Skip to contentSkip to conversation

ethtool: Add PHY type to ethtool get_drvinfo

PATCH In review Jeff Kirsher 4 participants 8 messages Nov 17, 2009 – Nov 18, 2009

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 ethtool: Add PHY type to ethtool get_drvinfo

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

---

 include/linux/ethtool.h |   24 +++++++++++++++++++++++-
 1 files changed, 23 insertions(+), 1 deletions(-)
include/linux/ethtool.h +23 -1
index edd03b7..d6606a5 100644
@@ -59,7 +59,8 @@ struct ethtool_drvinfo {
 	char	bus_info[ETHTOOL_BUSINFO_LEN];	/* Bus info for this IF. */
 				/* For PCI devices, use pci_name(pci_dev). */
 	char	reserved1[32];
-	char	reserved2[12];
+	char	reserved2[11];
+	__u8	phy_type;	/* PHY type present on the NIC */
 	__u32	n_priv_flags;	/* number of flags valid in ETHTOOL_GPFLAGS */
 	__u32	n_stats;	/* number of u64's from ETHTOOL_GSTATS */
 	__u32	testinfo_len;
@@ -291,6 +292,27 @@ enum ethtool_flags {
 	ETH_FLAG_LRO		= (1 << 15),	/* LRO is enabled */
 };
 
+/*
+ * PHY types supported
+ *
+ * 0    - PHY not specified
+ * 1    - SFP/SFP+ Fiber (SR and LR)
+ * 2    - XFP Fiber (SR and LR)
+ * 3    - SFP+ Direct Attach (TwinAX)
+ * 4    - BASE-T (RJ-45)
+ * MAX  - PHY not present
+ */
+enum ethtool_phy_type {
+	ETH_PHY_UNSPECIFIED = 0,
+	ETH_PHY_SFP_FIBER,
+	ETH_PHY_XFP_FIBER,
+	ETH_PHY_DA_TWINAX,
+	ETH_PHY_BASE_T,
+
+	/* This must be the last entry */
+	ETH_PHY_NOT_PRESENT,
+};
+
 /* The following structures are for supporting RX network flow
  * classification configuration. Note, all multibyte fields, e.g.,
  * ip4src, ip4dst, psrc, pdst, spi, etc. are expected to be in network
#1 lore raw

PATCH 2/2 ixgbe: Display currently attached PHY through ethtool get_drvinfo

From: PJ Waskiewicz <[email protected]>

This patch uses the new ethtool interface to display what PHY is currently connected to a NIC. The results can be viewed in ethtool -i ethX output.

---

 drivers/net/ixgbe/ixgbe_ethtool.c |   61 +++++++++++++++++++++++++++++++++++++
 1 files changed, 61 insertions(+), 0 deletions(-)
drivers/net/ixgbe/ixgbe_ethtool.c +61 -0
index 9d2cc83..0012c65 100644
@@ -782,6 +782,67 @@ static void ixgbe_get_drvinfo(struct net_device *netdev,
 	drvinfo->n_stats = IXGBE_STATS_LEN;
 	drvinfo->testinfo_len = IXGBE_TEST_LEN;
 	drvinfo->regdump_len = ixgbe_get_regs_len(netdev);
+
+	/* Get PHY type */
+
+	/*
+	 * If the adapter is not up, we should re-identify the PHY type.
+	 * It might be stale otherwise.
+	 */
+	if (test_bit(__IXGBE_DOWN, &adapter->state)) {
+		/* clear the PHY type to ensure we do a full scan */
+		adapter->hw.phy.type = ixgbe_phy_unknown;
+		adapter->hw.phy.ops.identify(&adapter->hw);
+	}
+
+	switch (adapter->hw.phy.type) {
+	case ixgbe_phy_tn:
+	case ixgbe_phy_cu_unknown:
+		/* Copper 10G-BASET */
+		drvinfo->phy_type = ETH_PHY_BASE_T;
+		break;
+	case ixgbe_phy_qt:
+		drvinfo->phy_type = ETH_PHY_XFP_FIBER;
+		break;
+	case ixgbe_phy_nl:
+	case ixgbe_phy_tw_tyco:
+	case ixgbe_phy_tw_unknown:
+	case ixgbe_phy_sfp_ftl:
+	case ixgbe_phy_sfp_avago:
+	case ixgbe_phy_sfp_intel:
+	case ixgbe_phy_sfp_unknown:
+	case ixgbe_phy_unknown:
+		switch (adapter->hw.phy.sfp_type) {
+		/* SFP+ devices, further checking needed */
+		case ixgbe_sfp_type_da_cu:
+		case ixgbe_sfp_type_da_cu_core0:
+		case ixgbe_sfp_type_da_cu_core1:
+			drvinfo->phy_type = ETH_PHY_DA_TWINAX;
+			break;
+		case ixgbe_sfp_type_sr:
+		case ixgbe_sfp_type_lr:
+		case ixgbe_sfp_type_srlr_core0:
+		case ixgbe_sfp_type_srlr_core1:
+			drvinfo->phy_type = ETH_PHY_SFP_FIBER;
+			break;
+		case ixgbe_sfp_type_not_present:
+			drvinfo->phy_type = ETH_PHY_NOT_PRESENT;
+			break;
+		case ixgbe_sfp_type_unknown:
+		default:
+			ETH_PHY_UNSPECIFIED;
+			break;
+		}
+		break;
+	case ixgbe_phy_xaui:
+		drvinfo->phy_type = ETH_PHY_NOT_PRESENT;
+		break;
+	case ixgbe_phy_generic:
+	case ixgbe_phy_sfp_unsupported:
+	default:
+		drvinfo->phy_type = ETH_PHY_UNSPECIFIED;
+		break;
+	}
 }
 
 static void ixgbe_get_ringparam(struct net_device *netdev,
#2 lore raw
In reply to Jeff Kirsher
Quoted 6 lines · Jeff Kirsher

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

[...]

I think this information belongs in ethtool_cmd, in the port and/or transceiver fields.

Ben.

Signature · 3 lines
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.
#3 lore raw
In reply to Jeff Kirsher
Quoted 10 lines · On Tue, 17 Nov 2009 08:13:24 -0800 Jeff Kirsher <[email protected]>…

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

---

Maybe revive usage of connector port in ethtool rather than adding new API? It is already there but values are out of date with current hardware.

#4 lore raw
Quoted 10 lines · Stephen Hemminger
Quoted 7 lines · On Tue, 17 Nov 2009 08:13:24 -0800 Jeff Kirsher <[email protected]>…

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

Quoted 6 lines
Quoted 3 lines

---

Maybe revive usage of connector port in ethtool rather than adding new API? It is already there but values are out of date with current hardware.

You're referring to the MODE_GSET port settings? If so, I like this approach, and will update my patches accordingly.

Thanks Stephen,
-PJ

#5 lore raw
Quoted 16 lines · Peter P
Quoted 14 lines · Stephen Hemminger
Quoted 10 lines · On Tue, 17 Nov 2009 08:13:24 -0800 Jeff Kirsher <[email protected]>…

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

---

Quoted 5 lines
Quoted 2 lines

Maybe revive usage of connector port in ethtool rather than adding new API? It is already there but values are out of date with current hardware.

You're referring to the MODE_GSET port settings? If so, I like this approach, and will update my patches accordingly.

Yes. cmd->port is always PORT_TP or PORT_FIBRE now but could be extended Likewise cmd->transceiver could be modified. I haven't seen an external transceiver since the old fat yellow cable with vampire taps (and that was 20 yrs ago).

#6 lore raw
Quoted 22 lines · Stephen Hemminger
Quoted 19 lines · Peter P
Quoted 16 lines · Stephen Hemminger
Quoted 10 lines · On Tue, 17 Nov 2009 08:13:24 -0800 Jeff Kirsher <[email protected]>…

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

---

Maybe revive usage of connector port in ethtool rather than adding new API? It is already there but values are out of date with current hardware.

Quoted 6 lines
Quoted 2 lines

You're referring to the MODE_GSET port settings? If so, I like this approach, and will update my patches accordingly.

Yes. cmd->port is always PORT_TP or PORT_FIBRE now but could be extended Likewise cmd->transceiver could be modified. I haven't seen an external transceiver since the old fat yellow cable with vampire taps (and that was 20 yrs ago).

I have patches that I sent to Jeff for internal testing that fixes up the cmd->port field (per your and Ben's suggestion). I like it much better too. :-)

-PJ

#7 lore raw
In reply to Stephen Hemminger
Quoted 22 lines · Stephen Hemminger
Quoted 19 lines · Peter P
Quoted 16 lines · Stephen Hemminger
Quoted 10 lines · On Tue, 17 Nov 2009 08:13:24 -0800 Jeff Kirsher <[email protected]>…

From: PJ Waskiewicz <[email protected]>

Allow the PHY type to be passed from a driver to ethtool when ethtool -i ethX is called. With newer network cards having SFP and SFP+ PHY cages, this information can be useful, especially if the NIC supports hot-swapping of the PHY modules.

---

Maybe revive usage of connector port in ethtool rather than adding new API? It is already there but values are out of date with current hardware.

Quoted 6 lines
Quoted 2 lines

You're referring to the MODE_GSET port settings? If so, I like this approach, and will update my patches accordingly.

Yes. cmd->port is always PORT_TP or PORT_FIBRE now but could be extended Likewise cmd->transceiver could be modified. I haven't seen an external transceiver since the old fat yellow cable with vampire taps (and that was 20 yrs ago).

It seems to me there should be a transceiver type code or codes for connectors like SFP where the bulk of the transceiver is internal but the optical part is an external module. SFP+ is a tricky case because it also supports passive cables.

Ben.

Signature · 3 lines
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.