVENT_FIELD(*event, CMDDONE_SEQ),
+				MCDI_EVENT_FIELD(*event, CMDDONE_DATALEN),
+				MCDI_EVENT_FIELD(*event, CMDDONE_ERRNO));
+		break;
+
+	case MCDI_EVENT_CODE_LINKCHANGE:
+		efx_mcdi_process_link_change(efx, event);
+		break;
+	case MCDI_EVENT_CODE_SENSOREVT:
+		efx_mcdi_sensor_event(efx, event);
+		break;
+	case MCDI_EVENT_CODE_SCHEDERR:
+		EFX_INFO(efx, "MC Scheduler error address=0x%x\n", data);
+		break;
+	case MCDI_EVENT_CODE_REBOOT:
+		EFX_INFO(efx, "MC Reboot\n");
+		efx_mcdi_ev_death(efx, EIO);
+		break;
+	case MCDI_EVENT_CODE_MAC_STATS_DMA:
+		/* MAC stats are gather lazily.  We can ignore this. */
+		break;
+
+	default:
+		EFX_ERR(efx, "Unknown MCDI event 0x%x\n", code);
+	}
+}
+
+/**************************************************************************
+ *
+ * Specific request functions
+ *
+ **************************************************************************
+ */
+
+int efx_mcdi_fwver(struct efx_nic *efx, u64 *version, u32 *build)
+{
+	u8 outbuf[ALIGN(MC_CMD_GET_VERSION_V1_OUT_LEN, 4)];
+	size_t outlength;
+	const __le16 *ver_words;
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_GET_VERSION_IN_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_VERSION, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlength);
+	if (rc)
+		goto fail;
+
+	if (outlength == MC_CMD_GET_VERSION_V0_OUT_LEN) {
+		*version = 0;
+		*build = MCDI_DWORD(outbuf, GET_VERSION_OUT_FIRMWARE);
+		return 0;
+	}
+
+	if (outlength < MC_CMD_GET_VERSION_V1_OUT_LEN) {
+		rc = -EMSGSIZE;
+		goto fail;
+	}
+
+	ver_words = (__le16 *)MCDI_PTR(outbuf, GET_VERSION_OUT_VERSION);
+	*version = (((u64)le16_to_cpu(ver_words[0]) << 48) |
+		    ((u64)le16_to_cpu(ver_words[1]) << 32) |
+		    ((u64)le16_to_cpu(ver_words[2]) << 16) |
+		    le16_to_cpu(ver_words[3]));
+	*build = MCDI_DWORD(outbuf, GET_VERSION_OUT_FIRMWARE);
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_drv_attach(struct efx_nic *efx, bool driver_operating,
+			bool *was_attached)
+{
+	u8 inbuf[MC_CMD_DRV_ATTACH_IN_LEN];
+	u8 outbuf[MC_CMD_DRV_ATTACH_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, DRV_ATTACH_IN_NEW_STATE,
+		       driver_operating ? 1 : 0);
+	MCDI_SET_DWORD(inbuf, DRV_ATTACH_IN_UPDATE, 1);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_DRV_ATTACH, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+	if (outlen < MC_CMD_DRV_ATTACH_OUT_LEN)
+		goto fail;
+
+	if (was_attached != NULL)
+		*was_attached = MCDI_DWORD(outbuf, DRV_ATTACH_OUT_OLD_STATE);
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_get_board_cfg(struct efx_nic *efx, u8 *mac_address,
+			   u16 *fw_subtype_list)
+{
+	uint8_t outbuf[MC_CMD_GET_BOARD_CFG_OUT_LEN];
+	size_t outlen;
+	int port_num = efx_port_num(efx);
+	int offset;
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_GET_BOARD_CFG_IN_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_BOARD_CFG, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	if (outlen < MC_CMD_GET_BOARD_CFG_OUT_LEN) {
+		rc = -EMSGSIZE;
+		goto fail;
+	}
+
+	offset = (port_num)
+		? MC_CMD_GET_BOARD_CFG_OUT_MAC_ADDR_BASE_PORT1_OFST
+		: MC_CMD_GET_BOARD_CFG_OUT_MAC_ADDR_BASE_PORT0_OFST;
+	if (mac_address)
+		memcpy(mac_address, outbuf + offset, ETH_ALEN);
+	if (fw_subtype_list)
+		memcpy(fw_subtype_list,
+		       outbuf + MC_CMD_GET_BOARD_CFG_OUT_FW_SUBTYPE_LIST_OFST,
+		       MC_CMD_GET_BOARD_CFG_OUT_FW_SUBTYPE_LIST_LEN);
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d len=%d\n", __func__, rc, (int)outlen);
+
+	return rc;
+}
+
+int efx_mcdi_log_ctrl(struct efx_nic *efx, bool evq, bool uart, u32 dest_evq)
+{
+	u8 inbuf[MC_CMD_LOG_CTRL_IN_LEN];
+	u32 dest = 0;
+	int rc;
+
+	if (uart)
+		dest |= MC_CMD_LOG_CTRL_IN_LOG_DEST_UART;
+	if (evq)
+		dest |= MC_CMD_LOG_CTRL_IN_LOG_DEST_EVQ;
+
+	MCDI_SET_DWORD(inbuf, LOG_CTRL_IN_LOG_DEST, dest);
+	MCDI_SET_DWORD(inbuf, LOG_CTRL_IN_LOG_DEST_EVQ, dest_evq);
+
+	BUILD_BUG_ON(MC_CMD_LOG_CTRL_OUT_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_LOG_CTRL, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_types(struct efx_nic *efx, u32 *nvram_types_out)
+{
+	u8 outbuf[MC_CMD_NVRAM_TYPES_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_NVRAM_TYPES_IN_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_TYPES, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+	if (outlen < MC_CMD_NVRAM_TYPES_OUT_LEN)
+		goto fail;
+
+	*nvram_types_out = MCDI_DWORD(outbuf, NVRAM_TYPES_OUT_TYPES);
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n",
+		__func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_info(struct efx_nic *efx, unsigned int type,
+			size_t *size_out, size_t *erase_size_out,
+			bool *protected_out)
+{
+	u8 inbuf[MC_CMD_NVRAM_INFO_IN_LEN];
+	u8 outbuf[MC_CMD_NVRAM_INFO_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, NVRAM_INFO_IN_TYPE, type);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_INFO, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+	if (outlen < MC_CMD_NVRAM_INFO_OUT_LEN)
+		goto fail;
+
+	*size_out = MCDI_DWORD(outbuf, NVRAM_INFO_OUT_SIZE);
+	*erase_size_out = MCDI_DWORD(outbuf, NVRAM_INFO_OUT_ERASESIZE);
+	*protected_out = !!(MCDI_DWORD(outbuf, NVRAM_INFO_OUT_FLAGS) &
+				(1 << MC_CMD_NVRAM_PROTECTED_LBN));
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_update_start(struct efx_nic *efx, unsigned int type)
+{
+	u8 inbuf[MC_CMD_NVRAM_UPDATE_START_IN_LEN];
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, NVRAM_UPDATE_START_IN_TYPE, type);
+
+	BUILD_BUG_ON(MC_CMD_NVRAM_UPDATE_START_OUT_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_UPDATE_START, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_read(struct efx_nic *efx, unsigned int type,
+			loff_t offset, u8 *buffer, size_t length)
+{
+	u8 inbuf[MC_CMD_NVRAM_READ_IN_LEN];
+	u8 outbuf[MC_CMD_NVRAM_READ_OUT_LEN(length)];
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, NVRAM_READ_IN_TYPE, type);
+	MCDI_SET_DWORD(inbuf, NVRAM_READ_IN_OFFSET, offset);
+	MCDI_SET_DWORD(inbuf, NVRAM_READ_IN_LENGTH, length);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_READ, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	memcpy(buffer, MCDI_PTR(outbuf, NVRAM_READ_OUT_READ_BUFFER), length);
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_write(struct efx_nic *efx, unsigned int type,
+			   loff_t offset, const u8 *buffer, size_t length)
+{
+	u8 inbuf[MC_CMD_NVRAM_WRITE_IN_LEN(length)];
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, NVRAM_WRITE_IN_TYPE, type);
+	MCDI_SET_DWORD(inbuf, NVRAM_WRITE_IN_OFFSET, offset);
+	MCDI_SET_DWORD(inbuf, NVRAM_WRITE_IN_LENGTH, length);
+	memcpy(MCDI_PTR(inbuf, NVRAM_WRITE_IN_WRITE_BUFFER), buffer, length);
+
+	BUILD_BUG_ON(MC_CMD_NVRAM_WRITE_OUT_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_WRITE, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_erase(struct efx_nic *efx, unsigned int type,
+			 loff_t offset, size_t length)
+{
+	u8 inbuf[MC_CMD_NVRAM_ERASE_IN_LEN];
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, NVRAM_ERASE_IN_TYPE, type);
+	MCDI_SET_DWORD(inbuf, NVRAM_ERASE_IN_OFFSET, offset);
+	MCDI_SET_DWORD(inbuf, NVRAM_ERASE_IN_LENGTH, length);
+
+	BUILD_BUG_ON(MC_CMD_NVRAM_ERASE_OUT_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_ERASE, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_nvram_update_finish(struct efx_nic *efx, unsigned int type)
+{
+	u8 inbuf[MC_CMD_NVRAM_UPDATE_FINISH_IN_LEN];
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, NVRAM_UPDATE_FINISH_IN_TYPE, type);
+
+	BUILD_BUG_ON(MC_CMD_NVRAM_UPDATE_FINISH_OUT_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_NVRAM_UPDATE_FINISH, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_handle_assertion(struct efx_nic *efx)
+{
+	union {
+		u8 asserts[MC_CMD_GET_ASSERTS_IN_LEN];
+		u8 reboot[MC_CMD_REBOOT_IN_LEN];
+	} inbuf;
+	u8 assertion[MC_CMD_GET_ASSERTS_OUT_LEN];
+	unsigned int flags, index, ofst;
+	const char *reason;
+	size_t outlen;
+	int retry;
+	int rc;
+
+	/* Check if the MC is in the assertion handler, retrying twice. Once
+	 * because a boot-time assertion might cause this command to fail
+	 * with EINTR. And once again because GET_ASSERTS can race with
+	 * MC_CMD_REBOOT running on the other port. */
+	retry = 2;
+	do {
+		MCDI_SET_DWORD(inbuf.asserts, GET_ASSERTS_IN_CLEAR, 0);
+		rc = efx_mcdi_rpc(efx, MC_CMD_GET_ASSERTS,
+				  inbuf.asserts, MC_CMD_GET_ASSERTS_IN_LEN,
+				  assertion, sizeof(assertion), &outlen);
+	} while ((rc == -EINTR || rc == -EIO) && retry-- > 0);
+
+	if (rc)
+		return rc;
+	if (outlen < MC_CMD_GET_ASSERTS_OUT_LEN)
+		return -EINVAL;
+
+	flags = MCDI_DWORD(assertion, GET_ASSERTS_OUT_GLOBAL_FLAGS);
+	if (flags == MC_CMD_GET_ASSERTS_FLAGS_NO_FAILS)
+		return 0;
+
+	/* Reset the hardware atomically such that only one port with succeed.
+	 * This command will succeed if a reboot is no longer required (because
+	 * the other port did it first), but fail with EIO if it succeeds.
+	 */
+	BUILD_BUG_ON(MC_CMD_REBOOT_OUT_LEN != 0);
+	MCDI_SET_DWORD(inbuf.reboot, REBOOT_IN_FLAGS,
+		       MC_CMD_REBOOT_FLAGS_AFTER_ASSERTION);
+	efx_mcdi_rpc(efx, MC_CMD_REBOOT, inbuf.reboot, MC_CMD_REBOOT_IN_LEN,
+		     NULL, 0, NULL);
+
+	/* Print out the assertion */
+	reason = (flags == MC_CMD_GET_ASSERTS_FLAGS_SYS_FAIL)
+		? "system-level assertion"
+		: (flags == MC_CMD_GET_ASSERTS_FLAGS_THR_FAIL)
+		? "thread-level assertion"
+		: (flags == MC_CMD_GET_ASSERTS_FLAGS_WDOG_FIRED)
+		? "watchdog reset"
+		: "unknown assertion";
+	EFX_ERR(efx, "MCPU %s at PC = 0x%.8x in thread 0x%.8x\n", reason,
+		MCDI_DWORD(assertion, GET_ASSERTS_OUT_SAVED_PC_OFFS),
+		MCDI_DWORD(assertion, GET_ASSERTS_OUT_THREAD_OFFS));
+
+	/* Print out the registers */
+	ofst = MC_CMD_GET_ASSERTS_OUT_GP_REGS_OFFS_OFST;
+	for (index = 1; index < 32; index++) {
+		EFX_ERR(efx, "R%.2d (?): 0x%.8x\n", index,
+			MCDI_DWORD2(assertion, ofst));
+		ofst += sizeof(efx_dword_t);
+	}
+
+	return 0;
+}
+
+void efx_mcdi_set_id_led(struct efx_nic *efx, enum efx_led_mode mode)
+{
+	u8 inbuf[MC_CMD_SET_ID_LED_IN_LEN];
+	int rc;
+
+	BUILD_BUG_ON(EFX_LED_OFF != MC_CMD_LED_OFF);
+	BUILD_BUG_ON(EFX_LED_ON != MC_CMD_LED_ON);
+	BUILD_BUG_ON(EFX_LED_DEFAULT != MC_CMD_LED_DEFAULT);
+
+	BUILD_BUG_ON(MC_CMD_SET_ID_LED_OUT_LEN != 0);
+
+	MCDI_SET_DWORD(inbuf, SET_ID_LED_IN_STATE, mode);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_SET_ID_LED, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+}
+
+int efx_mcdi_reset_port(struct efx_nic *efx)
+{
+	int rc = efx_mcdi_rpc(efx, MC_CMD_PORT_RESET, NULL, 0, NULL, 0, NULL);
+	if (rc)
+		EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_reset_mc(struct efx_nic *efx)
+{
+	u8 inbuf[MC_CMD_REBOOT_IN_LEN];
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_REBOOT_OUT_LEN != 0);
+	MCDI_SET_DWORD(inbuf, REBOOT_IN_FLAGS, 0);
+	rc = efx_mcdi_rpc(efx, MC_CMD_REBOOT, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	/* White is black, and up is down */
+	if (rc == -EIO)
+		return 0;
+	if (rc == 0)
+		rc = -EIO;
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_wol_filter_set(struct efx_nic *efx, u32 type,
+			    const u8 *mac, int *id_out)
+{
+	u8 inbuf[MC_CMD_WOL_FILTER_SET_IN_LEN];
+	u8 outbuf[MC_CMD_WOL_FILTER_SET_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, WOL_FILTER_SET_IN_WOL_TYPE, type);
+	MCDI_SET_DWORD(inbuf, WOL_FILTER_SET_IN_FILTER_MODE,
+		       MC_CMD_FILTER_MODE_SIMPLE);
+	memcpy(MCDI_PTR(inbuf, WOL_FILTER_SET_IN_MAGIC_MAC), mac, ETH_ALEN);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_WOL_FILTER_SET, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	if (outlen < MC_CMD_WOL_FILTER_SET_OUT_LEN) {
+		rc = -EMSGSIZE;
+		goto fail;
+	}
+
+	*id_out = (int)MCDI_DWORD(outbuf, WOL_FILTER_SET_OUT_FILTER_ID);
+
+	return 0;
+
+fail:
+	*id_out = -1;
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+
+}
+
+
+int
+efx_mcdi_wol_filter_set_magic(struct efx_nic *efx,  const u8 *mac, int *id_out)
+{
+	return efx_mcdi_wol_filter_set(efx, MC_CMD_WOL_TYPE_MAGIC, mac, id_out);
+}
+
+
+int efx_mcdi_wol_filter_get_magic(struct efx_nic *efx, int *id_out)
+{
+	u8 outbuf[MC_CMD_WOL_FILTER_GET_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_WOL_FILTER_GET, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	if (outlen < MC_CMD_WOL_FILTER_GET_OUT_LEN) {
+		rc = -EMSGSIZE;
+		goto fail;
+	}
+
+	*id_out = (int)MCDI_DWORD(outbuf, WOL_FILTER_GET_OUT_FILTER_ID);
+
+	return 0;
+
+fail:
+	*id_out = -1;
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+
+int efx_mcdi_wol_filter_remove(struct efx_nic *efx, int id)
+{
+	u8 inbuf[MC_CMD_WOL_FILTER_REMOVE_IN_LEN];
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, WOL_FILTER_REMOVE_IN_FILTER_ID, (u32)id);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_WOL_FILTER_REMOVE, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+
+int efx_mcdi_wol_filter_reset(struct efx_nic *efx)
+{
+	int rc;
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_WOL_FILTER_RESET, NULL, 0, NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
diff --git a/drivers/net/sfc/mcdi.h b/drivers/net/sfc/mcdi.h
new file mode 100644
index 0000000..de91672
--- /dev/null
+++ b/drivers/net/sfc/mcdi.h
@@ -0,0 +1,130 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2008-2009 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#ifndef EFX_MCDI_H
+#define EFX_MCDI_H
+
+/**
+ * enum efx_mcdi_state
+ * @MCDI_STATE_QUIESCENT: No pending MCDI requests. If the caller holds the
+ *	mcdi_lock then they are able to move to MCDI_STATE_RUNNING
+ * @MCDI_STATE_RUNNING: There is an MCDI request pending. Only the thread that
+ *	moved into this state is allowed to move out of it.
+ * @MCDI_STATE_COMPLETED: An MCDI request has completed, but the owning thread
+ *	has not yet consumed the result. For all other threads, equivalent to
+ *	MCDI_STATE_RUNNING.
+ */
+enum efx_mcdi_state {
+	MCDI_STATE_QUIESCENT,
+	MCDI_STATE_RUNNING,
+	MCDI_STATE_COMPLETED,
+};
+
+enum efx_mcdi_mode {
+	MCDI_MODE_POLL,
+	MCDI_MODE_EVENTS,
+};
+
+/**
+ * struct efx_mcdi_iface
+ * @state: Interface state. Waited for by mcdi_wq.
+ * @wq: Wait queue for threads waiting for state != STATE_RUNNING
+ * @iface_lock: Protects @credits, @seqno, @resprc, @resplen
+ * @mode: Poll for mcdi completion, or wait for an mcdi_event.
+ *	Serialised by @lock
+ * @seqno: The next sequence number to use for mcdi requests.
+ *	Serialised by @lock
+ * @credits: Number of spurious MCDI completion events allowed before we
+ *	trigger a fatal error. Protected by @lock
+ * @resprc: Returned MCDI completion
+ * @resplen: Returned payload length
+ */
+struct efx_mcdi_iface {
+	atomic_t state;
+	wait_queue_head_t wq;
+	spinlock_t iface_lock;
+	enum efx_mcdi_mode mode;
+	unsigned int credits;
+	unsigned int seqno;
+	unsigned int resprc;
+	size_t resplen;
+};
+
+extern void efx_mcdi_init(struct efx_nic *efx);
+
+extern int efx_mcdi_rpc(struct efx_nic *efx, unsigned cmd, const u8 *inbuf,
+			size_t inlen, u8 *outbuf, size_t outlen,
+			size_t *outlen_actual);
+
+extern int efx_mcdi_poll_reboot(struct efx_nic *efx);
+extern void efx_mcdi_mode_poll(struct efx_nic *efx);
+extern void efx_mcdi_mode_event(struct efx_nic *efx);
+
+extern void efx_mcdi_process_event(struct efx_channel *channel,
+				   efx_qword_t *event);
+
+#define MCDI_PTR2(_buf, _ofst)						\
+	(((u8 *)_buf) + _ofst)
+#define MCDI_SET_DWORD2(_buf, _ofst, _value)				\
+	EFX_POPULATE_DWORD_1(*((efx_dword_t *)MCDI_PTR2(_buf, _ofst)),	\
+			     EFX_DWORD_0, _value)
+#define MCDI_DWORD2(_buf, _ofst)					\
+	EFX_DWORD_FIELD(*((efx_dword_t *)MCDI_PTR2(_buf, _ofst)),	\
+			EFX_DWORD_0)
+#define MCDI_QWORD2(_buf, _ofst)					\
+	EFX_QWORD_FIELD64(*((efx_qword_t *)MCDI_PTR2(_buf, _ofst)),	\
+			  EFX_QWORD_0)
+
+#define MCDI_PTR(_buf, _ofst)						\
+	MCDI_PTR2(_buf, MC_CMD_ ## _ofst ## _OFST)
+#define MCDI_SET_DWORD(_buf, _ofst, _value)				\
+	MCDI_SET_DWORD2(_buf, MC_CMD_ ## _ofst ## _OFST, _value)
+#define MCDI_DWORD(_buf, _ofst)						\
+	MCDI_DWORD2(_buf, MC_CMD_ ## _ofst ## _OFST)
+#define MCDI_QWORD(_buf, _ofst)						\
+	MCDI_QWORD2(_buf, MC_CMD_ ## _ofst ## _OFST)
+
+#define MCDI_EVENT_FIELD(_ev, _field)			\
+	EFX_QWORD_FIELD(_ev, MCDI_EVENT_ ## _field)
+
+extern int efx_mcdi_fwver(struct efx_nic *efx, u64 *version, u32 *build);
+extern int efx_mcdi_drv_attach(struct efx_nic *efx, bool driver_operating,
+			       bool *was_attached_out);
+extern int efx_mcdi_get_board_cfg(struct efx_nic *efx, u8 *mac_address,
+				  u16 *fw_subtype_list);
+extern int efx_mcdi_log_ctrl(struct efx_nic *efx, bool evq, bool uart,
+			     u32 dest_evq);
+extern int efx_mcdi_nvram_types(struct efx_nic *efx, u32 *nvram_types_out);
+extern int efx_mcdi_nvram_info(struct efx_nic *efx, unsigned int type,
+			       size_t *size_out, size_t *erase_size_out,
+			       bool *protected_out);
+extern int efx_mcdi_nvram_update_start(struct efx_nic *efx,
+				       unsigned int type);
+extern int efx_mcdi_nvram_read(struct efx_nic *efx, unsigned int type,
+			       loff_t offset, u8 *buffer, size_t length);
+extern int efx_mcdi_nvram_write(struct efx_nic *efx, unsigned int type,
+				loff_t offset, const u8 *buffer,
+				size_t length);
+extern int efx_mcdi_nvram_erase(struct efx_nic *efx, unsigned int type,
+				loff_t offset, size_t length);
+extern int efx_mcdi_nvram_update_finish(struct efx_nic *efx,
+					unsigned int type);
+extern int efx_mcdi_handle_assertion(struct efx_nic *efx);
+extern void efx_mcdi_set_id_led(struct efx_nic *efx, enum efx_led_mode mode);
+extern int efx_mcdi_reset_port(struct efx_nic *efx);
+extern int efx_mcdi_reset_mc(struct efx_nic *efx);
+extern int efx_mcdi_wol_filter_set(struct efx_nic *efx, u32 type,
+				   const u8 *mac, int *id_out);
+extern int efx_mcdi_wol_filter_set_magic(struct efx_nic *efx,
+					 const u8 *mac, int *id_out);
+extern int efx_mcdi_wol_filter_get_magic(struct efx_nic *efx, int *id_out);
+extern int efx_mcdi_wol_filter_remove(struct efx_nic *efx, int id);
+extern int efx_mcdi_wol_filter_reset(struct efx_nic *efx);
+
+#endif /* EFX_MCDI_H */
diff --git a/drivers/net/sfc/mcdi_mac.c b/drivers/net/sfc/mcdi_mac.c
new file mode 100644
index 0000000..06d24a1
--- /dev/null
+++ b/drivers/net/sfc/mcdi_mac.c
@@ -0,0 +1,152 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2009 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#include "net_driver.h"
+#include "efx.h"
+#include "mac.h"
+#include "mcdi.h"
+#include "mcdi_pcol.h"
+
+static int efx_mcdi_set_mac(struct efx_nic *efx)
+{
+	u32 reject, fcntl;
+	u8 cmdbytes[MC_CMD_SET_MAC_IN_LEN];
+
+	memcpy(cmdbytes + MC_CMD_SET_MAC_IN_ADDR_OFST,
+	       efx->net_dev->dev_addr, ETH_ALEN);
+
+	MCDI_SET_DWORD(cmdbytes, SET_MAC_IN_MTU,
+			EFX_MAX_FRAME_LEN(efx->net_dev->mtu));
+	MCDI_SET_DWORD(cmdbytes, SET_MAC_IN_DRAIN, 0);
+
+	/* The MCDI command provides for controlling accept/reject
+	 * of broadcast packets too, but the driver doesn't currently
+	 * expose this. */
+	reject = (efx->promiscuous) ? 0 :
+		(1 << MC_CMD_SET_MAC_IN_REJECT_UNCST_LBN);
+	MCDI_SET_DWORD(cmdbytes, SET_MAC_IN_REJECT, reject);
+
+	switch (efx->wanted_fc) {
+	case EFX_FC_RX | EFX_FC_TX:
+		fcntl = MC_CMD_FCNTL_BIDIR;
+		break;
+	case EFX_FC_RX:
+		fcntl = MC_CMD_FCNTL_RESPOND;
+		break;
+	default:
+		fcntl = MC_CMD_FCNTL_OFF;
+		break;
+	}
+	if (efx->wanted_fc & EFX_FC_AUTO)
+		fcntl = MC_CMD_FCNTL_AUTO;
+
+	MCDI_SET_DWORD(cmdbytes, SET_MAC_IN_FCNTL, fcntl);
+
+	return efx_mcdi_rpc(efx, MC_CMD_SET_MAC, cmdbytes, sizeof(cmdbytes),
+			    NULL, 0, NULL);
+}
+
+static int efx_mcdi_get_mac_faults(struct efx_nic *efx, u32 *faults)
+{
+	u8 outbuf[MC_CMD_GET_LINK_OUT_LEN];
+	size_t outlength;
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_GET_LINK_IN_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_LINK, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlength);
+	if (rc)
+		goto fail;
+
+	*faults = MCDI_DWORD(outbuf, GET_LINK_OUT_MAC_FAULT);
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n",
+		__func__, rc);
+	return rc;
+}
+
+int efx_mcdi_mac_stats(struct efx_nic *efx, dma_addr_t dma_addr,
+		       u32 dma_len, int enable, int clear)
+{
+	u8 inbuf[MC_CMD_MAC_STATS_IN_LEN];
+	int rc;
+	efx_dword_t *cmd_ptr;
+	int period = 1000;
+	u32 addr_hi;
+	u32 addr_lo;
+
+	BUILD_BUG_ON(MC_CMD_MAC_STATS_OUT_LEN != 0);
+
+	addr_lo = ((u64)dma_addr) >> 0;
+	addr_hi = ((u64)dma_addr) >> 32;
+
+	MCDI_SET_DWORD(inbuf, MAC_STATS_IN_DMA_ADDR_LO, addr_lo);
+	MCDI_SET_DWORD(inbuf, MAC_STATS_IN_DMA_ADDR_HI, addr_hi);
+	cmd_ptr = (efx_dword_t *)MCDI_PTR(inbuf, MAC_STATS_IN_CMD);
+	if (enable)
+		EFX_POPULATE_DWORD_6(*cmd_ptr,
+				     MC_CMD_MAC_STATS_CMD_DMA, 1,
+				     MC_CMD_MAC_STATS_CMD_CLEAR, clear,
+				     MC_CMD_MAC_STATS_CMD_PERIODIC_CHANGE, 1,
+				     MC_CMD_MAC_STATS_CMD_PERIODIC_ENABLE, 1,
+				     MC_CMD_MAC_STATS_CMD_PERIODIC_CLEAR, 0,
+				     MC_CMD_MAC_STATS_CMD_PERIOD_MS, period);
+	else
+		EFX_POPULATE_DWORD_5(*cmd_ptr,
+				     MC_CMD_MAC_STATS_CMD_DMA, 0,
+				     MC_CMD_MAC_STATS_CMD_CLEAR, clear,
+				     MC_CMD_MAC_STATS_CMD_PERIODIC_CHANGE, 1,
+				     MC_CMD_MAC_STATS_CMD_PERIODIC_ENABLE, 0,
+				     MC_CMD_MAC_STATS_CMD_PERIODIC_CLEAR, 0);
+	MCDI_SET_DWORD(inbuf, MAC_STATS_IN_DMA_LEN, dma_len);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_MAC_STATS, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: %s failed rc=%d\n",
+		__func__, enable ? "enable" : "disable", rc);
+	return rc;
+}
+
+static int efx_mcdi_mac_reconfigure(struct efx_nic *efx)
+{
+	int rc;
+
+	rc = efx_mcdi_set_mac(efx);
+	if (rc != 0)
+		return rc;
+
+	/* Restore the multicast hash registers. */
+	efx->type->push_multicast_hash(efx);
+
+	return 0;
+}
+
+
+static bool efx_mcdi_mac_check_fault(struct efx_nic *efx)
+{
+	u32 faults;
+	int rc = efx_mcdi_get_mac_faults(efx, &faults);
+	return (rc != 0) || (faults != 0);
+}
+
+
+struct efx_mac_operations efx_mcdi_mac_operations = {
+	.reconfigure	= efx_mcdi_mac_reconfigure,
+	.update_stats	= efx_port_dummy_op_void,
+	.check_fault 	= efx_mcdi_mac_check_fault,
+};
diff --git a/drivers/net/sfc/mcdi_phy.c b/drivers/net/sfc/mcdi_phy.c
new file mode 100644
index 0000000..0e1bcc5
--- /dev/null
+++ b/drivers/net/sfc/mcdi_phy.c
@@ -0,0 +1,597 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2009 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+/*
+ * Driver for PHY related operations via MCDI.
+ */
+
+#include "efx.h"
+#include "phy.h"
+#include "mcdi.h"
+#include "mcdi_pcol.h"
+#include "mdio_10g.h"
+
+struct efx_mcdi_phy_cfg {
+	u32 flags;
+	u32 type;
+	u32 supported_cap;
+	u32 channel;
+	u32 port;
+	u32 stats_mask;
+	u8 name[20];
+	u32 media;
+	u32 mmd_mask;
+	u8 revision[20];
+	u32 forced_cap;
+};
+
+static int
+efx_mcdi_get_phy_cfg(struct efx_nic *efx, struct efx_mcdi_phy_cfg *cfg)
+{
+	u8 outbuf[MC_CMD_GET_PHY_CFG_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_GET_PHY_CFG_IN_LEN != 0);
+	BUILD_BUG_ON(MC_CMD_GET_PHY_CFG_OUT_NAME_LEN != sizeof(cfg->name));
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_PHY_CFG, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	if (outlen < MC_CMD_GET_PHY_CFG_OUT_LEN) {
+		rc = -EMSGSIZE;
+		goto fail;
+	}
+
+	cfg->flags = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_FLAGS);
+	cfg->type = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_TYPE);
+	cfg->supported_cap =
+		MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_SUPPORTED_CAP);
+	cfg->channel = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_CHANNEL);
+	cfg->port = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_PRT);
+	cfg->stats_mask = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_STATS_MASK);
+	memcpy(cfg->name, MCDI_PTR(outbuf, GET_PHY_CFG_OUT_NAME),
+	       sizeof(cfg->name));
+	cfg->media = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_MEDIA_TYPE);
+	cfg->mmd_mask = MCDI_DWORD(outbuf, GET_PHY_CFG_OUT_MMD_MASK);
+	memcpy(cfg->revision, MCDI_PTR(outbuf, GET_PHY_CFG_OUT_REVISION),
+	       sizeof(cfg->revision));
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+static int efx_mcdi_set_link(struct efx_nic *efx, u32 capabilities,
+			     u32 flags, u32 loopback_mode,
+			     u32 loopback_speed)
+{
+	u8 inbuf[MC_CMD_SET_LINK_IN_LEN];
+	int rc;
+
+	BUILD_BUG_ON(MC_CMD_SET_LINK_OUT_LEN != 0);
+
+	MCDI_SET_DWORD(inbuf, SET_LINK_IN_CAP, capabilities);
+	MCDI_SET_DWORD(inbuf, SET_LINK_IN_FLAGS, flags);
+	MCDI_SET_DWORD(inbuf, SET_LINK_IN_LOOPBACK_MODE, loopback_mode);
+	MCDI_SET_DWORD(inbuf, SET_LINK_IN_LOOPBACK_SPEED, loopback_speed);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_SET_LINK, inbuf, sizeof(inbuf),
+			  NULL, 0, NULL);
+	if (rc)
+		goto fail;
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+static int efx_mcdi_loopback_modes(struct efx_nic *efx, u64 *loopback_modes)
+{
+	u8 outbuf[MC_CMD_GET_LOOPBACK_MODES_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_LOOPBACK_MODES, NULL, 0,
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	if (outlen < MC_CMD_GET_LOOPBACK_MODES_OUT_LEN) {
+		rc = -EMSGSIZE;
+		goto fail;
+	}
+
+	*loopback_modes = MCDI_QWORD(outbuf, GET_LOOPBACK_MODES_SUGGESTED);
+
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_mdio_read(struct efx_nic *efx, unsigned int bus,
+			 unsigned int prtad, unsigned int devad, u16 addr,
+			 u16 *value_out, u32 *status_out)
+{
+	u8 inbuf[MC_CMD_MDIO_READ_IN_LEN];
+	u8 outbuf[MC_CMD_MDIO_READ_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, MDIO_READ_IN_BUS, bus);
+	MCDI_SET_DWORD(inbuf, MDIO_READ_IN_PRTAD, prtad);
+	MCDI_SET_DWORD(inbuf, MDIO_READ_IN_DEVAD, devad);
+	MCDI_SET_DWORD(inbuf, MDIO_READ_IN_ADDR, addr);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_MDIO_READ, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	*value_out = (u16)MCDI_DWORD(outbuf, MDIO_READ_OUT_VALUE);
+	*status_out = MCDI_DWORD(outbuf, MDIO_READ_OUT_STATUS);
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+int efx_mcdi_mdio_write(struct efx_nic *efx, unsigned int bus,
+			  unsigned int prtad, unsigned int devad, u16 addr,
+			  u16 value, u32 *status_out)
+{
+	u8 inbuf[MC_CMD_MDIO_WRITE_IN_LEN];
+	u8 outbuf[MC_CMD_MDIO_WRITE_OUT_LEN];
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, MDIO_WRITE_IN_BUS, bus);
+	MCDI_SET_DWORD(inbuf, MDIO_WRITE_IN_PRTAD, prtad);
+	MCDI_SET_DWORD(inbuf, MDIO_WRITE_IN_DEVAD, devad);
+	MCDI_SET_DWORD(inbuf, MDIO_WRITE_IN_ADDR, addr);
+	MCDI_SET_DWORD(inbuf, MDIO_WRITE_IN_VALUE, value);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_MDIO_WRITE, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		goto fail;
+
+	*status_out = MCDI_DWORD(outbuf, MDIO_WRITE_OUT_STATUS);
+	return 0;
+
+fail:
+	EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+	return rc;
+}
+
+static u32 mcdi_to_ethtool_cap(u32 media, u32 cap)
+{
+	u32 result = 0;
+
+	switch (media) {
+	case MC_CMD_MEDIA_KX4:
+		result |= SUPPORTED_Backplane;
+		if (cap & (1 << MC_CMD_PHY_CAP_1000FDX_LBN))
+			result |= SUPPORTED_1000baseKX_Full;
+		if (cap & (1 << MC_CMD_PHY_CAP_10000FDX_LBN))
+			result |= SUPPORTED_10000baseKX4_Full;
+		break;
+
+	case MC_CMD_MEDIA_XFP:
+	case MC_CMD_MEDIA_SFP_PLUS:
+		result |= SUPPORTED_FIBRE;
+		break;
+
+	case MC_CMD_MEDIA_BASE_T:
+		result |= SUPPORTED_TP;
+		if (cap & (1 << MC_CMD_PHY_CAP_10HDX_LBN))
+			result |= SUPPORTED_10baseT_Half;
+		if (cap & (1 << MC_CMD_PHY_CAP_10FDX_LBN))
+			result |= SUPPORTED_10baseT_Full;
+		if (cap & (1 << MC_CMD_PHY_CAP_100HDX_LBN))
+			result |= SUPPORTED_100baseT_Half;
+		if (cap & (1 << MC_CMD_PHY_CAP_100FDX_LBN))
+			result |= SUPPORTED_100baseT_Full;
+		if (cap & (1 << MC_CMD_PHY_CAP_1000HDX_LBN))
+			result |= SUPPORTED_1000baseT_Half;
+		if (cap & (1 << MC_CMD_PHY_CAP_1000FDX_LBN))
+			result |= SUPPORTED_1000baseT_Full;
+		if (cap & (1 << MC_CMD_PHY_CAP_10000FDX_LBN))
+			result |= SUPPORTED_10000baseT_Full;
+		break;
+	}
+
+	if (cap & (1 << MC_CMD_PHY_CAP_PAUSE_LBN))
+		result |= SUPPORTED_Pause;
+	if (cap & (1 << MC_CMD_PHY_CAP_ASYM_LBN))
+		result |= SUPPORTED_Asym_Pause;
+	if (cap & (1 << MC_CMD_PHY_CAP_AN_LBN))
+		result |= SUPPORTED_Autoneg;
+
+	return result;
+}
+
+static u32 ethtool_to_mcdi_cap(u32 cap)
+{
+	u32 result = 0;
+
+	if (cap & SUPPORTED_10baseT_Half)
+		result |= (1 << MC_CMD_PHY_CAP_10HDX_LBN);
+	if (cap & SUPPORTED_10baseT_Full)
+		result |= (1 << MC_CMD_PHY_CAP_10FDX_LBN);
+	if (cap & SUPPORTED_100baseT_Half)
+		result |= (1 << MC_CMD_PHY_CAP_100HDX_LBN);
+	if (cap & SUPPORTED_100baseT_Full)
+		result |= (1 << MC_CMD_PHY_CAP_100FDX_LBN);
+	if (cap & SUPPORTED_1000baseT_Half)
+		result |= (1 << MC_CMD_PHY_CAP_1000HDX_LBN);
+	if (cap & (SUPPORTED_1000baseT_Full | SUPPORTED_1000baseKX_Full))
+		result |= (1 << MC_CMD_PHY_CAP_1000FDX_LBN);
+	if (cap & (SUPPORTED_10000baseT_Full | SUPPORTED_10000baseKX4_Full))
+		result |= (1 << MC_CMD_PHY_CAP_10000FDX_LBN);
+	if (cap & SUPPORTED_Pause)
+		result |= (1 << MC_CMD_PHY_CAP_PAUSE_LBN);
+	if (cap & SUPPORTED_Asym_Pause)
+		result |= (1 << MC_CMD_PHY_CAP_ASYM_LBN);
+	if (cap & SUPPORTED_Autoneg)
+		result |= (1 << MC_CMD_PHY_CAP_AN_LBN);
+
+	return result;
+}
+
+static u32 efx_get_mcdi_phy_flags(struct efx_nic *efx)
+{
+	struct efx_mcdi_phy_cfg *phy_cfg = efx->phy_data;
+	enum efx_phy_mode mode, supported;
+	u32 flags;
+
+	/* TODO: Advertise the capabilities supported by this PHY */
+	supported = 0;
+	if (phy_cfg->flags & (1 << MC_CMD_GET_PHY_CFG_TXDIS_LBN))
+		supported |= PHY_MODE_TX_DISABLED;
+	if (phy_cfg->flags & (1 << MC_CMD_GET_PHY_CFG_LOWPOWER_LBN))
+		supported |= PHY_MODE_LOW_POWER;
+	if (phy_cfg->flags & (1 << MC_CMD_GET_PHY_CFG_POWEROFF_LBN))
+		supported |= PHY_MODE_OFF;
+
+	mode = efx->phy_mode & supported;
+
+	flags = 0;
+	if (mode & PHY_MODE_TX_DISABLED)
+		flags |= (1 << MC_CMD_SET_LINK_TXDIS_LBN);
+	if (mode & PHY_MODE_LOW_POWER)
+		flags |= (1 << MC_CMD_SET_LINK_LOWPOWER_LBN);
+	if (mode & PHY_MODE_OFF)
+		flags |= (1 << MC_CMD_SET_LINK_POWEROFF_LBN);
+
+	return flags;
+}
+
+static u32 mcdi_to_ethtool_media(u32 media)
+{
+	switch (media) {
+	case MC_CMD_MEDIA_XAUI:
+	case MC_CMD_MEDIA_CX4:
+	case MC_CMD_MEDIA_KX4:
+		return PORT_OTHER;
+
+	case MC_CMD_MEDIA_XFP:
+	case MC_CMD_MEDIA_SFP_PLUS:
+		return PORT_FIBRE;
+
+	case MC_CMD_MEDIA_BASE_T:
+		return PORT_TP;
+
+	default:
+		return PORT_OTHER;
+	}
+}
+
+static int efx_mcdi_phy_probe(struct efx_nic *efx)
+{
+	struct efx_mcdi_phy_cfg *phy_cfg;
+	int rc;
+
+	/* TODO: Move phy_data initialisation to
+	 * phy_op->probe/remove, rather than init/fini */
+	phy_cfg = kzalloc(sizeof(*phy_cfg), GFP_KERNEL);
+	if (phy_cfg == NULL) {
+		rc = -ENOMEM;
+		goto fail_alloc;
+	}
+	rc = efx_mcdi_get_phy_cfg(efx, phy_cfg);
+	if (rc != 0)
+		goto fail;
+
+	efx->phy_type = phy_cfg->type;
+
+	efx->mdio_bus = phy_cfg->channel;
+	efx->mdio.prtad = phy_cfg->port;
+	efx->mdio.mmds = phy_cfg->mmd_mask & ~(1 << MC_CMD_MMD_CLAUSE22);
+	efx->mdio.mode_support = 0;
+	if (phy_cfg->mmd_mask & (1 << MC_CMD_MMD_CLAUSE22))
+		efx->mdio.mode_support |= MDIO_SUPPORTS_C22;
+	if (phy_cfg->mmd_mask & ~(1 << MC_CMD_MMD_CLAUSE22))
+		efx->mdio.mode_support |= MDIO_SUPPORTS_C45 | MDIO_EMULATE_C22;
+
+	/* Assert that we can map efx -> mcdi loopback modes */
+	BUILD_BUG_ON(LOOPBACK_NONE != MC_CMD_LOOPBACK_NONE);
+	BUILD_BUG_ON(LOOPBACK_DATA != MC_CMD_LOOPBACK_DATA);
+	BUILD_BUG_ON(LOOPBACK_GMAC != MC_CMD_LOOPBACK_GMAC);
+	BUILD_BUG_ON(LOOPBACK_XGMII != MC_CMD_LOOPBACK_XGMII);
+	BUILD_BUG_ON(LOOPBACK_XGXS != MC_CMD_LOOPBACK_XGXS);
+	BUILD_BUG_ON(LOOPBACK_XAUI != MC_CMD_LOOPBACK_XAUI);
+	BUILD_BUG_ON(LOOPBACK_GMII != MC_CMD_LOOPBACK_GMII);
+	BUILD_BUG_ON(LOOPBACK_SGMII != MC_CMD_LOOPBACK_SGMII);
+	BUILD_BUG_ON(LOOPBACK_XGBR != MC_CMD_LOOPBACK_XGBR);
+	BUILD_BUG_ON(LOOPBACK_XFI != MC_CMD_LOOPBACK_XFI);
+	BUILD_BUG_ON(LOOPBACK_XAUI_FAR != MC_CMD_LOOPBACK_XAUI_FAR);
+	BUILD_BUG_ON(LOOPBACK_GMII_FAR != MC_CMD_LOOPBACK_GMII_FAR);
+	BUILD_BUG_ON(LOOPBACK_SGMII_FAR != MC_CMD_LOOPBACK_SGMII_FAR);
+	BUILD_BUG_ON(LOOPBACK_XFI_FAR != MC_CMD_LOOPBACK_XFI_FAR);
+	BUILD_BUG_ON(LOOPBACK_GPHY != MC_CMD_LOOPBACK_GPHY);
+	BUILD_BUG_ON(LOOPBACK_PHYXS != MC_CMD_LOOPBACK_PHYXS);
+	BUILD_BUG_ON(LOOPBACK_PCS != MC_CMD_LOOPBACK_PCS);
+	BUILD_BUG_ON(LOOPBACK_PMAPMD != MC_CMD_LOOPBACK_PMAPMD);
+	BUILD_BUG_ON(LOOPBACK_XPORT != MC_CMD_LOOPBACK_XPORT);
+	BUILD_BUG_ON(LOOPBACK_XGMII_WS != MC_CMD_LOOPBACK_XGMII_WS);
+	BUILD_BUG_ON(LOOPBACK_XAUI_WS != MC_CMD_LOOPBACK_XAUI_WS);
+	BUILD_BUG_ON(LOOPBACK_XAUI_WS_FAR != MC_CMD_LOOPBACK_XAUI_WS_FAR);
+	BUILD_BUG_ON(LOOPBACK_XAUI_WS_NEAR != MC_CMD_LOOPBACK_XAUI_WS_NEAR);
+	BUILD_BUG_ON(LOOPBACK_GMII_WS != MC_CMD_LOOPBACK_GMII_WS);
+	BUILD_BUG_ON(LOOPBACK_XFI_WS != MC_CMD_LOOPBACK_XFI_WS);
+	BUILD_BUG_ON(LOOPBACK_XFI_WS_FAR != MC_CMD_LOOPBACK_XFI_WS_FAR);
+	BUILD_BUG_ON(LOOPBACK_PHYXS_WS != MC_CMD_LOOPBACK_PHYXS_WS);
+
+	rc = efx_mcdi_loopback_modes(efx, &efx->loopback_modes);
+	if (rc != 0)
+		goto fail;
+	/* The MC indicates that LOOPBACK_NONE is a valid loopback mode,
+	 * but by convention we don't */
+	efx->loopback_modes &= ~(1 << LOOPBACK_NONE);
+
+	kfree(phy_cfg);
+
+	return 0;
+
+fail:
+	kfree(phy_cfg);
+fail_alloc:
+	return rc;
+}
+
+static int efx_mcdi_phy_init(struct efx_nic *efx)
+{
+	struct efx_mcdi_phy_cfg *phy_data;
+	u8 outbuf[MC_CMD_GET_LINK_OUT_LEN];
+	u32 caps;
+	int rc;
+
+	phy_data = kzalloc(sizeof(*phy_data), GFP_KERNEL);
+	if (phy_data == NULL)
+		return -ENOMEM;
+
+	rc = efx_mcdi_get_phy_cfg(efx, phy_data);
+	if (rc != 0)
+		goto fail;
+
+	efx->phy_data = phy_data;
+
+	BUILD_BUG_ON(MC_CMD_GET_LINK_IN_LEN != 0);
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_LINK, NULL, 0,
+			  outbuf, sizeof(outbuf), NULL);
+	if (rc)
+		goto fail;
+
+	caps = MCDI_DWORD(outbuf, GET_LINK_OUT_CAP);
+	if (caps & (1 << MC_CMD_PHY_CAP_AN_LBN))
+		efx->link_advertising =
+			mcdi_to_ethtool_cap(phy_data->media, caps);
+	else
+		phy_data->forced_cap = caps;
+
+	return 0;
+
+fail:
+	kfree(phy_data);
+	return rc;
+}
+
+int efx_mcdi_phy_reconfigure(struct efx_nic *efx)
+{
+	struct efx_mcdi_phy_cfg *phy_cfg = efx->phy_data;
+	u32 caps = (efx->link_advertising ?
+		    ethtool_to_mcdi_cap(efx->link_advertising) :
+		    phy_cfg->forced_cap);
+
+	return efx_mcdi_set_link(efx, caps, efx_get_mcdi_phy_flags(efx),
+				 efx->loopback_mode, 0);
+}
+
+void efx_mcdi_phy_decode_link(struct efx_nic *efx,
+			      struct efx_link_state *link_state,
+			      u32 speed, u32 flags, u32 fcntl)
+{
+	switch (fcntl) {
+	case MC_CMD_FCNTL_AUTO:
+		WARN_ON(1);	/* This is not a link mode */
+		link_state->fc = EFX_FC_AUTO | EFX_FC_TX | EFX_FC_RX;
+		break;
+	case MC_CMD_FCNTL_BIDIR:
+		link_state->fc = EFX_FC_TX | EFX_FC_RX;
+		break;
+	case MC_CMD_FCNTL_RESPOND:
+		link_state->fc = EFX_FC_RX;
+		break;
+	default:
+		WARN_ON(1);
+	case MC_CMD_FCNTL_OFF:
+		link_state->fc = 0;
+		break;
+	}
+
+	link_state->up = !!(flags & (1 << MC_CMD_GET_LINK_LINK_UP_LBN));
+	link_state->fd = !!(flags & (1 << MC_CMD_GET_LINK_FULL_DUPLEX_LBN));
+	link_state->speed = speed;
+}
+
+/* Verify that the forced flow control settings (!EFX_FC_AUTO) are
+ * supported by the link partner. Warn the user if this isn't the case
+ */
+void efx_mcdi_phy_check_fcntl(struct efx_nic *efx, u32 lpa)
+{
+	struct efx_mcdi_phy_cfg *phy_cfg = efx->phy_data;
+	u32 rmtadv;
+
+	/* The link partner capabilities are only relevent if the
+	 * link supports flow control autonegotiation */
+	if (~phy_cfg->supported_cap & (1 << MC_CMD_PHY_CAP_ASYM_LBN))
+		return;
+
+	/* If flow control autoneg is supported and enabled, then fine */
+	if (efx->wanted_fc & EFX_FC_AUTO)
+		return;
+
+	rmtadv = 0;
+	if (lpa & (1 << MC_CMD_PHY_CAP_PAUSE_LBN))
+		rmtadv |= ADVERTISED_Pause;
+	if (lpa & (1 << MC_CMD_PHY_CAP_ASYM_LBN))
+		rmtadv |=  ADVERTISED_Asym_Pause;
+
+	if ((efx->wanted_fc & EFX_FC_TX) && rmtadv == ADVERTISED_Asym_Pause)
+		EFX_ERR(efx, "warning: link partner doesn't support "
+			"pause frames");
+}
+
+static bool efx_mcdi_phy_poll(struct efx_nic *efx)
+{
+	struct efx_link_state old_state = efx->link_state;
+	u8 outbuf[MC_CMD_GET_LINK_OUT_LEN];
+	int rc;
+
+	WARN_ON(!mutex_is_locked(&efx->mac_lock));
+
+	BUILD_BUG_ON(MC_CMD_GET_LINK_IN_LEN != 0);
+
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_LINK, NULL, 0,
+			  outbuf, sizeof(outbuf), NULL);
+	if (rc) {
+		EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+		efx->link_state.up = false;
+	} else {
+		efx_mcdi_phy_decode_link(
+			efx, &efx->link_state,
+			MCDI_DWORD(outbuf, GET_LINK_OUT_LINK_SPEED),
+			MCDI_DWORD(outbuf, GET_LINK_OUT_FLAGS),
+			MCDI_DWORD(outbuf, GET_LINK_OUT_FCNTL));
+	}
+
+	return !efx_link_state_equal(&efx->link_state, &old_state);
+}
+
+static void efx_mcdi_phy_fini(struct efx_nic *efx)
+{
+	struct efx_mcdi_phy_data *phy_data = efx->phy_data;
+
+	efx->phy_data = NULL;
+	kfree(phy_data);
+}
+
+static void efx_mcdi_phy_get_settings(struct efx_nic *efx, struct ethtool_cmd *ecmd)
+{
+	struct efx_mcdi_phy_cfg *phy_cfg = efx->phy_data;
+	u8 outbuf[MC_CMD_GET_LINK_OUT_LEN];
+	int rc;
+
+	ecmd->supported =
+		mcdi_to_ethtool_cap(phy_cfg->media, phy_cfg->supported_cap);
+	ecmd->advertising = efx->link_advertising;
+	ecmd->speed = efx->link_state.speed;
+	ecmd->duplex = efx->link_state.fd;
+	ecmd->port = mcdi_to_ethtool_media(phy_cfg->media);
+	ecmd->phy_address = phy_cfg->port;
+	ecmd->transceiver = XCVR_INTERNAL;
+	ecmd->autoneg = !!(efx->link_advertising & ADVERTISED_Autoneg);
+	ecmd->mdio_support = (efx->mdio.mode_support &
+			      (MDIO_SUPPORTS_C45 | MDIO_SUPPORTS_C22));
+
+	BUILD_BUG_ON(MC_CMD_GET_LINK_IN_LEN != 0);
+	rc = efx_mcdi_rpc(efx, MC_CMD_GET_LINK, NULL, 0,
+			  outbuf, sizeof(outbuf), NULL);
+	if (rc) {
+		EFX_ERR(efx, "%s: failed rc=%d\n", __func__, rc);
+		return;
+	}
+	ecmd->lp_advertising =
+		mcdi_to_ethtool_cap(phy_cfg->media,
+				    MCDI_DWORD(outbuf, GET_LINK_OUT_LP_CAP));
+}
+
+static int efx_mcdi_phy_set_settings(struct efx_nic *efx, struct ethtool_cmd *ecmd)
+{
+	struct efx_mcdi_phy_cfg *phy_cfg = efx->phy_data;
+	u32 caps;
+	int rc;
+
+	if (ecmd->autoneg) {
+		caps = (ethtool_to_mcdi_cap(ecmd->advertising) |
+			 1 << MC_CMD_PHY_CAP_AN_LBN);
+	} else if (ecmd->duplex) {
+		switch (ecmd->speed) {
+		case 10:    caps = 1 << MC_CMD_PHY_CAP_10FDX_LBN;    break;
+		case 100:   caps = 1 << MC_CMD_PHY_CAP_100FDX_LBN;   break;
+		case 1000:  caps = 1 << MC_CMD_PHY_CAP_1000FDX_LBN;  break;
+		case 10000: caps = 1 << MC_CMD_PHY_CAP_10000FDX_LBN; break;
+		default:    return -EINVAL;
+		}
+	} else {
+		switch (ecmd->speed) {
+		case 10:    caps = 1 << MC_CMD_PHY_CAP_10HDX_LBN;    break;
+		case 100:   caps = 1 << MC_CMD_PHY_CAP_100HDX_LBN;   break;
+		case 1000:  caps = 1 << MC_CMD_PHY_CAP_1000HDX_LBN;  break;
+		default:    return -EINVAL;
+		}
+	}
+
+	rc = efx_mcdi_set_link(efx, caps, efx_get_mcdi_phy_flags(efx),
+			       efx->loopback_mode, 0);
+	if (rc)
+		return rc;
+
+	if (ecmd->autoneg) {
+		efx_link_set_advertising(
+			efx, ecmd->advertising | ADVERTISED_Autoneg);
+		phy_cfg->forced_cap = 0;
+	} else {
+		efx_link_set_advertising(efx, 0);
+		phy_cfg->forced_cap = caps;
+	}
+	return 0;
+}
+
+struct efx_phy_operations efx_mcdi_phy_ops = {
+	.probe		= efx_mcdi_phy_probe,
+	.init 	 	= efx_mcdi_phy_init,
+	.reconfigure	= efx_mcdi_phy_reconfigure,
+	.poll		= efx_mcdi_phy_poll,
+	.fini		= efx_mcdi_phy_fini,
+	.get_settings	= efx_mcdi_phy_get_settings,
+	.set_settings	= efx_mcdi_phy_set_settings,
+	.run_tests	= NULL,
+	.test_name	= NULL,
+};
diff --git a/drivers/net/sfc/siena.c b/drivers/net/sfc/siena.c
new file mode 100644
index 0000000..de07a4f
--- /dev/null
+++ b/drivers/net/sfc/siena.c
@@ -0,0 +1,604 @@
+/****************************************************************************
+ * Driver for Solarflare Solarstorm network controllers and boards
+ * Copyright 2005-2006 Fen Systems Ltd.
+ * Copyright 2006-2009 Solarflare Communications Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/module.h>
+#include "net_driver.h"
+#include "bitfield.h"
+#include "efx.h"
+#include "nic.h"
+#include "mac.h"
+#include "spi.h"
+#include "regs.h"
+#include "io.h"
+#include "phy.h"
+#include "workarounds.h"
+#include "mcdi.h"
+#include "mcdi_pcol.h"
+
+/* Hardware control for SFC9000 family including SFL9021 (aka Siena). */
+
+static void siena_init_wol(struct efx_nic *efx);
+
+
+static void siena_push_irq_moderation(struct efx_channel *channel)
+{
+	efx_dword_t timer_cmd;
+
+	if (channel->irq_moderation)
+		EFX_POPULATE_DWORD_2(timer_cmd,
+				     FRF_CZ_TC_TIMER_MODE,
+				     FFE_CZ_TIMER_MODE_INT_HLDOFF,
+				     FRF_CZ_TC_TIMER_VAL,
+				     channel->irq_moderation - 1);
+	else
+		EFX_POPULATE_DWORD_2(timer_cmd,
+				     FRF_CZ_TC_TIMER_MODE,
+				     FFE_CZ_TIMER_MODE_DIS,
+				     FRF_CZ_TC_TIMER_VAL, 0);
+	efx_writed_page_locked(channel->efx, &timer_cmd, FR_BZ_TIMER_COMMAND_P0,
+			       channel->channel);
+}
+
+static void siena_push_multicast_hash(struct efx_nic *efx)
+{
+	WARN_ON(!mutex_is_locked(&efx->mac_lock));
+
+	efx_mcdi_rpc(efx, MC_CMD_SET_MCAST_HASH,
+		     efx->multicast_hash.byte, sizeof(efx->multicast_hash),
+		     NULL, 0, NULL);
+}
+
+static int siena_mdio_write(struct net_device *net_dev,
+			    int prtad, int devad, u16 addr, u16 value)
+{
+	struct efx_nic *efx = netdev_priv(net_dev);
+	uint32_t status;
+	int rc;
+
+	rc = efx_mcdi_mdio_write(efx, efx->mdio_bus, prtad, devad,
+				 addr, value, &status);
+	if (rc)
+		return rc;
+	if (status != MC_CMD_MDIO_STATUS_GOOD)
+		return -EIO;
+
+	return 0;
+}
+
+static int siena_mdio_read(struct net_device *net_dev,
+			   int prtad, int devad, u16 addr)
+{
+	struct efx_nic *efx = netdev_priv(net_dev);
+	uint16_t value;
+	uint32_t status;
+	int rc;
+
+	rc = efx_mcdi_mdio_read(efx, efx->mdio_bus, prtad, devad,
+				addr, &value, &status);
+	if (rc)
+		return rc;
+	if (status != MC_CMD_MDIO_STATUS_GOOD)
+		return -EIO;
+
+	return (int)value;
+}
+
+/* This call is responsible for hooking in the MAC and PHY operations */
+static int siena_probe_port(struct efx_nic *efx)
+{
+	int rc;
+
+	/* Hook in PHY operations table */
+	efx->phy_op = &efx_mcdi_phy_ops;
+
+	/* Set up MDIO structure for PHY */
+	efx->mdio.mode_support = MDIO_SUPPORTS_C45 | MDIO_EMULATE_C22;
+	efx->mdio.mdio_read = siena_mdio_read;
+	efx->mdio.mdio_write = siena_mdio_write;
+
+	/* Fill out MDIO structure and loopback modes */
+	rc = efx->phy_op->probe(efx);
+	if (rc != 0)
+		return rc;
+
+	/* Initial assumption */
+	efx->link_state.speed = 10000;
+	efx->link_state.fd = true;
+	efx->wanted_fc = EFX_FC_RX | EFX_FC_TX;
+
+	/* Allocate buffer for stats */
+	rc = efx_nic_alloc_buffer(efx, &efx->stats_buffer,
+				  MC_CMD_MAC_NSTATS * sizeof(u64));
+	if (rc)
+		return rc;
+	EFX_LOG(efx, "stats buffer at %llx (virt %p phys %llx)\n",
+		(u64)efx->stats_buffer.dma_addr,
+		efx->stats_buffer.addr,
+		(u64)virt_to_phys(efx->stats_buffer.addr));
+
+	efx_mcdi_mac_stats(efx, efx->stats_buffer.dma_addr, 0, 0, 1);
+
+	return 0;
+}
+
+void siena_remove_port(struct efx_nic *efx)
+{
+	efx_nic_free_buffer(efx, &efx->stats_buffer);
+}
+
+static const struct efx_nic_register_test siena_register_tests[] = {
+	{ FR_AZ_ADR_REGION,
+	  EFX_OWORD32(0x0001FFFF, 0x0001FFFF, 0x0001FFFF, 0x0001FFFF) },
+	{ FR_CZ_USR_EV_CFG,
+	  EFX_OWORD32(0x000103FF, 0x00000000, 0x00000000, 0x00000000) },
+	{ FR_AZ_RX_CFG,
+	  EFX_OWORD32(0xFFFFFFFE, 0xFFFFFFFF, 0x0003FFFF, 0x00000000) },
+	{ FR_AZ_TX_CFG,
+	  EFX_OWORD32(0x7FFF0037, 0xFFFF8000, 0xFFFFFFFF, 0x03FFFFFF) },
+	{ FR_AZ_TX_RESERVED,
+	  EFX_OWORD32(0xFFFEFE80, 0x1FFFFFFF, 0x020000FE, 0x007FFFFF) },
+	{ FR_AZ_SRM_TX_DC_CFG,
+	  EFX_OWORD32(0x001FFFFF, 0x00000000, 0x00000000, 0x00000000) },
+	{ FR_AZ_RX_DC_CFG,
+	  EFX_OWORD32(0x00000003, 0x00000000, 0x00000000, 0x00000000) },
+	{ FR_AZ_RX_DC_PF_WM,
+	  EFX_OWORD32(0x000003FF, 0x00000000, 0x00000000, 0x00000000) },
+	{ FR_BZ_DP_CTRL,
+	  EFX_OWORD32(0x00000FFF, 0x00000000, 0x00000000, 0x00000000) },
+	{ FR_BZ_RX_RSS_TKEY,
+	  EFX_OWORD32(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF) },
+	{ FR_CZ_RX_RSS_IPV6_REG1,
+	  EFX_OWORD32(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF) },
+	{ FR_CZ_RX_RSS_IPV6_REG2,
+	  EFX_OWORD32(0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF) },
+	{ FR_CZ_RX_RSS_IPV6_REG3,
+	  EFX_OWORD32(0xFFFFFFFF, 0xFFFFFFFF, 0x00000007, 0x00000000) },
+};
+
+static int siena_test_registers(struct efx_nic *efx)
+{
+	return efx_nic_test_registers(efx, siena_register_tests,
+				      ARRAY_SIZE(siena_register_tests));
+}
+
+/**************************************************************************
+ *
+ * Device reset
+ *
+ **************************************************************************
+ */
+
+static int siena_reset_hw(struct efx_nic *efx, enum reset_type method)
+{
+
+	if (method == RESET_TYPE_WORLD)
+		return efx_mcdi_reset_mc(efx);
+	else
+		return efx_mcdi_reset_port(efx);
+}
+
+static int siena_probe_nvconfig(struct efx_nic *efx)
+{
+	int rc;
+
+	rc = efx_mcdi_get_board_cfg(efx, efx->mac_address, NULL);
+	if (rc)
+		return rc;
+
+	return 0;
+}
+
+static int siena_probe_nic(struct efx_nic *efx)
+{
+	struct siena_nic_data *nic_data;
+	bool already_attached = 0;
+	int rc;
+
+	/* Allocate storage for hardware specific data */
+	nic_data = kzalloc(sizeof(struct siena_nic_data), GFP_KERNEL);
+	if (!nic_data)
+		return -ENOMEM;
+	efx->nic_data = nic_data;
+
+	if (efx_nic_fpga_ver(efx) != 0) {
+		EFX_ERR(efx, "Siena FPGA not supported\n");
+		rc = -ENODEV;
+		goto fail1;
+	}
+
+	efx_mcdi_init(efx);
+
+	/* Recover from a failed assertion before probing */
+	rc = efx_mcdi_handle_assertion(efx);
+	if (rc)
+		goto fail1;
+
+	rc = efx_mcdi_fwver(efx, &nic_data->fw_version, &nic_data->fw_build);
+	if (rc) {
+		EFX_ERR(efx, "Failed to read MCPU firmware version - "
+			"rc %d\n", rc);
+		goto fail1; /* MCPU absent? */
+	}
+
+	/* Let the BMC know that the driver is now in charge of link and
+	 * filter settings. We must do this before we reset the NIC */
+	rc = efx_mcdi_drv_attach(efx, true, &already_attached);
+	if (rc) {
+		EFX_ERR(efx, "Unable to register driver with MCPU\n");
+		goto fail2;
+	}
+	if (already_attached)
+		/* Not a fatal error */
+		EFX_ERR(efx, "Host already registered with MCPU\n");
+
+	/* Now we can reset the NIC */
+	rc = siena_reset_hw(efx, RESET_TYPE_ALL);
+	if (rc) {
+		EFX_ERR(efx, "failed to reset NIC\n");
+		goto fail3;
+	}
+
+	siena_init_wol(efx);
+
+	/* Allocate memory for INT_KER */
+	rc = efx_nic_alloc_buffer(efx, &efx->irq_status, sizeof(efx_oword_t));
+	if (rc)
+		goto fail4;
+	BUG_ON(efx->irq_status.dma_addr & 0x0f);
+
+	EFX_LOG(efx, "INT_KER at %llx (virt %p phys %llx)\n",
+		(unsigned long long)efx->irq_status.dma_addr,
+		efx->irq_status.addr,
+		(unsigned long long)virt_to_phys(efx->irq_status.addr));
+
+	/* Read in the non-volatile configuration */
+	rc = siena_probe_nvconfig(efx);
+	if (rc == -EINVAL) {
+		EFX_ERR(efx, "NVRAM is invalid therefore using defaults\n");
+		efx->phy_type = PHY_TYPE_NONE;
+		efx->mdio.prtad = MDIO_PRTAD_NONE;
+	} else if (rc) {
+		goto fail5;
+	}
+
+	return 0;
+
+fail5:
+	efx_nic_free_buffer(efx, &efx->irq_status);
+fail4:
+fail3:
+	efx_mcdi_drv_attach(efx, false, NULL);
+fail2:
+fail1:
+	kfree(efx->nic_data);
+	return rc;
+}
+
+/* This call performs hardware-specific global initialisation, such as
+ * defining the descriptor cache sizes and number of RSS channels.
+ * It does not set up any buffers, descriptor rings or event queues.
+ */
+static int siena_init_nic(struct efx_nic *efx)
+{
+	efx_oword_t temp;
+	int rc;
+
+	/* Recover from a failed assertion post-reset */
+	rc = efx_mcdi_handle_assertion(efx);
+	if (rc)
+		return rc;
+
+	/* Squash TX of packets of 16 bytes or less */
+	efx_reado(efx, &temp, FR_AZ_TX_RESERVED);
+	EFX_SET_OWORD_FIELD(temp, FRF_BZ_TX_FLUSH_MIN_LEN_EN, 1);
+	efx_writeo(efx, &temp, FR_AZ_TX_RESERVED);
+
+	/* Do not enable TX_NO_EOP_DISC_EN, since it limits packets to 16
+	 * descriptors (which is bad).
+	 */
+	efx_reado(efx, &temp, FR_AZ_TX_CFG);
+	EFX_SET_OWORD_FIELD(temp, FRF_AZ_TX_NO_EOP_DISC_EN, 0);
+	EFX_SET_OWORD_FIELD(temp, FRF_CZ_TX_FILTER_EN_BIT, 1);
+	efx_writeo(efx, &temp, FR_AZ_TX_CFG);
+
+	efx_reado(efx, &temp, FR_AZ_RX_CFG);
+	EFX_SET_OWORD_FIELD(temp, FRF_BZ_RX_DESC_PUSH_EN, 0);
+	EFX_SET_OWORD_FIELD(temp, FRF_BZ_RX_INGR_EN, 1);
+	efx_writeo(efx, &temp, FR_AZ_RX_CFG);
+
+	if (efx_nic_rx_xoff_thresh >= 0 || efx_nic_rx_xon_thresh >= 0)
+		/* No MCDI operation has been defined to set thresholds */
+		EFX_ERR(efx, "ignoring RX flow control thresholds\n");
+
+	/* Enable event logging */
+	rc = efx_mcdi_log_ctrl(efx, true, false, 0);
+	if (rc)
+		return rc;
+
+	/* Set destination of both TX and RX Flush events */
+	EFX_POPULATE_OWORD_1(temp, FRF_BZ_FLS_EVQ_ID, 0);
+	efx_writeo(efx, &temp, FR_BZ_DP_CTRL);
+
+	EFX_POPULATE_OWORD_1(temp, FRF_CZ_USREV_DIS, 1);
+	efx_writeo(efx, &temp, FR_CZ_USR_EV_CFG);
+
+	efx_nic_init_common(efx);
+	return 0;
+}
+
+static void siena_remove_nic(struct efx_nic *efx)
+{
+	efx_nic_free_buffer(efx, &efx->irq_status);
+
+	siena_reset_hw(efx, RESET_TYPE_ALL);
+
+	/* Relinquish the device back to the BMC */
+	if (efx_nic_has_mc(efx))
+		efx_mcdi_drv_attach(efx, false, NULL);
+
+	/* Tear down the private nic state */
+	kfree(efx->nic_data);
+	efx->nic_data = NULL;
+}
+
+#define STATS_GENERATION_INVALID ((u64)(-1))
+
+static int siena_try_update_nic_stats(struct efx_nic *efx)
+{
+	u64 *dma_stats;
+	struct efx_mac_stats *mac_stats;
+	u64 generation_start;
+	u64 generation_end;
+
+	mac_stats = &efx->mac_stats;
+	dma_stats = (u64 *)efx->stats_buffer.addr;
+
+	generation_end = dma_stats[MC_CMD_MAC_GENERATION_END];
+	if (generation_end == STATS_GENERATION_INVALID)
+		return 0;
+	rmb();
+
+#define MAC_STAT(M, D) \
+	mac_stats->M = dma_stats[MC_CMD_MAC_ ## D]
+
+	MAC_STAT(tx_bytes, TX_BYTES);
+	MAC_STAT(tx_bad_bytes, TX_BAD_BYTES);
+	mac_stats->tx_good_bytes = (mac_stats->tx_bytes -
+				    mac_stats->tx_bad_bytes);
+	MAC_STAT(tx_packets, TX_PKTS);
+	MAC_STAT(tx_bad, TX_BAD_FCS_PKTS);
+	MAC_STAT(tx_pause, TX_PAUSE_PKTS);
+	MAC_STAT(tx_control, TX_CONTROL_PKTS);
+	MAC_STAT(tx_unicast, TX_UNICAST_PKTS);
+	MAC_STAT(tx_multicast, TX_MULTICAST_PKTS);
+	MAC_STAT(tx_broadcast, TX_BROADCAST_PKTS);
+	MAC_STAT(tx_lt64, TX_LT64_PKTS);
+	MAC_STAT(tx_64, TX_64_PKTS);
+	MAC_STAT(tx_65_to_127, TX_65_TO_127_PKTS);
+	MAC_STAT(tx_128_to_255, TX_128_TO_255_PKTS);
+	MAC_STAT(tx_256_to_511, TX_256_TO_511_PKTS);
+	MAC_STAT(tx_512_to_1023, TX_512_TO_1023_PKTS);
+	MAC_STAT(tx_1024_to_15xx, TX_1024_TO_15XX_PKTS);
+	MAC_STAT(tx_15xx_to_jumbo, TX_15XX_TO_JUMBO_PKTS);
+	MAC_STAT(tx_gtjumbo, TX_GTJUMBO_PKTS);
+	mac_stats->tx_collision = 0;
+	MAC_STAT(tx_single_collision, TX_SINGLE_COLLISION_PKTS);
+	MAC_STAT(tx_multiple_collision, TX_MULTIPLE_COLLISION_PKTS);
+	MAC_STAT(tx_excessive_collision, TX_EXCESSIVE_COLLISION_PKTS);
+	MAC_STAT(tx_deferred, TX_DEFERRED_PKTS);
+	MAC_STAT(tx_late_collision, TX_LATE_COLLISION_PKTS);
+	mac_stats->tx_collision = (mac_stats->tx_single_collision +
+				   mac_stats->tx_multiple_collision +
+				   mac_stats->tx_excessive_collision +
+				   mac_stats->tx_late_collision);
+	MAC_STAT(tx_excessive_deferred, TX_EXCESSIVE_DEFERRED_PKTS);
+	MAC_STAT(tx_non_tcpudp, TX_NON_TCPUDP_PKTS);
+	MAC_STAT(tx_mac_src_error, TX_MAC_SRC_ERR_PKTS);
+	MAC_STAT(tx_ip_src_error, TX_IP_SRC_ERR_PKTS);
+	MAC_STAT(rx_bytes, RX_BYTES);
+	MAC_STAT(rx_bad_bytes, RX_BAD_BYTES);
+	mac_stats->rx_good_bytes = (mac_stats->rx_bytes -
+				    mac_stats->rx_bad_bytes);
+	MAC_STAT(rx_packets, RX_PKTS);
+	MAC_STAT(rx_good, RX_GOOD_PKTS);
+	mac_stats->rx_bad = mac_stats->rx_packets - mac_stats->rx_good;
+	MAC_STAT(rx_pause, RX_PAUSE_PKTS);
+	MAC_STAT(rx_control, RX_CONTROL_PKTS);
+	MAC_STAT(rx_unicast, RX_UNICAST_PKTS);
+	MAC_STAT(rx_multicast, RX_MULTICAST_PKTS);
+	MAC_STAT(rx_broadcast, RX_BROADCAST_PKTS);
+	MAC_STAT(rx_lt64, RX_UNDERSIZE_PKTS);
+	MAC_STAT(rx_64, RX_64_PKTS);
+	MAC_STAT(rx_65_to_127, RX_65_TO_127_PKTS);
+	MAC_STAT(rx_128_to_255, RX_128_TO_255_PKTS);
+	MAC_STAT(rx_256_to_511, RX_256_TO_511_PKTS);
+	MAC_STAT(rx_512_to_1023, RX_512_TO_1023_PKTS);
+	MAC_STAT(rx_1024_to_15xx, RX_1024_TO_15XX_PKTS);
+	MAC_STAT(rx_15xx_to_jumbo, RX_15XX_TO_JUMBO_PKTS);
+	MAC_STAT(rx_gtjumbo, RX_GTJUMBO_PKTS);
+	mac_stats->rx_bad_lt64 = 0;
+	mac_stats->rx_bad_64_to_15xx = 0;
+	mac_stats->rx_bad_15xx_to_jumbo = 0;
+	MAC_STAT(rx_bad_gtjumbo, RX_JABBER_PKTS);
+	MAC_STAT(rx_overflow, RX_OVERFLOW_PKTS);
+	mac_stats->rx_missed = 0;
+	MAC_STAT(rx_false_carrier, RX_FALSE_CARRIER_PKTS);
+	MAC_STAT(rx_symbol_error, RX_SYMBOL_ERROR_PKTS);
+	MAC_STAT(rx_align_error, RX_ALIGN_ERROR_PKTS);
+	MAC_STAT(rx_length_error, RX_LENGTH_ERROR_PKTS);
+	MAC_STAT(rx_internal_error, RX_INTERNAL_ERROR_PKTS);
+	mac_stats->rx_good_lt64 = 0;
+
+	efx->n_rx_nodesc_drop_cnt = dma_stats[MC_CMD_MAC_RX_NODESC_DROPS];
+
+#undef MAC_STAT
+
+	rmb();
+	generation_start = dma_stats[MC_CMD_MAC_GENERATION_START];
+	if (generation_end != generation_start)
+		return -EAGAIN;
+
+	return 0;
+}
+
+static void siena_update_nic_stats(struct efx_nic *efx)
+{
+	while (siena_try_update_nic_stats(efx) == -EAGAIN)
+		cpu_relax();
+}
+
+static void siena_start_nic_stats(struct efx_nic *efx)
+{
+	u64 *dma_stats = (u64 *)efx->stats_buffer.addr;
+
+	dma_stats[MC_CMD_MAC_GENERATION_END] = STATS_GENERATION_INVALID;
+
+	efx_mcdi_mac_stats(efx, efx->stats_buffer.dma_addr,
+			   MC_CMD_MAC_NSTATS * sizeof(u64), 1, 0);
+}
+
+static void siena_stop_nic_stats(struct efx_nic *efx)
+{
+	efx_mcdi_mac_stats(efx, efx->stats_buffer.dma_addr, 0, 0, 0);
+}
+
+void siena_print_fwver(struct efx_nic *efx, char *buf, size_t len)
+{
+	struct siena_nic_data *nic_data = efx->nic_data;
+	snprintf(buf, len, "%u.%u.%u.%u",
+		 (unsigned int)(nic_data->fw_version >> 48),
+		 (unsigned int)(nic_data->fw_version >> 32 & 0xffff),
+		 (unsigned int)(nic_data->fw_version >> 16 & 0xffff),
+		 (unsigned int)(nic_data->fw_version & 0xffff));
+}
+
+/**************************************************************************
+ *
+ * Wake on LAN
+ *
+ **************************************************************************
+ */
+
+static void siena_get_wol(struct efx_nic *efx, struct ethtool_wolinfo *wol)
+{
+	struct siena_nic_data *nic_data = efx->nic_data;
+
+	wol->supported = WAKE_MAGIC;
+	if (nic_data->wol_filter_id != -1)
+		wol->wolopts = WAKE_MAGIC;
+	else
+		wol->wolopts = 0;
+	memset(&wol->sopass, 0, sizeof(wol->sopass));
+}
+
+
+static int siena_set_wol(struct efx_nic *efx, u32 type)
+{
+	struct siena_nic_data *nic_data = efx->nic_data;
+	int rc;
+
+	if (type & ~WAKE_MAGIC)
+		return -EINVAL;
+
+	if (type & WAKE_MAGIC) {
+		if (nic_data->wol_filter_id != -1)
+			efx_mcdi_wol_filter_remove(efx,
+						   nic_data->wol_filter_id);
+		rc = efx_mcdi_wol_filter_set_magic(efx, efx->mac_address,
+						   &nic_data->wol_filter_id);
+		if (rc)
+			goto fail;
+
+		pci_wake_from_d3(efx->pci_dev, true);
+	} else {
+		rc = efx_mcdi_wol_filter_reset(efx);
+		nic_data->wol_filter_id = -1;
+		pci_wake_from_d3(efx->pci_dev, false);
+		if (rc)
+			goto fail;
+	}
+
+	return 0;
+ fail:
+	EFX_ERR(efx, "%s failed: type=%d rc=%d\n", __func__, type, rc);
+	return rc;
+}
+
+
+static void siena_init_wol(struct efx_nic *efx)
+{
+	struct siena_nic_data *nic_data = efx->nic_data;
+	int rc;
+
+	rc = efx_mcdi_wol_filter_get_magic(efx, &nic_data->wol_filter_id);
+
+	if (rc != 0) {
+		/* If it failed, attempt to get into a synchronised
+		 * state with MC by resetting any set WoL filters */
+		efx_mcdi_wol_filter_reset(efx);
+		nic_data->wol_filter_id = -1;
+	} else if (nic_data->wol_filter_id != -1) {
+		pci_wake_from_d3(efx->pci_dev, true);
+	}
+}
+
+
+/**************************************************************************
+ *
+ * Revision-dependent attributes used by efx.c and nic.c
+ *
+ **************************************************************************
+ */
+
+struct efx_nic_type siena_a0_nic_type = {
+	.probe = siena_probe_nic,
+	.remove = siena_remove_nic,
+	.init = siena_init_nic,
+	.fini = efx_port_dummy_op_void,
+	.monitor = NULL,
+	.reset = siena_reset_hw,
+	.probe_port = siena_probe_port,
+	.remove_port = siena_remove_port,
+	.prepare_flush = efx_port_dummy_op_void,
+	.update_stats = siena_update_nic_stats,
+	.start_stats = siena_start_nic_stats,
+	.stop_stats = siena_stop_nic_stats,
+	.set_id_led = efx_mcdi_set_id_led,
+	.push_irq_moderation = siena_push_irq_moderation,
+	.push_multicast_hash = siena_push_multicast_hash,
+	.reconfigure_port = efx_mcdi_phy_reconfigure,
+	.get_wol = siena_get_wol,
+	.set_wol = siena_set_wol,
+	.resume_wol = siena_init_wol,
+	.test_registers = siena_test_registers,
+	.default_mac_ops = &efx_mcdi_mac_operations,
+
+	.revision = EFX_REV_SIENA_A0,
+	.mem_map_size = (FR_CZ_MC_TREG_SMEM +
+			 FR_CZ_MC_TREG_SMEM_STEP * FR_CZ_MC_TREG_SMEM_ROWS),
+	.txd_ptr_tbl_base = FR_BZ_TX_DESC_PTR_TBL,
+	.rxd_ptr_tbl_base = FR_BZ_RX_DESC_PTR_TBL,
+	.buf_tbl_base = FR_BZ_BUF_FULL_TBL,
+	.evq_ptr_tbl_base = FR_BZ_EVQ_PTR_TBL,
+	.evq_rptr_tbl_base = FR_BZ_EVQ_RPTR,
+	.max_dma_mask = DMA_BIT_MASK(FSF_AZ_TX_KER_BUF_ADDR_WIDTH),
+	.rx_buffer_padding = 0,
+	.max_interrupt_mode = EFX_INT_MODE_MSIX,
+	.phys_addr_channels = 32, /* Hardware limit is 64, but the legacy
+				   * interrupt handler only supports 32
+				   * channels */
+	.tx_dc_base = 0x88000,
+	.rx_dc_base = 0x68000,
+	.offload_features = NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM,
+	.reset_world_flags = ETH_RESET_MGMT << ETH_RESET_SHARED_SHIFT,
+};
-- 
1.6.5.3




-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply related


This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox