iseries_veth.c

来自「linux 内核源代码」· C语言 代码 · 共 1,730 行 · 第 1/3 页

/* File veth.c created by Kyle A. Lucke on Mon Aug  7 2000. */
/*
 * IBM eServer iSeries Virtual Ethernet Device Driver
 * Copyright (C) 2001 Kyle A. Lucke ([e-mail]), IBM Corp.
 * Substantially cleaned up by:
 * Copyright (C) 2003 David Gibson <[e-mail]>, IBM Corporation.
 * Copyright (C) 2004-2005 Michael Ellerman, IBM Corporation.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation; either version 2 of the
 * License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
 * USA
 *
 *
 * This module implements the virtual ethernet device for iSeries LPAR
 * Linux.  It uses hypervisor message passing to implement an
 * ethernet-like network device communicating between partitions on
 * the iSeries.
 *
 * The iSeries LPAR hypervisor currently allows for up to 16 different
 * virtual ethernets.  These are all dynamically configurable on
 * OS/400 partitions, but dynamic configuration is not supported under
 * Linux yet.  An ethXX network device will be created for each
 * virtual ethernet this partition is connected to.
 *
 * - This driver is responsible for routing packets to and from other
 *   partitions.  The MAC addresses used by the virtual ethernets
 *   contains meaning and must not be modified.
 *
 * - Having 2 virtual ethernets to the same remote partition DOES NOT
 *   double the available bandwidth.  The 2 devices will share the
 *   available hypervisor bandwidth.
 *
 * - If you send a packet to your own mac address, it will just be
 *   dropped, you won't get it on the receive side.
 *
 * - Multicast is implemented by sending the frame frame to every
 *   other partition.  It is the responsibility of the receiving
 *   partition to filter the addresses desired.
 *
 * Tunable parameters:
 *
 * VETH_NUMBUFFERS: This compile time option defaults to 120.  It
 * controls how much memory Linux will allocate per remote partition
 * it is communicating with.  It can be thought of as the maximum
 * number of packets outstanding to a remote partition at a time.
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/errno.h>
#include <linux/ioport.h>
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/skbuff.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/mm.h>
#include <linux/ethtool.h>
#include <linux/if_ether.h>

#include <asm/abs_addr.h>
#include <asm/iseries/mf.h>
#include <asm/uaccess.h>
#include <asm/firmware.h>
#include <asm/iseries/hv_lp_config.h>
#include <asm/iseries/hv_types.h>
#include <asm/iseries/hv_lp_event.h>
#include <asm/iommu.h>
#include <asm/vio.h>

#undef DEBUG

MODULE_AUTHOR("Kyle Lucke <[e-mail]>");
MODULE_DESCRIPTION("iSeries Virtual ethernet driver");
MODULE_LICENSE("GPL");

#define VETH_EVENT_CAP	(0)
#define VETH_EVENT_FRAMES	(1)
#define VETH_EVENT_MONITOR	(2)
#define VETH_EVENT_FRAMES_ACK	(3)

#define VETH_MAX_ACKS_PER_MSG	(20)
#define VETH_MAX_FRAMES_PER_MSG	(6)

struct veth_frames_data {
	u32 addr[VETH_MAX_FRAMES_PER_MSG];
	u16 len[VETH_MAX_FRAMES_PER_MSG];
	u32 eofmask;
};
#define VETH_EOF_SHIFT		(32-VETH_MAX_FRAMES_PER_MSG)

struct veth_frames_ack_data {
	u16 token[VETH_MAX_ACKS_PER_MSG];
};

struct veth_cap_data {
	u8 caps_version;
	u8 rsvd1;
	u16 num_buffers;
	u16 ack_threshold;
	u16 rsvd2;
	u32 ack_timeout;
	u32 rsvd3;
	u64 rsvd4[3];
};

struct veth_lpevent {
	struct HvLpEvent base_event;
	union {
		struct veth_cap_data caps_data;
		struct veth_frames_data frames_data;
		struct veth_frames_ack_data frames_ack_data;
	} u;

};

#define DRV_NAME	"iseries_veth"
#define DRV_VERSION	"2.0"

#define VETH_NUMBUFFERS		(120)
#define VETH_ACKTIMEOUT 	(1000000) /* microseconds */
#define VETH_MAX_MCAST		(12)

#define VETH_MAX_MTU		(9000)

#if VETH_NUMBUFFERS < 10
#define ACK_THRESHOLD 		(1)
#elif VETH_NUMBUFFERS < 20
#define ACK_THRESHOLD 		(4)
#elif VETH_NUMBUFFERS < 40
#define ACK_THRESHOLD 		(10)
#else
#define ACK_THRESHOLD 		(20)
#endif

#define	VETH_STATE_SHUTDOWN	(0x0001)
#define VETH_STATE_OPEN		(0x0002)
#define VETH_STATE_RESET	(0x0004)
#define VETH_STATE_SENTMON	(0x0008)
#define VETH_STATE_SENTCAPS	(0x0010)
#define VETH_STATE_GOTCAPACK	(0x0020)
#define VETH_STATE_GOTCAPS	(0x0040)
#define VETH_STATE_SENTCAPACK	(0x0080)
#define VETH_STATE_READY	(0x0100)

struct veth_msg {
	struct veth_msg *next;
	struct veth_frames_data data;
	int token;
	int in_use;
	struct sk_buff *skb;
	struct device *dev;
};

struct veth_lpar_connection {
	HvLpIndex remote_lp;
	struct delayed_work statemachine_wq;
	struct veth_msg *msgs;
	int num_events;
	struct veth_cap_data local_caps;

	struct kobject kobject;
	struct timer_list ack_timer;

	struct timer_list reset_timer;
	unsigned int reset_timeout;
	unsigned long last_contact;
	int outstanding_tx;

	spinlock_t lock;
	unsigned long state;
	HvLpInstanceId src_inst;
	HvLpInstanceId dst_inst;
	struct veth_lpevent cap_event, cap_ack_event;
	u16 pending_acks[VETH_MAX_ACKS_PER_MSG];
	u32 num_pending_acks;

	int num_ack_events;
	struct veth_cap_data remote_caps;
	u32 ack_timeout;

	struct veth_msg *msg_stack_head;
};

struct veth_port {
	struct device *dev;
	u64 mac_addr;
	HvLpIndexMap lpar_map;

	/* queue_lock protects the stopped_map and dev's queue. */
	spinlock_t queue_lock;
	HvLpIndexMap stopped_map;

	/* mcast_gate protects promiscuous, num_mcast & mcast_addr. */
	rwlock_t mcast_gate;
	int promiscuous;
	int num_mcast;
	u64 mcast_addr[VETH_MAX_MCAST];

	struct kobject kobject;
};

static HvLpIndex this_lp;
static struct veth_lpar_connection *veth_cnx[HVMAXARCHITECTEDLPS]; /* = 0 */
static struct net_device *veth_dev[HVMAXARCHITECTEDVIRTUALLANS]; /* = 0 */

static int veth_start_xmit(struct sk_buff *skb, struct net_device *dev);
static void veth_recycle_msg(struct veth_lpar_connection *, struct veth_msg *);
static void veth_wake_queues(struct veth_lpar_connection *cnx);
static void veth_stop_queues(struct veth_lpar_connection *cnx);
static void veth_receive(struct veth_lpar_connection *, struct veth_lpevent *);
static void veth_release_connection(struct kobject *kobject);
static void veth_timed_ack(unsigned long ptr);
static void veth_timed_reset(unsigned long ptr);

/*
 * Utility functions
 */

#define veth_info(fmt, args...) \
	printk(KERN_INFO DRV_NAME ": " fmt, ## args)

#define veth_error(fmt, args...) \
	printk(KERN_ERR DRV_NAME ": Error: " fmt, ## args)

#ifdef DEBUG
#define veth_debug(fmt, args...) \
	printk(KERN_DEBUG DRV_NAME ": " fmt, ## args)
#else
#define veth_debug(fmt, args...) do {} while (0)
#endif

/* You must hold the connection's lock when you call this function. */
static inline void veth_stack_push(struct veth_lpar_connection *cnx,
				   struct veth_msg *msg)
{
	msg->next = cnx->msg_stack_head;
	cnx->msg_stack_head = msg;
}

/* You must hold the connection's lock when you call this function. */
static inline struct veth_msg *veth_stack_pop(struct veth_lpar_connection *cnx)
{
	struct veth_msg *msg;

	msg = cnx->msg_stack_head;
	if (msg)
		cnx->msg_stack_head = cnx->msg_stack_head->next;

	return msg;
}

/* You must hold the connection's lock when you call this function. */
static inline int veth_stack_is_empty(struct veth_lpar_connection *cnx)
{
	return cnx->msg_stack_head == NULL;
}

static inline HvLpEvent_Rc
veth_signalevent(struct veth_lpar_connection *cnx, u16 subtype,
		 HvLpEvent_AckInd ackind, HvLpEvent_AckType acktype,
		 u64 token,
		 u64 data1, u64 data2, u64 data3, u64 data4, u64 data5)
{
	return HvCallEvent_signalLpEventFast(cnx->remote_lp,
					     HvLpEvent_Type_VirtualLan,
					     subtype, ackind, acktype,
					     cnx->src_inst,
					     cnx->dst_inst,
					     token, data1, data2, data3,
					     data4, data5);
}

static inline HvLpEvent_Rc veth_signaldata(struct veth_lpar_connection *cnx,
					   u16 subtype, u64 token, void *data)
{
	u64 *p = (u64 *) data;

	return veth_signalevent(cnx, subtype, HvLpEvent_AckInd_NoAck,
				HvLpEvent_AckType_ImmediateAck,
				token, p[0], p[1], p[2], p[3], p[4]);
}

struct veth_allocation {
	struct completion c;
	int num;
};

static void veth_complete_allocation(void *parm, int number)
{
	struct veth_allocation *vc = (struct veth_allocation *)parm;

	vc->num = number;
	complete(&vc->c);
}

static int veth_allocate_events(HvLpIndex rlp, int number)
{
	struct veth_allocation vc = { COMPLETION_INITIALIZER(vc.c), 0 };

	mf_allocate_lp_events(rlp, HvLpEvent_Type_VirtualLan,
			    sizeof(struct veth_lpevent), number,
			    &veth_complete_allocation, &vc);
	wait_for_completion(&vc.c);

	return vc.num;
}

/*
 * sysfs support
 */

struct veth_cnx_attribute {
	struct attribute attr;
	ssize_t (*show)(struct veth_lpar_connection *, char *buf);
	ssize_t (*store)(struct veth_lpar_connection *, const char *buf);
};

static ssize_t veth_cnx_attribute_show(struct kobject *kobj,
		struct attribute *attr, char *buf)
{
	struct veth_cnx_attribute *cnx_attr;
	struct veth_lpar_connection *cnx;

	cnx_attr = container_of(attr, struct veth_cnx_attribute, attr);
	cnx = container_of(kobj, struct veth_lpar_connection, kobject);

	if (!cnx_attr->show)
		return -EIO;

	return cnx_attr->show(cnx, buf);
}

#define CUSTOM_CNX_ATTR(_name, _format, _expression)			\
static ssize_t _name##_show(struct veth_lpar_connection *cnx, char *buf)\
{									\
	return sprintf(buf, _format, _expression);			\
}									\
struct veth_cnx_attribute veth_cnx_attr_##_name = __ATTR_RO(_name)

#define SIMPLE_CNX_ATTR(_name)	\
	CUSTOM_CNX_ATTR(_name, "%lu\n", (unsigned long)cnx->_name)

SIMPLE_CNX_ATTR(outstanding_tx);
SIMPLE_CNX_ATTR(remote_lp);
SIMPLE_CNX_ATTR(num_events);
SIMPLE_CNX_ATTR(src_inst);
SIMPLE_CNX_ATTR(dst_inst);
SIMPLE_CNX_ATTR(num_pending_acks);
SIMPLE_CNX_ATTR(num_ack_events);
CUSTOM_CNX_ATTR(ack_timeout, "%d\n", jiffies_to_msecs(cnx->ack_timeout));
CUSTOM_CNX_ATTR(reset_timeout, "%d\n", jiffies_to_msecs(cnx->reset_timeout));
CUSTOM_CNX_ATTR(state, "0x%.4lX\n", cnx->state);
CUSTOM_CNX_ATTR(last_contact, "%d\n", cnx->last_contact ?
		jiffies_to_msecs(jiffies - cnx->last_contact) : 0);

#define GET_CNX_ATTR(_name)	(&veth_cnx_attr_##_name.attr)

static struct attribute *veth_cnx_default_attrs[] = {
	GET_CNX_ATTR(outstanding_tx),
	GET_CNX_ATTR(remote_lp),
	GET_CNX_ATTR(num_events),
	GET_CNX_ATTR(reset_timeout),
	GET_CNX_ATTR(last_contact),
	GET_CNX_ATTR(state),
	GET_CNX_ATTR(src_inst),
	GET_CNX_ATTR(dst_inst),
	GET_CNX_ATTR(num_pending_acks),
	GET_CNX_ATTR(num_ack_events),
	GET_CNX_ATTR(ack_timeout),
	NULL
};

static struct sysfs_ops veth_cnx_sysfs_ops = {
		.show = veth_cnx_attribute_show
};

static struct kobj_type veth_lpar_connection_ktype = {
	.release	= veth_release_connection,
	.sysfs_ops	= &veth_cnx_sysfs_ops,
	.default_attrs	= veth_cnx_default_attrs
};

struct veth_port_attribute {
	struct attribute attr;
	ssize_t (*show)(struct veth_port *, char *buf);
	ssize_t (*store)(struct veth_port *, const char *buf);
};

static ssize_t veth_port_attribute_show(struct kobject *kobj,
		struct attribute *attr, char *buf)
{
	struct veth_port_attribute *port_attr;
	struct veth_port *port;

	port_attr = container_of(attr, struct veth_port_attribute, attr);
	port = container_of(kobj, struct veth_port, kobject);

	if (!port_attr->show)
		return -EIO;

	return port_attr->show(port, buf);
}

#define CUSTOM_PORT_ATTR(_name, _format, _expression)			\
static ssize_t _name##_show(struct veth_port *port, char *buf)		\
{									\
	return sprintf(buf, _format, _expression);			\
}									\
struct veth_port_attribute veth_port_attr_##_name = __ATTR_RO(_name)

#define SIMPLE_PORT_ATTR(_name)	\
	CUSTOM_PORT_ATTR(_name, "%lu\n", (unsigned long)port->_name)

SIMPLE_PORT_ATTR(promiscuous);
SIMPLE_PORT_ATTR(num_mcast);
CUSTOM_PORT_ATTR(lpar_map, "0x%X\n", port->lpar_map);
CUSTOM_PORT_ATTR(stopped_map, "0x%X\n", port->stopped_map);
CUSTOM_PORT_ATTR(mac_addr, "0x%lX\n", port->mac_addr);

#define GET_PORT_ATTR(_name)	(&veth_port_attr_##_name.attr)
static struct attribute *veth_port_default_attrs[] = {
	GET_PORT_ATTR(mac_addr),
	GET_PORT_ATTR(lpar_map),
	GET_PORT_ATTR(stopped_map),
	GET_PORT_ATTR(promiscuous),
	GET_PORT_ATTR(num_mcast),
	NULL
};

static struct sysfs_ops veth_port_sysfs_ops = {
	.show = veth_port_attribute_show
};

static struct kobj_type veth_port_ktype = {
	.sysfs_ops	= &veth_port_sysfs_ops,
	.default_attrs	= veth_port_default_attrs
};

/*
 * LPAR connection code
 */

static inline void veth_kick_statemachine(struct veth_lpar_connection *cnx)
{
	schedule_delayed_work(&cnx->statemachine_wq, 0);
}

static void veth_take_cap(struct veth_lpar_connection *cnx,
			  struct veth_lpevent *event)
{
	unsigned long flags;

	spin_lock_irqsave(&cnx->lock, flags);
	/* Receiving caps may mean the other end has just come up, so
	 * we need to reload the instance ID of the far end */
	cnx->dst_inst =
		HvCallEvent_getTargetLpInstanceId(cnx->remote_lp,
						  HvLpEvent_Type_VirtualLan);

	if (cnx->state & VETH_STATE_GOTCAPS) {
		veth_error("Received a second capabilities from LPAR %d.\n",
			   cnx->remote_lp);
		event->base_event.xRc = HvLpEvent_Rc_BufferNotAvailable;
		HvCallEvent_ackLpEvent((struct HvLpEvent *) event);
	} else {
		memcpy(&cnx->cap_event, event, sizeof(cnx->cap_event));
		cnx->state |= VETH_STATE_GOTCAPS;
		veth_kick_statemachine(cnx);
	}
	spin_unlock_irqrestore(&cnx->lock, flags);
}

static void veth_take_cap_ack(struct veth_lpar_connection *cnx,
			      struct veth_lpevent *event)
{
	unsigned long flags;

	spin_lock_irqsave(&cnx->lock, flags);
	if (cnx->state & VETH_STATE_GOTCAPACK) {
		veth_error("Received a second capabilities ack from LPAR %d.\n",
			   cnx->remote_lp);
	} else {
		memcpy(&cnx->cap_ack_event, event,
		       sizeof(&cnx->cap_ack_event));
		cnx->state |= VETH_STATE_GOTCAPACK;
		veth_kick_statemachine(cnx);
	}
	spin_unlock_irqrestore(&cnx->lock, flags);
}

static void veth_take_monitor_ack(struct veth_lpar_connection *cnx,
				  struct veth_lpevent *event)
{
	unsigned long flags;

	spin_lock_irqsave(&cnx->lock, flags);
	veth_debug("cnx %d: lost connection.\n", cnx->remote_lp);

	/* Avoid kicking the statemachine once we're shutdown.
	 * It's unnecessary and it could break veth_stop_connection(). */

	if (! (cnx->state & VETH_STATE_SHUTDOWN)) {
		cnx->state |= VETH_STATE_RESET;
		veth_kick_statemachine(cnx);
	}
	spin_unlock_irqrestore(&cnx->lock, flags);
}

static void veth_handle_ack(struct veth_lpevent *event)
{
	HvLpIndex rlp = event->base_event.xTargetLp;
	struct veth_lpar_connection *cnx = veth_cnx[rlp];

	BUG_ON(! cnx);

	switch (event->base_event.xSubtype) {
	case VETH_EVENT_CAP:
		veth_take_cap_ack(cnx, event);
		break;
	case VETH_EVENT_MONITOR:
		veth_take_monitor_ack(cnx, event);
		break;
	default:
		veth_error("Unknown ack type %d from LPAR %d.\n",
				event->base_event.xSubtype, rlp);
	};
}

static void veth_handle_int(struct veth_lpevent *event)
{
	HvLpIndex rlp = event->base_event.xSourceLp;
	struct veth_lpar_connection *cnx = veth_cnx[rlp];
	unsigned long flags;
	int i, acked = 0;

	BUG_ON(! cnx);

	switch (event->base_event.xSubtype) {
	case VETH_EVENT_CAP:
		veth_take_cap(cnx, event);
		break;
	case VETH_EVENT_MONITOR:
		/* do nothing... this'll hang out here til we're dead,
		 * and the hypervisor will return it for us. */
		break;
	case VETH_EVENT_FRAMES_ACK:
		spin_lock_irqsave(&cnx->lock, flags);

		for (i = 0; i < VETH_MAX_ACKS_PER_MSG; ++i) {
			u16 msgnum = event->u.frames_ack_data.token[i];

			if (msgnum < VETH_NUMBUFFERS) {
				veth_recycle_msg(cnx, cnx->msgs + msgnum);
				cnx->outstanding_tx--;
				acked++;
			}
		}

		if (acked > 0) {
			cnx->last_contact = jiffies;
			veth_wake_queues(cnx);
		}

		spin_unlock_irqrestore(&cnx->lock, flags);