iseries_veth.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 1,424 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.
 *
 * 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/config.h>
#include <linux/module.h>
#include <linux/version.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 <asm/iSeries/mf.h>
#include <asm/iSeries/iSeries_pci.h>
#include <asm/uaccess.h>

#include <asm/iSeries/HvLpConfig.h>
#include <asm/iSeries/HvTypes.h>
#include <asm/iSeries/HvLpEvent.h>
#include <asm/iommu.h>
#include <asm/vio.h>

#include "iseries_veth.h"

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

#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 VethFramesData data;
	int token;
	unsigned long in_use;
	struct sk_buff *skb;
	struct device *dev;
};

struct veth_lpar_connection {
	HvLpIndex remote_lp;
	struct work_struct statemachine_wq;
	struct veth_msg *msgs;
	int num_events;
	struct VethCapData local_caps;

	struct timer_list ack_timer;

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

	int num_ack_events;
	struct VethCapData remote_caps;
	u32 ack_timeout;

	spinlock_t msg_stack_lock;
	struct veth_msg *msg_stack_head;
};

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

	spinlock_t pending_gate;
	struct sk_buff *pending_skb;
	HvLpIndexMap pending_lpmask;

	rwlock_t mcast_gate;
	int promiscuous;
	int all_mcast;
	int num_mcast;
	u64 mcast_addr[VETH_MAX_MCAST];
};

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_flush_pending(struct veth_lpar_connection *cnx);
static void veth_receive(struct veth_lpar_connection *, struct VethLpEvent *);
static void veth_timed_ack(unsigned long connectionPtr);

/*
 * Utility functions
 */

#define veth_printk(prio, fmt, args...) \
	printk(prio "%s: " fmt, __FILE__, ## args)

#define veth_error(fmt, args...) \
	printk(KERN_ERR "(%s:%3.3d) ERROR: " fmt, __FILE__, __LINE__ , ## args)

static inline void veth_stack_push(struct veth_lpar_connection *cnx,
				   struct veth_msg *msg)
{
	unsigned long flags;

	spin_lock_irqsave(&cnx->msg_stack_lock, flags);
	msg->next = cnx->msg_stack_head;
	cnx->msg_stack_head = msg;
	spin_unlock_irqrestore(&cnx->msg_stack_lock, flags);
}

static inline struct veth_msg *veth_stack_pop(struct veth_lpar_connection *cnx)
{
	unsigned long flags;
	struct veth_msg *msg;

	spin_lock_irqsave(&cnx->msg_stack_lock, flags);
	msg = cnx->msg_stack_head;
	if (msg)
		cnx->msg_stack_head = cnx->msg_stack_head->next;
	spin_unlock_irqrestore(&cnx->msg_stack_lock, flags);
	return msg;
}

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_allocateLpEvents(rlp, HvLpEvent_Type_VirtualLan,
			    sizeof(struct VethLpEvent), number,
			    &veth_complete_allocation, &vc);
	wait_for_completion(&vc.c);

	return vc.num;
}

/*
 * LPAR connection code
 */

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

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

	spin_lock_irqsave(&cnx->lock, flags);
	veth_printk(KERN_DEBUG, "Monitor ack returned for lpar %d\n",
		    cnx->remote_lp);
	cnx->state |= VETH_STATE_RESET;
	veth_kick_statemachine(cnx);
	spin_unlock_irqrestore(&cnx->lock, flags);
}

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

	BUG_ON(! cnx);

	switch (event->base_event.xSubtype) {
	case VethEventTypeCap:
		veth_take_cap_ack(cnx, event);
		break;
	case VethEventTypeMonitor:
		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 VethLpEvent *event)
{
	HvLpIndex rlp = event->base_event.xSourceLp;
	struct veth_lpar_connection *cnx = veth_cnx[rlp];
	unsigned long flags;
	int i;

	BUG_ON(! cnx);

	switch (event->base_event.xSubtype) {
	case VethEventTypeCap:
		veth_take_cap(cnx, event);
		break;
	case VethEventTypeMonitor:
		/* do nothing... this'll hang out here til we're dead,
		 * and the hypervisor will return it for us. */
		break;
	case VethEventTypeFramesAck:
		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);
		}
		spin_unlock_irqrestore(&cnx->lock, flags);
		veth_flush_pending(cnx);
		break;
	case VethEventTypeFrames:
		veth_receive(cnx, event);
		break;
	default:
		veth_error("Unknown interrupt type %d from lpar %d\n",
			   event->base_event.xSubtype, rlp);
	};
}

static void veth_handle_event(struct HvLpEvent *event, struct pt_regs *regs)
{
	struct VethLpEvent *veth_event = (struct VethLpEvent *)event;

	if (event->xFlags.xFunction == HvLpEvent_Function_Ack)
		veth_handle_ack(veth_event);
	else if (event->xFlags.xFunction == HvLpEvent_Function_Int)
		veth_handle_int(veth_event);
}

static int veth_process_caps(struct veth_lpar_connection *cnx)
{
	struct VethCapData *remote_caps = &cnx->remote_caps;
	int num_acks_needed;

	/* Convert timer to jiffies */
	cnx->ack_timeout = remote_caps->ack_timeout * HZ / 1000000;

	if ( (remote_caps->num_buffers == 0)
	     || (remote_caps->ack_threshold > VETH_MAX_ACKS_PER_MSG)
	     || (remote_caps->ack_threshold == 0)
	     || (cnx->ack_timeout == 0) ) {
		veth_error("Received incompatible capabilities from lpar %d\n",
			   cnx->remote_lp);
		return HvLpEvent_Rc_InvalidSubtypeData;
	}

	num_acks_needed = (remote_caps->num_buffers
			   / remote_caps->ack_threshold) + 1;

	/* FIXME: locking on num_ack_events? */
	if (cnx->num_ack_events < num_acks_needed) {
		int num;

		num = veth_allocate_events(cnx->remote_lp,
					   num_acks_needed-cnx->num_ack_events);
		if (num > 0)
			cnx->num_ack_events += num;

		if (cnx->num_ack_events < num_acks_needed) {
			veth_error("Couldn't allocate enough ack events for lpar %d\n",
				   cnx->remote_lp);

			return HvLpEvent_Rc_BufferNotAvailable;
		}
	}


	return HvLpEvent_Rc_Good;
}

/* FIXME: The gotos here are a bit dubious */
static void veth_statemachine(void *p)
{
	struct veth_lpar_connection *cnx = (struct veth_lpar_connection *)p;
	int rlp = cnx->remote_lp;
	int rc;

	spin_lock_irq(&cnx->lock);

 restart:
	if (cnx->state & VETH_STATE_RESET) {
		int i;

		del_timer(&cnx->ack_timer);

		if (cnx->state & VETH_STATE_OPEN)
			HvCallEvent_closeLpEventPath(cnx->remote_lp,
						     HvLpEvent_Type_VirtualLan);

		/* reset ack data */
		memset(&cnx->pending_acks, 0xff, sizeof (cnx->pending_acks));
		cnx->num_pending_acks = 0;

		cnx->state &= ~(VETH_STATE_RESET | VETH_STATE_SENTMON
				| VETH_STATE_OPEN | VETH_STATE_SENTCAPS
				| VETH_STATE_GOTCAPACK | VETH_STATE_GOTCAPS
				| VETH_STATE_SENTCAPACK | VETH_STATE_READY);

		/* Clean up any leftover messages */
		if (cnx->msgs)
			for (i = 0; i < VETH_NUMBUFFERS; ++i)
				veth_recycle_msg(cnx, cnx->msgs + i);
		spin_unlock_irq(&cnx->lock);
		veth_flush_pending(cnx);
		spin_lock_irq(&cnx->lock);
		if (cnx->state & VETH_STATE_RESET)
			goto restart;
	}

	if (cnx->state & VETH_STATE_SHUTDOWN)
		/* It's all over, do nothing */
		goto out;

	if ( !(cnx->state & VETH_STATE_OPEN) ) {