Skip to contentSkip to conversation

NET: ethernet/netlogic: Netlogic XLR/XLS GMAC driver

PATCH In review [email protected] 3 participants 3 messages Feb 5, 2013 – Feb 6, 2013

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

PATCH v2 1/2 NET: ethernet/netlogic: Netlogic XLR/XLS GMAC driver

From: Ganesan Ramalingam <[email protected]>

Add support for the Network Accelerator Engine on Netlogic XLR/XLS MIPS SoCs. The XLR/XLS NAE blocks can be configured as one 10G interface or four 1G interfaces. This driver supports blocks with 1G ports.

---
This patch has to be merged through netdev tree.

Changes since v1:
* Implementing ethtool using phylib
* Free skb buf incase of nlm_fmn_send() fail
* Use of ndo_get_stats64 instead of ndo_get_stats
* Watchdog timeo fix

 drivers/net/ethernet/Kconfig            |    1 +
 drivers/net/ethernet/Makefile           |    1 +
 drivers/net/ethernet/netlogic/Kconfig   |    8 +
 drivers/net/ethernet/netlogic/Makefile  |    1 +
 drivers/net/ethernet/netlogic/xlr_net.c | 1109 +++++++++++++++++++++++++++++++
 drivers/net/ethernet/netlogic/xlr_net.h | 1099 ++++++++++++++++++++++++++++++
 6 files changed, 2219 insertions(+), 0 deletions(-)
 create mode 100644 drivers/net/ethernet/netlogic/Kconfig
 create mode 100644 drivers/net/ethernet/netlogic/Makefile
 create mode 100644 drivers/net/ethernet/netlogic/xlr_net.c
 create mode 100644 drivers/net/ethernet/netlogic/xlr_net.h
drivers/net/ethernet/Kconfig +1 -0
index e4ff389..5a8c9dc 100644
@@ -168,5 +168,6 @@ source "drivers/net/ethernet/via/Kconfig"
 source "drivers/net/ethernet/wiznet/Kconfig"
 source "drivers/net/ethernet/xilinx/Kconfig"
 source "drivers/net/ethernet/xircom/Kconfig"
+source "drivers/net/ethernet/netlogic/Kconfig"
 
 endif # ETHERNET
drivers/net/ethernet/Makefile +1 -0
index d447307..cef1593 100644
@@ -75,3 +75,4 @@ obj-$(CONFIG_NET_VENDOR_VIA) += via/
 obj-$(CONFIG_NET_VENDOR_WIZNET) += wiznet/
 obj-$(CONFIG_NET_VENDOR_XILINX) += xilinx/
 obj-$(CONFIG_NET_VENDOR_XIRCOM) += xircom/
+obj-$(CONFIG_NETLOGIC_XLR_NET) += netlogic/
drivers/net/ethernet/netlogic/Kconfig new file +8 -0
new file mode 100644
index 0000000..a6c8a2f
@@ -0,0 +1,8 @@
+config NETLOGIC_XLR_NET
+	tristate "Netlogic XLR/XLS network device"
+	default y
+	select PHYLIB
+	depends on CPU_XLR
+	---help---
+	This driver support Netlogic XLR/XLS on chip gigabit
+	Ethernet.
drivers/net/ethernet/netlogic/Makefile new file +1 -0
new file mode 100644
index 0000000..7fc927a
@@ -0,0 +1 @@
+obj-$(CONFIG_NETLOGIC_XLR_NET) += xlr_net.o
drivers/net/ethernet/netlogic/xlr_net.c new file +1109 -0

1,109 lines not on this page · raw message

drivers/net/ethernet/netlogic/xlr_net.h new file +1099 -0

1,099 lines not on this page · raw message

Signature · 1 line
1.7.6
#1 lore raw
Quoted 16 lines · On Tue, 2013-02-05 at 17:00 +0530, [email protected] wrote:

From: Ganesan Ramalingam <[email protected]>

Add support for the Network Accelerator Engine on Netlogic XLR/XLS MIPS SoCs. The XLR/XLS NAE blocks can be configured as one 10G interface or four 1G interfaces. This driver supports blocks with 1G ports.

---
This patch has to be merged through netdev tree.

Changes since v1:
* Implementing ethtool using phylib
* Free skb buf incase of nlm_fmn_send() fail
* Use of ndo_get_stats64 instead of ndo_get_stats
* Watchdog timeo fix

[...]

drivers/net/ethernet/netlogic/xlr_net.c line 378

Quoted 13 lines
drivers/net/ethernet/netlogic/xlr_net.c new file +13 -0
+static void xlr_stats(struct net_device *ndev, struct rtnl_link_stats64 *stats)
+{
+	struct xlr_net_priv *priv = netdev_priv(ndev);
+
+	stats->rx_packets = xlr_nae_rdreg(priv->base_addr, RX_PACKET_COUNTER);
+	stats->tx_packets = xlr_nae_rdreg(priv->base_addr, TX_PACKET_COUNTER);
+	stats->rx_bytes = xlr_nae_rdreg(priv->base_addr, RX_BYTE_COUNTER);
+	stats->tx_bytes = xlr_nae_rdreg(priv->base_addr, TX_BYTE_COUNTER);
+	stats->tx_errors = xlr_nae_rdreg(priv->base_addr, TX_FCS_ERROR_COUNTER);
+	stats->rx_dropped = xlr_nae_rdreg(priv->base_addr,
+			RX_DROP_PACKET_COUNTER);
+	stats->tx_dropped = xlr_nae_rdreg(priv->base_addr,
+			TX_DROP_FRAME_COUNTER);

[...]

But these are still 32-bit counters in hardware, yes? So I think you need to arrange to poll at least the byte counters every few seconds and maintain the higher bits in software e.g.:

struct xlr_stats {
	spinlock_t lock;
	struct timer timer;
	u64 rx_bytes;
	u64 tx_bytes;
	...
};

static void xlr_poll_stats(struct xlr_net_priv *priv)

{
	struct xlr_stats = &priv->stats;
	u32 counter;

	spin_lock_bh(&stats->lock);
	counter = xlr_nae_rdreg(priv->base_addr, RX_BYTE_COUNTER);
	stats->rx_bytes += (u32)(counter - stats->rx_bytes);
	counter = xlr_nae_rdreg(priv->base_addr, TX_BYTE_COUNTER);
	stats->tx_bytes += (u32)(counter - stats->tx_bytes);
	...
	spin_unlock_bh(&stats->lock);
}

static void xlr_poll_stats_timer(void *data)

{
	struct xlr_net_priv *priv = data;

	xlr_poll_stats(priv);
	mod_timer(&priv->stats.timer, jiffies + 2 * HZ);
}

static void xlr_stats(struct net_device *ndev, struct rtnl_link_stats64 *stats)

{
	struct xlr_net_priv *priv = netdev_priv(ndev);

	xlr_poll_stats(priv);
	...
}

static int xlr_net_probe(struct platform_device *pdev)

{
	...
	spin_lock_init(&priv->stats.lock);
	setup_timer(&priv->stats.timer, xlr_poll_stats_timer, priv);
	...
}

Plus add_timer() and del_timer_sync() when starting/stopping the interface. Also you need to clear the software stats if you have to reset the hardware. All of which is a fair amount of work but you can't avoid it if the hardware counters just aren't wide enough.

Ben.

Signature · 3 lines
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
maintainer
#2 lore raw
In reply to [email protected]

From: [email protected]
Date: Tue, 5 Feb 2013 17:00:19 +0530

drivers/net/ethernet/netlogic/Kconfig line 8

Quoted 8 lines
drivers/net/ethernet/netlogic/Kconfig new file +8 -0
+config NETLOGIC_XLR_NET
+	tristate "Netlogic XLR/XLS network device"
+	default y
+	select PHYLIB
+	depends on CPU_XLR
+	---help---
+	This driver support Netlogic XLR/XLS on chip gigabit
+	Ethernet.

No individual device driver should default to 'y'. Vendor guards, yes, can
default to 'y', but not individual drivers.

Quoted 5 lines

+/*
+ * The readl/writel implementation byteswaps on XLR/XLS, so
+ * we need to use __raw_ IO to read the NAE registers
+ * because they are in the big-endian MMIO area on the SoC.
+ */

Comments in the networking are to be formatted:

	/* Like
	 * this.
	 */

Please fix this up in your entire driver.

drivers/net/ethernet/netlogic/xlr_net.c line 82

Quoted 7 lines
drivers/net/ethernet/netlogic/xlr_net.c new file +7 -0
+/*
+ * Table of net_device pointers indexed by port, this will be used to
+ * lookup the net_device corresponding to a port by the message ring handler.
+ *
+ * Maximum ports in XLR/XLS is 8(8 GMAC on XLS, 4 GMAC + 2 XGMAC on XLR)
+ */
+static struct net_device *mac_to_ndev[8];

Make this a dynamic data structure, a parent device that the individual netdevs are hung off of, it can still be an array. That way you can have a bonafide struct device instance and associated hierarchy of devices in the kernel device list.

Also avoid this strange and non-standard usage of "MAC" as an integer port index. The canonical meaning of MAC is the link-layer address of the device.

drivers/net/ethernet/netlogic/xlr_net.c line 94

Quoted 12 lines
drivers/net/ethernet/netlogic/xlr_net.c new file +12 -0
+
+static inline struct sk_buff *mac_get_skb_back_ptr(void *addr)
+{
+	struct sk_buff **back_ptr;
+
+	/* this function should be used only for newly allocated packets.
+	 * It assumes the first cacheline is for the back pointer related
+	 * book keeping info.
+	 */
+	back_ptr = (struct sk_buff **)(addr - MAC_SKB_BACK_PTR_SIZE);
+	return *back_ptr;
+}

Use the skb->cb[] control block rather than mis-using the skb data area for storing internal driver state.

drivers/net/ethernet/netlogic/xlr_net.c line 114

Quoted 1 line
drivers/net/ethernet/netlogic/xlr_net.c new file +1 -0
+	paddr = virt_to_bus(addr);

virt_to_bus() is verboten, use the proper DMA APIs. I don't care if this is a specialized driver for a special platform.

drivers/net/ethernet/netlogic/xlr_net.c line 157

Quoted 1 line
drivers/net/ethernet/netlogic/xlr_net.c new file +1 -0
+		addr = bus_to_virt(msg->msg0 & 0xffffffffffULL);

bus_to_virt() is verbotten, use the proper DMA APIs and store correct references to packets in a translation data structure in your per-netdev software state.

Quoted 1 line

+static void __maybe_unused xlr_wakeup_queue(unsigned long dev)

This is really unused, just delete it.

That's enough for me, this driver needs a lot of work.