[PATCH] net/pch_gbe: supports eg20t ptp clock

[PATCH] net/pch_gbe: supports eg20t ptp clock

[Takahiro Shimizu]

From: Takahiroi Shimizu <tshimizu818@gmail.com>

Supports EG20T ptp clock in the driver

Signed-off-by: Takahiro Shimizu <tshimizu818@gmail.com>
---
 drivers/net/ethernet/oki-semi/pch_gbe/Kconfig      |   13 ++
 drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe.h    |   13 ++
 .../net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c   |  217 +++++++++++++++++++-
 3 files changed, 240 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/oki-semi/pch_gbe/Kconfig b/drivers/net/ethernet/oki-semi/pch_gbe/Kconfig
index 00bc4fc..bce0164 100644
--- a/drivers/net/ethernet/oki-semi/pch_gbe/Kconfig
+++ b/drivers/net/ethernet/oki-semi/pch_gbe/Kconfig
@@ -20,3 +20,16 @@ config PCH_GBE
 	  purpose use.
 	  ML7223/ML7831 is companion chip for Intel Atom E6xx series.
 	  ML7223/ML7831 is completely compatible for Intel EG20T PCH.
+
+if PCH_GBE
+
+config PCH_PTP
+	bool "PCH PTP clock support"
+	default n
+	depends on PTP_1588_CLOCK_PCH
+	---help---
+	  Say Y here if you want to use Precision Time Protocol (PTP) in the
+	  driver. PTP is a method to precisely synchronize distributed clocks
+	  over Ethernet networks.
+
+endif # PCH_GBE
diff --git a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe.h b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe.h
index a09a071..dd14915 100644
--- a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe.h
+++ b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe.h
@@ -630,6 +630,9 @@ struct pch_gbe_adapter {
 	unsigned long tx_queue_len;
 	bool have_msi;
 	bool rx_stop_flag;
+	int hwts_tx_en;
+	int hwts_rx_en;
+	struct pci_dev *ptp_pdev;
 };
 
 extern const char pch_driver_version[];
@@ -648,6 +651,16 @@ extern void pch_gbe_free_tx_resources(struct pch_gbe_adapter *adapter,
 extern void pch_gbe_free_rx_resources(struct pch_gbe_adapter *adapter,
 				       struct pch_gbe_rx_ring *rx_ring);
 extern void pch_gbe_update_stats(struct pch_gbe_adapter *adapter);
+#ifdef CONFIG_PCH_PTP
+extern u32 pch_ch_control_read(struct pci_dev *pdev);
+extern void pch_ch_control_write(struct pci_dev *pdev, u32 val);
+extern u32 pch_ch_event_read(struct pci_dev *pdev);
+extern void pch_ch_event_write(struct pci_dev *pdev, u32 val);
+extern u32 pch_src_uuid_lo_read(struct pci_dev *pdev);
+extern u32 pch_src_uuid_hi_read(struct pci_dev *pdev);
+extern u64 pch_rx_snap_read(struct pci_dev *pdev);
+extern u64 pch_tx_snap_read(struct pci_dev *pdev);
+#endif
 
 /* pch_gbe_param.c */
 extern void pch_gbe_check_options(struct pch_gbe_adapter *adapter);
diff --git a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
index 3ead111..aa54ede 100644
--- a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
+++ b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 1999 - 2010 Intel Corporation.
- * Copyright (C) 2010 OKI SEMICONDUCTOR CO., LTD.
+ * Copyright (C) 2010 - 2012 LAPIS SEMICONDUCTOR CO., LTD.
  *
  * This code was derived from the Intel e1000e Linux driver.
  *
@@ -21,6 +21,10 @@
 #include "pch_gbe.h"
 #include "pch_gbe_api.h"
 #include <linux/module.h>
+#ifdef CONFIG_PCH_PTP
+#include <linux/net_tstamp.h>
+#include <linux/ptp_classify.h>
+#endif
 
 #define DRV_VERSION     "1.00"
 const char pch_driver_version[] = DRV_VERSION;
@@ -95,12 +99,195 @@ const char pch_driver_version[] = DRV_VERSION;
 
 #define PCH_GBE_INT_DISABLE_ALL		0
 
+#ifdef CONFIG_PCH_PTP
+/* Macros for ieee1588 */
+#define TICKS_NS_SHIFT  5
+
+/* 0x40 Time Synchronization Channel Control Register Bits */
+#define MASTER_MODE   (1<<0)
+#define SLAVE_MODE    (0<<0)
+#define V2_MODE       (1<<31)
+#define CAP_MODE0     (0<<16)
+#define CAP_MODE2     (1<<17)
+
+/* 0x44 Time Synchronization Channel Event Register Bits */
+#define TX_SNAPSHOT_LOCKED (1<<0)
+#define RX_SNAPSHOT_LOCKED (1<<1)
+#endif
+
 static unsigned int copybreak __read_mostly = PCH_GBE_COPYBREAK_DEFAULT;
 
 static int pch_gbe_mdio_read(struct net_device *netdev, int addr, int reg);
 static void pch_gbe_mdio_write(struct net_device *netdev, int addr, int reg,
 			       int data);
 
+#ifdef CONFIG_PCH_PTP
+static struct sock_filter ptp_filter[] = {
+	PTP_FILTER
+};
+
+static int pch_ptp_match(struct sk_buff *skb, u16 uid_hi, u32 uid_lo, u16 seqid)
+{
+	u8 *data = skb->data;
+	unsigned int offset;
+	u16 *hi, *id;
+	u32 lo;
+
+	if ((sk_run_filter(skb, ptp_filter) != PTP_CLASS_V2_IPV4) &&
+		(sk_run_filter(skb, ptp_filter) != PTP_CLASS_V1_IPV4)) {
+		return 0;
+	}
+
+	offset = ETH_HLEN + IPV4_HLEN(data) + UDP_HLEN;
+
+	if (skb->len < offset + OFF_PTP_SEQUENCE_ID + sizeof(seqid))
+		return 0;
+
+	hi = (u16 *)(data + offset + OFF_PTP_SOURCE_UUID);
+	id = (u16 *)(data + offset + OFF_PTP_SEQUENCE_ID);
+
+	memcpy(&lo, &hi[1], sizeof(lo));
+
+	return (uid_hi == *hi &&
+		uid_lo == lo &&
+		seqid  == *id);
+}
+
+static void pch_rx_timestamp(
+			struct pch_gbe_adapter *adapter, struct sk_buff *skb)
+{
+	struct skb_shared_hwtstamps *shhwtstamps;
+	struct pci_dev *pdev;
+	u64 ns;
+	u32 hi, lo, val;
+	u16 uid, seq;
+
+	if (!adapter->hwts_rx_en)
+		return;
+
+	/* Get ieee1588's dev information */
+	pdev = adapter->ptp_pdev;
+
+	val = pch_ch_event_read(pdev);
+
+	if (!(val & RX_SNAPSHOT_LOCKED))
+		return;
+
+	lo = pch_src_uuid_lo_read(pdev);
+	hi = pch_src_uuid_hi_read(pdev);
+
+	uid = hi & 0xffff;
+	seq = (hi >> 16) & 0xffff;
+
+	if (!pch_ptp_match(skb, htons(uid), htonl(lo), htons(seq)))
+		goto out;
+
+	ns = pch_rx_snap_read(pdev);
+	ns <<= TICKS_NS_SHIFT;
+
+	shhwtstamps = skb_hwtstamps(skb);
+	memset(shhwtstamps, 0, sizeof(*shhwtstamps));
+	shhwtstamps->hwtstamp = ns_to_ktime(ns);
+out:
+	pch_ch_event_write(pdev, RX_SNAPSHOT_LOCKED);
+}
+
+static void pch_tx_timestamp(
+			struct pch_gbe_adapter *adapter, struct sk_buff *skb)
+{
+	struct skb_shared_hwtstamps shhwtstamps;
+	struct pci_dev *pdev;
+	struct skb_shared_info *shtx;
+	u64 ns;
+	u32 cnt, val;
+
+	shtx = skb_shinfo(skb);
+	if (unlikely(shtx->tx_flags & SKBTX_HW_TSTAMP && adapter->hwts_tx_en))
+		shtx->tx_flags |= SKBTX_IN_PROGRESS;
+	else
+		return;
+
+	/* Get ieee1588's dev information */
+	pdev = adapter->ptp_pdev;
+
+	/*
+	 * This really stinks, but we have to poll for the Tx time stamp.
+	 * Usually, the time stamp is ready after 4 to 6 microseconds.
+	 */
+	for (cnt = 0; cnt < 100; cnt++) {
+		val = pch_ch_event_read(pdev);
+		if (val & TX_SNAPSHOT_LOCKED)
+			break;
+		udelay(1);
+	}
+	if (!(val & TX_SNAPSHOT_LOCKED)) {
+		shtx->tx_flags &= ~SKBTX_IN_PROGRESS;
+		return;
+	}
+
+	ns = pch_tx_snap_read(pdev);
+	ns <<= TICKS_NS_SHIFT;
+
+	memset(&shhwtstamps, 0, sizeof(shhwtstamps));
+	shhwtstamps.hwtstamp = ns_to_ktime(ns);
+	skb_tstamp_tx(skb, &shhwtstamps);
+
+	pch_ch_event_write(pdev, TX_SNAPSHOT_LOCKED);
+}
+
+static int hwtstamp_ioctl(struct net_device *netdev, struct ifreq *ifr, int cmd)
+{
+	struct hwtstamp_config cfg;
+	struct pch_gbe_adapter *adapter = netdev_priv(netdev);
+	struct pci_dev *pdev;
+
+	if (copy_from_user(&cfg, ifr->ifr_data, sizeof(cfg)))
+		return -EFAULT;
+
+	if (cfg.flags) /* reserved for future extensions */
+		return -EINVAL;
+
+	/* Get ieee1588's dev information */
+	pdev = adapter->ptp_pdev;
+
+	switch (cfg.tx_type) {
+	case HWTSTAMP_TX_OFF:
+		adapter->hwts_tx_en = 0;
+		break;
+	case HWTSTAMP_TX_ON:
+		adapter->hwts_tx_en = 1;
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	switch (cfg.rx_filter) {
+	case HWTSTAMP_FILTER_NONE:
+		adapter->hwts_rx_en = 0;
+		break;
+	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+		adapter->hwts_rx_en = 0;
+		pch_ch_control_write(pdev, (SLAVE_MODE | CAP_MODE0));
+		break;
+	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+		adapter->hwts_rx_en = 1;
+		pch_ch_control_write(pdev, (MASTER_MODE | CAP_MODE0));
+		break;
+	case HWTSTAMP_FILTER_PTP_V2_EVENT:
+		adapter->hwts_rx_en = 1;
+		pch_ch_control_write(pdev, (V2_MODE | CAP_MODE2));
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	/* Clear out any old time stamps. */
+	pch_ch_event_write(pdev, TX_SNAPSHOT_LOCKED | RX_SNAPSHOT_LOCKED);
+
+	return copy_to_user(ifr->ifr_data, &cfg, sizeof(cfg)) ? -EFAULT : 0;
+}
+#endif
+
 inline void pch_gbe_mac_load_mac_addr(struct pch_gbe_hw *hw)
 {
 	iowrite32(0x01, &hw->reg->MAC_ADDR_LOAD);
@@ -1072,6 +1259,11 @@ static void pch_gbe_tx_queue(struct pch_gbe_adapter *adapter,
 	iowrite32(tx_ring->dma +
 		  (int)sizeof(struct pch_gbe_tx_desc) * ring_num,
 		  &hw->reg->TX_DSC_SW_P);
+
+#ifdef CONFIG_PCH_PTP
+	pch_tx_timestamp(adapter, skb);
+#endif
+
 	dev_kfree_skb_any(skb);
 }
 
@@ -1543,6 +1735,11 @@ pch_gbe_clean_rx(struct pch_gbe_adapter *adapter,
 				adapter->stats.multicast++;
 			/* Write meta date of skb */
 			skb_put(skb, length);
+
+#ifdef CONFIG_PCH_PTP
+			pch_rx_timestamp(adapter, skb);
+#endif
+
 			skb->protocol = eth_type_trans(skb, netdev);
 			if (tcp_ip_status & PCH_GBE_RXD_ACC_STAT_TCPIPOK)
 				skb->ip_summed = CHECKSUM_NONE;
@@ -2147,6 +2344,11 @@ static int pch_gbe_ioctl(struct net_device *netdev, struct ifreq *ifr, int cmd)
 
 	pr_debug("cmd : 0x%04x\n", cmd);
 
+#ifdef CONFIG_PCH_PTP
+	if (cmd == SIOCSHWTSTAMP)
+		return hwtstamp_ioctl(netdev, ifr, cmd);
+#endif
+
 	return generic_mii_ioctl(&adapter->mii, if_mii(ifr), cmd, NULL);
 }
 
@@ -2440,6 +2642,15 @@ static int pch_gbe_probe(struct pci_dev *pdev,
 		goto err_free_netdev;
 	}
 
+#ifdef CONFIG_PCH_PTP
+	adapter->ptp_pdev = pci_get_bus_and_slot(adapter->pdev->bus->number,
+					       PCI_DEVFN(12, 4));
+	if (ptp_filter_init(ptp_filter, ARRAY_SIZE(ptp_filter))) {
+		pr_err("Bad ptp filter\n");
+		return -EINVAL;
+	}
+#endif
+
 	netdev->netdev_ops = &pch_gbe_netdev_ops;
 	netdev->watchdog_timeo = PCH_GBE_WATCHDOG_PERIOD;
 	netif_napi_add(netdev, &adapter->napi,
@@ -2504,7 +2715,7 @@ static int pch_gbe_probe(struct pci_dev *pdev,
 	netif_carrier_off(netdev);
 	netif_stop_queue(netdev);
 
-	dev_dbg(&pdev->dev, "OKIsemi(R) PCH Network Connection\n");
+	dev_dbg(&pdev->dev, "PCH Network Connection\n");
 
 	device_set_wakeup_enable(&pdev->dev, 1);
 	return 0;
@@ -2605,7 +2816,7 @@ module_init(pch_gbe_init_module);
 module_exit(pch_gbe_exit_module);
 
 MODULE_DESCRIPTION("EG20T PCH Gigabit ethernet Driver");
-MODULE_AUTHOR("OKI SEMICONDUCTOR, <toshiharu-linux@dsn.okisemi.com>");
+MODULE_AUTHOR("LAPIS SEMICONDUCTOR, <tshimizu818@gmail.com>");
 MODULE_LICENSE("GPL");
 MODULE_VERSION(DRV_VERSION);
 MODULE_DEVICE_TABLE(pci, pch_gbe_pcidev_id);
-- 
1.7.4.4

Re: [PATCH] net/pch_gbe: supports eg20t ptp clock

[David Miller]

So how the heck is this supposed to work?

You post a patch for the e20tp clock source, you don't CC: netdev.

Then you post this network driver patch which depends upon that on
and you do CC: netdev.

You've made it impossible for me to apply them both properly into my
tree so that the dependency goes in.

Re: [PATCH] net/pch_gbe: supports eg20t ptp clock

[Joe Perches]

On Mon, 2012-03-05 at 14:46 +0900, Takahiro Shimizu wrote:
> From: Takahiroi Shimizu <tshimizu818@gmail.com>
> Supports EG20T ptp clock in the driver

just some trivial comments...

> diff --git a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
> +/* 0x40 Time Synchronization Channel Control Register Bits */
> +#define MASTER_MODE   (1<<0)
> +#define SLAVE_MODE    (0<<0)

> +#define V2_MODE       (1<<31)
> +#define CAP_MODE0     (0<<16)

0 << anything is pretty ugly.

[]

> +static void pch_rx_timestamp(
> +			struct pch_gbe_adapter *adapter, struct sk_buff *skb)

Also not too pretty.

Maybe

static void
pch_rx_timestamp(struct pch_gbe_adapter *adapter, struct sk_buff *skb)

[]
> +static void pch_tx_timestamp(
> +			struct pch_gbe_adapter *adapter, struct sk_buff *skb)

here too.
[]
> +	shtx = skb_shinfo(skb);
> +	if (unlikely(shtx->tx_flags & SKBTX_HW_TSTAMP && adapter->hwts_tx_en))
> +		shtx->tx_flags |= SKBTX_IN_PROGRESS;
> +	else
> +		return;

probably better as

	if (likely(!(shtx->tx_flags & SKBTX_HW_TSTAMP &&
		     adapter->hwts_tx_en)))
		return;

	shtx->tx_flags |= etc...

> +
> +	/* Get ieee1588's dev information */
> +	pdev = adapter->ptp_pdev;
> +
> +	/*
> +	 * This really stinks, but we have to poll for the Tx time stamp.
> +	 * Usually, the time stamp is ready after 4 to 6 microseconds.
> +	 */
> +	for (cnt = 0; cnt < 100; cnt++) {
> +		val = pch_ch_event_read(pdev);
> +		if (val & TX_SNAPSHOT_LOCKED)
> +			break;
> +		udelay(1);
> +	}

why not just continually poll instead of udelay?

[]

> +static int hwtstamp_ioctl(struct net_device *netdev, struct ifreq *ifr, int cmd)
[]
> +	switch (cfg.rx_filter) {
> +	case HWTSTAMP_FILTER_NONE:
> +		adapter->hwts_rx_en = 0;
> +		break;
> +	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
> +		adapter->hwts_rx_en = 0;
> +		pch_ch_control_write(pdev, (SLAVE_MODE | CAP_MODE0));

parentheses are not nececessary around SLAVE_MODE | CAP_MODE0

> +		break;
> +	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
> +		adapter->hwts_rx_en = 1;
> +		pch_ch_control_write(pdev, (MASTER_MODE | CAP_MODE0));

here too.

Re: [PATCH] net/pch_gbe: supports eg20t ptp clock

[Joe Perches]

On Mon, 2012-03-05 at 14:46 +0900, Takahiro Shimizu wrote:
> From: Takahiroi Shimizu <tshimizu818@gmail.com>
> Supports EG20T ptp clock in the driver

just some trivial comments...

> diff --git a/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c b/drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c
> +/* 0x40 Time Synchronization Channel Control Register Bits */
> +#define MASTER_MODE   (1<<0)
> +#define SLAVE_MODE    (0<<0)

> +#define V2_MODE       (1<<31)
> +#define CAP_MODE0     (0<<16)

0 << anything is pretty ugly.

[]

> +static void pch_rx_timestamp(
> +			struct pch_gbe_adapter *adapter, struct sk_buff *skb)

Also not too pretty.

Maybe

static void
pch_rx_timestamp(struct pch_gbe_adapter *adapter, struct sk_buff *skb)

[]

> +static void pch_tx_timestamp(
> +			struct pch_gbe_adapter *adapter, struct sk_buff *skb)

here too.

[]

> +	shtx = skb_shinfo(skb);
> +	if (unlikely(shtx->tx_flags & SKBTX_HW_TSTAMP && adapter->hwts_tx_en))
> +		shtx->tx_flags |= SKBTX_IN_PROGRESS;
> +	else
> +		return;

probably better as

	if (likely(!(shtx->tx_flags & SKBTX_HW_TSTAMP &&
		     adapter->hwts_tx_en)))
		return;

	shtx->tx_flags |= etc...

> +
> +	/* Get ieee1588's dev information */
> +	pdev = adapter->ptp_pdev;
> +
> +	/*
> +	 * This really stinks, but we have to poll for the Tx time stamp.
> +	 * Usually, the time stamp is ready after 4 to 6 microseconds.
> +	 */
> +	for (cnt = 0; cnt < 100; cnt++) {
> +		val = pch_ch_event_read(pdev);
> +		if (val & TX_SNAPSHOT_LOCKED)
> +			break;
> +		udelay(1);
> +	}

why not just continually poll until timeout instead of udelay?

[]

> +static int hwtstamp_ioctl(struct net_device *netdev, struct ifreq *ifr, int cmd)

[]

> +	switch (cfg.rx_filter) {
> +	case HWTSTAMP_FILTER_NONE:
> +		adapter->hwts_rx_en = 0;
> +		break;
> +	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
> +		adapter->hwts_rx_en = 0;
> +		pch_ch_control_write(pdev, (SLAVE_MODE | CAP_MODE0));

parentheses are not necessary around SLAVE_MODE | CAP_MODE0

> +		break;
> +	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
> +		adapter->hwts_rx_en = 1;
> +		pch_ch_control_write(pdev, (MASTER_MODE | CAP_MODE0));

here too.

Re: [PATCH] net/pch_gbe: supports eg20t ptp clock

[Richard Cochran]

On Mon, Mar 05, 2012 at 12:57:09AM -0500, David Miller wrote:
> 
> So how the heck is this supposed to work?
> 
> You post a patch for the e20tp clock source, you don't CC: netdev.
> 
> Then you post this network driver patch which depends upon that on
> and you do CC: netdev.
> 
> You've made it impossible for me to apply them both properly into my
> tree so that the dependency goes in.

I'm confused. Are you talking about a patch from last August?

In any case, if there two patches, then they should of course appear
together.

Thanks,
Richard