Linux-2.6.12-rc2v2.6.12-rc2

Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!

8 files changed, 1775 insertions, 0 deletions

diff --git a/drivers/block/aoe/Makefile b/drivers/block/aoe/Makefile
new file mode 100644
index 00000000000..e76d997183c
--- /dev/null
+++ b/drivers/block/aoe/Makefile
@@ -0,0 +1,6 @@
+#
+# Makefile for ATA over Ethernet
+#
+
+obj-$(CONFIG_ATA_OVER_ETH)	+= aoe.o
+aoe-objs := aoeblk.o aoechr.o aoecmd.o aoedev.o aoemain.o aoenet.o
diff --git a/drivers/block/aoe/aoe.h b/drivers/block/aoe/aoe.h
new file mode 100644
index 00000000000..db78f826d40
--- /dev/null
+++ b/drivers/block/aoe/aoe.h
@@ -0,0 +1,165 @@
+/* Copyright (c) 2004 Coraid, Inc.  See COPYING for GPL terms. */
+#define VERSION "5"
+#define AOE_MAJOR 152
+#define DEVICE_NAME "aoe"
+#ifndef AOE_PARTITIONS
+#define AOE_PARTITIONS 16
+#endif
+#define SYSMINOR(aoemajor, aoeminor) ((aoemajor) * 10 + (aoeminor))
+#define AOEMAJOR(sysminor) ((sysminor) / 10)
+#define AOEMINOR(sysminor) ((sysminor) % 10)
+#define WHITESPACE " \t\v\f\n"
+
+enum {
+	AOECMD_ATA,
+	AOECMD_CFG,
+
+	AOEFL_RSP = (1<<3),
+	AOEFL_ERR = (1<<2),
+
+	AOEAFL_EXT = (1<<6),
+	AOEAFL_DEV = (1<<4),
+	AOEAFL_ASYNC = (1<<1),
+	AOEAFL_WRITE = (1<<0),
+
+	AOECCMD_READ = 0,
+	AOECCMD_TEST,
+	AOECCMD_PTEST,
+	AOECCMD_SET,
+	AOECCMD_FSET,
+
+	AOE_HVER = 0x10,
+};
+
+struct aoe_hdr {
+	unsigned char dst[6];
+	unsigned char src[6];
+	unsigned char type[2];
+	unsigned char verfl;
+	unsigned char err;
+	unsigned char major[2];
+	unsigned char minor;
+	unsigned char cmd;
+	unsigned char tag[4];
+};
+
+struct aoe_atahdr {
+	unsigned char aflags;
+	unsigned char errfeat;
+	unsigned char scnt;
+	unsigned char cmdstat;
+	unsigned char lba0;
+	unsigned char lba1;
+	unsigned char lba2;
+	unsigned char lba3;
+	unsigned char lba4;
+	unsigned char lba5;
+	unsigned char res[2];
+};
+
+struct aoe_cfghdr {
+	unsigned char bufcnt[2];
+	unsigned char fwver[2];
+	unsigned char res;
+	unsigned char aoeccmd;
+	unsigned char cslen[2];
+};
+
+enum {
+	DEVFL_UP = 1,	/* device is installed in system and ready for AoE->ATA commands */
+	DEVFL_TKILL = (1<<1),	/* flag for timer to know when to kill self */
+	DEVFL_EXT = (1<<2),	/* device accepts lba48 commands */
+	DEVFL_CLOSEWAIT = (1<<3), /* device is waiting for all closes to revalidate */
+	DEVFL_WC_UPDATE = (1<<4), /* this device needs to update write cache status */
+	DEVFL_WORKON = (1<<4),
+
+	BUFFL_FAIL = 1,
+};
+
+enum {
+	MAXATADATA = 1024,
+	NPERSHELF = 10,
+	FREETAG = -1,
+	MIN_BUFS = 8,
+};
+
+struct buf {
+	struct list_head bufs;
+	ulong flags;
+	ulong nframesout;
+	char *bufaddr;
+	ulong resid;
+	ulong bv_resid;
+	sector_t sector;
+	struct bio *bio;
+	struct bio_vec *bv;
+};
+
+struct frame {
+	int tag;
+	ulong waited;
+	struct buf *buf;
+	char *bufaddr;
+	int writedatalen;
+	int ndata;
+
+	/* largest possible */
+	unsigned char data[sizeof(struct aoe_hdr) + sizeof(struct aoe_atahdr)];
+};
+
+struct aoedev {
+	struct aoedev *next;
+	unsigned char addr[6];	/* remote mac addr */
+	ushort flags;
+	ulong sysminor;
+	ulong aoemajor;
+	ulong aoeminor;
+	ulong nopen;		/* (bd_openers isn't available without sleeping) */
+	ulong rttavg;		/* round trip average of requests/responses */
+	u16 fw_ver;		/* version of blade's firmware */
+	struct work_struct work;/* disk create work struct */
+	struct gendisk *gd;
+	request_queue_t blkq;
+	struct hd_geometry geo; 
+	sector_t ssize;
+	struct timer_list timer;
+	spinlock_t lock;
+	struct net_device *ifp;	/* interface ed is attached to */
+	struct sk_buff *skblist;/* packets needing to be sent */
+	mempool_t *bufpool;	/* for deadlock-free Buf allocation */
+	struct list_head bufq;	/* queue of bios to work on */
+	struct buf *inprocess;	/* the one we're currently working on */
+	ulong lasttag;		/* last tag sent */
+	ulong nframes;		/* number of frames below */
+	struct frame *frames;
+};
+
+
+int aoeblk_init(void);
+void aoeblk_exit(void);
+void aoeblk_gdalloc(void *);
+void aoedisk_rm_sysfs(struct aoedev *d);
+
+int aoechr_init(void);
+void aoechr_exit(void);
+void aoechr_error(char *);
+
+void aoecmd_work(struct aoedev *d);
+void aoecmd_cfg(ushort, unsigned char);
+void aoecmd_ata_rsp(struct sk_buff *);
+void aoecmd_cfg_rsp(struct sk_buff *);
+
+int aoedev_init(void);
+void aoedev_exit(void);
+struct aoedev *aoedev_bymac(unsigned char *);
+void aoedev_downdev(struct aoedev *d);
+struct aoedev *aoedev_set(ulong, unsigned char *, struct net_device *, ulong);
+int aoedev_busy(void);
+
+int aoenet_init(void);
+void aoenet_exit(void);
+void aoenet_xmit(struct sk_buff *);
+int is_aoe_netif(struct net_device *ifp);
+int set_aoe_iflist(const char __user *str, size_t size);
+
+u64 mac_addr(char addr[6]);
diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c
new file mode 100644
index 00000000000..63561b280bc
--- /dev/null
+++ b/drivers/block/aoe/aoeblk.c
@@ -0,0 +1,267 @@
+/* Copyright (c) 2004 Coraid, Inc.  See COPYING for GPL terms. */
+/*
+ * aoeblk.c
+ * block device routines
+ */
+
+#include <linux/hdreg.h>
+#include <linux/blkdev.h>
+#include <linux/fs.h>
+#include <linux/ioctl.h>
+#include <linux/genhd.h>
+#include <linux/netdevice.h>
+#include "aoe.h"
+
+static kmem_cache_t *buf_pool_cache;
+
+/* add attributes for our block devices in sysfs */
+static ssize_t aoedisk_show_state(struct gendisk * disk, char *page)
+{
+	struct aoedev *d = disk->private_data;
+
+	return snprintf(page, PAGE_SIZE,
+			"%s%s\n",
+			(d->flags & DEVFL_UP) ? "up" : "down",
+			(d->flags & DEVFL_CLOSEWAIT) ? ",closewait" : "");
+}
+static ssize_t aoedisk_show_mac(struct gendisk * disk, char *page)
+{
+	struct aoedev *d = disk->private_data;
+
+	return snprintf(page, PAGE_SIZE, "%012llx\n",
+			(unsigned long long)mac_addr(d->addr));
+}
+static ssize_t aoedisk_show_netif(struct gendisk * disk, char *page)
+{
+	struct aoedev *d = disk->private_data;
+
+	return snprintf(page, PAGE_SIZE, "%s\n", d->ifp->name);
+}
+
+static struct disk_attribute disk_attr_state = {
+	.attr = {.name = "state", .mode = S_IRUGO },
+	.show = aoedisk_show_state
+};
+static struct disk_attribute disk_attr_mac = {
+	.attr = {.name = "mac", .mode = S_IRUGO },
+	.show = aoedisk_show_mac
+};
+static struct disk_attribute disk_attr_netif = {
+	.attr = {.name = "netif", .mode = S_IRUGO },
+	.show = aoedisk_show_netif
+};
+
+static void
+aoedisk_add_sysfs(struct aoedev *d)
+{
+	sysfs_create_file(&d->gd->kobj, &disk_attr_state.attr);
+	sysfs_create_file(&d->gd->kobj, &disk_attr_mac.attr);
+	sysfs_create_file(&d->gd->kobj, &disk_attr_netif.attr);
+}
+void
+aoedisk_rm_sysfs(struct aoedev *d)
+{
+	sysfs_remove_link(&d->gd->kobj, "state");
+	sysfs_remove_link(&d->gd->kobj, "mac");
+	sysfs_remove_link(&d->gd->kobj, "netif");
+}
+
+static int
+aoeblk_open(struct inode *inode, struct file *filp)
+{
+	struct aoedev *d;
+	ulong flags;
+
+	d = inode->i_bdev->bd_disk->private_data;
+
+	spin_lock_irqsave(&d->lock, flags);
+	if (d->flags & DEVFL_UP) {
+		d->nopen++;
+		spin_unlock_irqrestore(&d->lock, flags);
+		return 0;
+	}
+	spin_unlock_irqrestore(&d->lock, flags);
+	return -ENODEV;
+}
+
+static int
+aoeblk_release(struct inode *inode, struct file *filp)
+{
+	struct aoedev *d;
+	ulong flags;
+
+	d = inode->i_bdev->bd_disk->private_data;
+
+	spin_lock_irqsave(&d->lock, flags);
+
+	if (--d->nopen == 0 && (d->flags & DEVFL_CLOSEWAIT)) {
+		d->flags &= ~DEVFL_CLOSEWAIT;
+		spin_unlock_irqrestore(&d->lock, flags);
+		aoecmd_cfg(d->aoemajor, d->aoeminor);
+		return 0;
+	}
+	spin_unlock_irqrestore(&d->lock, flags);
+
+	return 0;
+}
+
+static int
+aoeblk_make_request(request_queue_t *q, struct bio *bio)
+{
+	struct aoedev *d;
+	struct buf *buf;
+	struct sk_buff *sl;
+	ulong flags;
+
+	blk_queue_bounce(q, &bio);
+
+	d = bio->bi_bdev->bd_disk->private_data;
+	buf = mempool_alloc(d->bufpool, GFP_NOIO);
+	if (buf == NULL) {
+		printk(KERN_INFO "aoe: aoeblk_make_request: buf allocation "
+			"failure\n");
+		bio_endio(bio, bio->bi_size, -ENOMEM);
+		return 0;
+	}
+	memset(buf, 0, sizeof(*buf));
+	INIT_LIST_HEAD(&buf->bufs);
+	buf->bio = bio;
+	buf->resid = bio->bi_size;
+	buf->sector = bio->bi_sector;
+	buf->bv = buf->bio->bi_io_vec;
+	buf->bv_resid = buf->bv->bv_len;
+	buf->bufaddr = page_address(buf->bv->bv_page) + buf->bv->bv_offset;
+
+	spin_lock_irqsave(&d->lock, flags);
+
+	if ((d->flags & DEVFL_UP) == 0) {
+		printk(KERN_INFO "aoe: aoeblk_make_request: device %ld.%ld is not up\n",
+			d->aoemajor, d->aoeminor);
+		spin_unlock_irqrestore(&d->lock, flags);
+		mempool_free(buf, d->bufpool);
+		bio_endio(bio, bio->bi_size, -ENXIO);
+		return 0;
+	}
+
+	list_add_tail(&buf->bufs, &d->bufq);
+	aoecmd_work(d);
+
+	sl = d->skblist;
+	d->skblist = NULL;
+
+	spin_unlock_irqrestore(&d->lock, flags);
+
+	aoenet_xmit(sl);
+	return 0;
+}
+
+/* This ioctl implementation expects userland to have the device node
+ * permissions set so that only priviledged users can open an aoe
+ * block device directly.
+ */
+static int
+aoeblk_ioctl(struct inode *inode, struct file *filp, uint cmd, ulong arg)
+{
+	struct aoedev *d;
+
+	if (!arg)
+		return -EINVAL;
+
+	d = inode->i_bdev->bd_disk->private_data;
+	if ((d->flags & DEVFL_UP) == 0) {
+		printk(KERN_ERR "aoe: aoeblk_ioctl: disk not up\n");
+		return -ENODEV;
+	}
+
+	if (cmd == HDIO_GETGEO) {
+		d->geo.start = get_start_sect(inode->i_bdev);
+		if (!copy_to_user((void __user *) arg, &d->geo, sizeof d->geo))
+			return 0;
+		return -EFAULT;
+	}
+	printk(KERN_INFO "aoe: aoeblk_ioctl: unknown ioctl %d\n", cmd);
+	return -EINVAL;
+}
+
+static struct block_device_operations aoe_bdops = {
+	.open = aoeblk_open,
+	.release = aoeblk_release,
+	.ioctl = aoeblk_ioctl,
+	.owner = THIS_MODULE,
+};
+
+/* alloc_disk and add_disk can sleep */
+void
+aoeblk_gdalloc(void *vp)
+{
+	struct aoedev *d = vp;
+	struct gendisk *gd;
+	ulong flags;
+
+	gd = alloc_disk(AOE_PARTITIONS);
+	if (gd == NULL) {
+		printk(KERN_ERR "aoe: aoeblk_gdalloc: cannot allocate disk "
+			"structure for %ld.%ld\n", d->aoemajor, d->aoeminor);
+		spin_lock_irqsave(&d->lock, flags);
+		d->flags &= ~DEVFL_WORKON;
+		spin_unlock_irqrestore(&d->lock, flags);
+		return;
+	}
+
+	d->bufpool = mempool_create(MIN_BUFS,
+				    mempool_alloc_slab, mempool_free_slab,
+				    buf_pool_cache);
+	if (d->bufpool == NULL) {
+		printk(KERN_ERR "aoe: aoeblk_gdalloc: cannot allocate bufpool "
+			"for %ld.%ld\n", d->aoemajor, d->aoeminor);
+		put_disk(gd);
+		spin_lock_irqsave(&d->lock, flags);
+		d->flags &= ~DEVFL_WORKON;
+		spin_unlock_irqrestore(&d->lock, flags);
+		return;
+	}
+
+	spin_lock_irqsave(&d->lock, flags);
+	blk_queue_make_request(&d->blkq, aoeblk_make_request);
+	gd->major = AOE_MAJOR;
+	gd->first_minor = d->sysminor * AOE_PARTITIONS;
+	gd->fops = &aoe_bdops;
+	gd->private_data = d;
+	gd->capacity = d->ssize;
+	snprintf(gd->disk_name, sizeof gd->disk_name, "etherd/e%ld.%ld",
+		d->aoemajor, d->aoeminor);
+
+	gd->queue = &d->blkq;
+	d->gd = gd;
+	d->flags &= ~DEVFL_WORKON;
+	d->flags |= DEVFL_UP;
+
+	spin_unlock_irqrestore(&d->lock, flags);
+
+	add_disk(gd);
+	aoedisk_add_sysfs(d);
+	
+	printk(KERN_INFO "aoe: %012llx e%lu.%lu v%04x has %llu "
+		"sectors\n", (unsigned long long)mac_addr(d->addr),
+		d->aoemajor, d->aoeminor,
+		d->fw_ver, (long long)d->ssize);
+}
+
+void
+aoeblk_exit(void)
+{
+	kmem_cache_destroy(buf_pool_cache);
+}
+
+int __init
+aoeblk_init(void)
+{
+	buf_pool_cache = kmem_cache_create("aoe_bufs", 
+					   sizeof(struct buf),
+					   0, 0, NULL, NULL);
+	if (buf_pool_cache == NULL)
+		return -ENOMEM;
+
+	return 0;
+}
+
diff --git a/drivers/block/aoe/aoechr.c b/drivers/block/aoe/aoechr.c
new file mode 100644
index 00000000000..14aeca3e2e8
--- /dev/null
+++ b/drivers/block/aoe/aoechr.c
@@ -0,0 +1,244 @@
+/* Copyright (c) 2004 Coraid, Inc.  See COPYING for GPL terms. */
+/*
+ * aoechr.c
+ * AoE character device driver
+ */
+
+#include <linux/hdreg.h>
+#include <linux/blkdev.h>
+#include "aoe.h"
+
+enum {
+	//MINOR_STAT = 1, (moved to sysfs)
+	MINOR_ERR = 2,
+	MINOR_DISCOVER,
+	MINOR_INTERFACES,
+	MSGSZ = 2048,
+	NARGS = 10,
+	NMSG = 100,		/* message backlog to retain */
+};
+
+struct aoe_chardev {
+	ulong minor;
+	char name[32];
+};
+
+enum { EMFL_VALID = 1 };
+
+struct ErrMsg {
+	short flags;
+	short len;
+	char *msg;
+};
+
+static struct ErrMsg emsgs[NMSG];
+static int emsgs_head_idx, emsgs_tail_idx;
+static struct semaphore emsgs_sema;
+static spinlock_t emsgs_lock;
+static int nblocked_emsgs_readers;
+static struct class_simple *aoe_class;
+static struct aoe_chardev chardevs[] = {
+	{ MINOR_ERR, "err" },
+	{ MINOR_DISCOVER, "discover" },
+	{ MINOR_INTERFACES, "interfaces" },
+};
+
+static int
+discover(void)
+{
+	aoecmd_cfg(0xffff, 0xff);
+	return 0;
+}
+
+static int
+interfaces(const char __user *str, size_t size)
+{
+	if (set_aoe_iflist(str, size)) {
+		printk(KERN_CRIT
+		       "%s: could not set interface list: %s\n",
+		       __FUNCTION__, "too many interfaces");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+void
+aoechr_error(char *msg)
+{
+	struct ErrMsg *em;
+	char *mp;
+	ulong flags, n;
+
+	n = strlen(msg);
+
+	spin_lock_irqsave(&emsgs_lock, flags);
+
+	em = emsgs + emsgs_tail_idx;
+	if ((em->flags & EMFL_VALID)) {
+bail:		spin_unlock_irqrestore(&emsgs_lock, flags);
+		return;
+	}
+
+	mp = kmalloc(n, GFP_ATOMIC);
+	if (mp == NULL) {
+		printk(KERN_CRIT "aoe: aoechr_error: allocation failure, len=%ld\n", n);
+		goto bail;
+	}
+
+	memcpy(mp, msg, n);
+	em->msg = mp;
+	em->flags |= EMFL_VALID;
+	em->len = n;
+
+	emsgs_tail_idx++;
+	emsgs_tail_idx %= ARRAY_SIZE(emsgs);
+
+	spin_unlock_irqrestore(&emsgs_lock, flags);
+
+	if (nblocked_emsgs_readers)
+		up(&emsgs_sema);
+}
+
+static ssize_t
+aoechr_write(struct file *filp, const char __user *buf, size_t cnt, loff_t *offp)
+{
+	int ret = -EINVAL;
+
+	switch ((unsigned long) filp->private_data) {
+	default:
+		printk(KERN_INFO "aoe: aoechr_write: can't write to that file.\n");
+		break;
+	case MINOR_DISCOVER:
+		ret = discover();
+		break;
+	case MINOR_INTERFACES:
+		ret = interfaces(buf, cnt);
+		break;
+	}
+	if (ret == 0)
+		ret = cnt;
+	return ret;
+}
+
+static int
+aoechr_open(struct inode *inode, struct file *filp)
+{
+	int n, i;
+
+	n = MINOR(inode->i_rdev);
+	filp->private_data = (void *) (unsigned long) n;
+
+	for (i = 0; i < ARRAY_SIZE(chardevs); ++i)
+		if (chardevs[i].minor == n)
+			return 0;
+	return -EINVAL;
+}
+
+static int
+aoechr_rel(struct inode *inode, struct file *filp)
+{
+	return 0;
+}
+
+static ssize_t
+aoechr_read(struct file *filp, char __user *buf, size_t cnt, loff_t *off)
+{
+	unsigned long n;
+	char *mp;
+	struct ErrMsg *em;
+	ssize_t len;
+	ulong flags;
+
+	n = (unsigned long) filp->private_data;
+	switch (n) {
+	case MINOR_ERR:
+		spin_lock_irqsave(&emsgs_lock, flags);
+loop:
+		em = emsgs + emsgs_head_idx;
+		if ((em->flags & EMFL_VALID) == 0) {
+			if (filp->f_flags & O_NDELAY) {
+				spin_unlock_irqrestore(&emsgs_lock, flags);
+				return -EAGAIN;
+			}
+			nblocked_emsgs_readers++;
+
+			spin_unlock_irqrestore(&emsgs_lock, flags);
+
+			n = down_interruptible(&emsgs_sema);
+
+			spin_lock_irqsave(&emsgs_lock, flags);
+
+			nblocked_emsgs_readers--;
+
+			if (n) {
+				spin_unlock_irqrestore(&emsgs_lock, flags);
+				return -ERESTARTSYS;
+			}
+			goto loop;
+		}
+		if (em->len > cnt) {
+			spin_unlock_irqrestore(&emsgs_lock, flags);
+			return -EAGAIN;
+		}
+		mp = em->msg;
+		len = em->len;
+		em->msg = NULL;
+		em->flags &= ~EMFL_VALID;
+
+		emsgs_head_idx++;
+		emsgs_head_idx %= ARRAY_SIZE(emsgs);
+
+		spin_unlock_irqrestore(&emsgs_lock, flags);
+
+		n = copy_to_user(buf, mp, len);
+		kfree(mp);
+		return n == 0 ? len : -EFAULT;
+	default:
+		return -EFAULT;
+	}
+}
+
+static struct file_operations aoe_fops = {
+	.write = aoechr_write,
+	.read = aoechr_read,
+	.open = aoechr_open,
+	.release = aoechr_rel,
+	.owner = THIS_MODULE,
+};
+
+int __init
+aoechr_init(void)
+{
+	int n, i;
+
+	n = register_chrdev(AOE_MAJOR, "aoechr", &aoe_fops);
+	if (n < 0) { 
+		printk(KERN_ERR "aoe: aoechr_init: can't register char device\n");
+		return n;
+	}
+	sema_init(&emsgs_sema, 0);
+	spin_lock_init(&emsgs_lock);
+	aoe_class = class_simple_create(THIS_MODULE, "aoe");
+	if (IS_ERR(aoe_class)) {
+		unregister_chrdev(AOE_MAJOR, "aoechr");
+		return PTR_ERR(aoe_class);
+	}
+	for (i = 0; i < ARRAY_SIZE(chardevs); ++i)
+		class_simple_device_add(aoe_class,
+					MKDEV(AOE_MAJOR, chardevs[i].minor),
+					NULL, chardevs[i].name);
+
+	return 0;
+}
+
+void
+aoechr_exit(void)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(chardevs); ++i)
+		class_simple_device_remove(MKDEV(AOE_MAJOR, chardevs[i].minor));
+	class_simple_destroy(aoe_class);
+	unregister_chrdev(AOE_MAJOR, "aoechr");
+}
+
diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c
new file mode 100644
index 00000000000..fb6d942a456
--- /dev/null
+++ b/drivers/block/aoe/aoecmd.c
@@ -0,0 +1,629 @@
+/* Copyright (c) 2004 Coraid, Inc.  See COPYING for GPL terms. */
+/*
+ * aoecmd.c
+ * Filesystem request handling methods
+ */
+
+#include <linux/hdreg.h>
+#include <linux/blkdev.h>
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include "aoe.h"
+
+#define TIMERTICK (HZ / 10)
+#define MINTIMER (2 * TIMERTICK)
+#define MAXTIMER (HZ << 1)
+#define MAXWAIT (60 * 3)	/* After MAXWAIT seconds, give up and fail dev */
+
+static struct sk_buff *
+new_skb(struct net_device *if_dev, ulong len)
+{
+	struct sk_buff *skb;
+
+	skb = alloc_skb(len, GFP_ATOMIC);
+	if (skb) {
+		skb->nh.raw = skb->mac.raw = skb->data;
+		skb->dev = if_dev;
+		skb->protocol = __constant_htons(ETH_P_AOE);
+		skb->priority = 0;
+		skb_put(skb, len);
+		skb->next = skb->prev = NULL;
+
+		/* tell the network layer not to perform IP checksums
+		 * or to get the NIC to do it
+		 */
+		skb->ip_summed = CHECKSUM_NONE;
+	}
+	return skb;
+}
+
+static struct sk_buff *
+skb_prepare(struct aoedev *d, struct frame *f)
+{
+	struct sk_buff *skb;
+	char *p;
+
+	skb = new_skb(d->ifp, f->ndata + f->writedatalen);
+	if (!skb) {
+		printk(KERN_INFO "aoe: skb_prepare: failure to allocate skb\n");
+		return NULL;
+	}
+
+	p = skb->mac.raw;
+	memcpy(p, f->data, f->ndata);
+
+	if (f->writedatalen) {
+		p += sizeof(struct aoe_hdr) + sizeof(struct aoe_atahdr);
+		memcpy(p, f->bufaddr, f->writedatalen);
+	}
+
+	return skb;
+}
+
+static struct frame *
+getframe(struct aoedev *d, int tag)
+{
+	struct frame *f, *e;
+
+	f = d->frames;
+	e = f + d->nframes;
+	for (; f<e; f++)
+		if (f->tag == tag)
+			return f;
+	return NULL;
+}
+
+/*
+ * Leave the top bit clear so we have tagspace for userland.
+ * The bottom 16 bits are the xmit tick for rexmit/rttavg processing.
+ * This driver reserves tag -1 to mean "unused frame."
+ */
+static int
+newtag(struct aoedev *d)
+{
+	register ulong n;
+
+	n = jiffies & 0xffff;
+	return n |= (++d->lasttag & 0x7fff) << 16;
+}
+
+static int
+aoehdr_atainit(struct aoedev *d, struct aoe_hdr *h)
+{
+	u16 type = __constant_cpu_to_be16(ETH_P_AOE);
+	u16 aoemajor = __cpu_to_be16(d->aoemajor);
+	u32 host_tag = newtag(d);
+	u32 tag = __cpu_to_be32(host_tag);
+
+	memcpy(h->src, d->ifp->dev_addr, sizeof h->src);
+	memcpy(h->dst, d->addr, sizeof h->dst);
+	memcpy(h->type, &type, sizeof type);
+	h->verfl = AOE_HVER;
+	memcpy(h->major, &aoemajor, sizeof aoemajor);
+	h->minor = d->aoeminor;
+	h->cmd = AOECMD_ATA;
+	memcpy(h->tag, &tag, sizeof tag);
+
+	return host_tag;
+}
+
+static void
+aoecmd_ata_rw(struct aoedev *d, struct frame *f)
+{
+	struct aoe_hdr *h;
+	struct aoe_atahdr *ah;
+	struct buf *buf;
+	struct sk_buff *skb;
+	ulong bcnt;
+	register sector_t sector;
+	char writebit, extbit;
+
+	writebit = 0x10;
+	extbit = 0x4;
+
+	buf = d->inprocess;
+
+	sector = buf->sector;
+	bcnt = buf->bv_resid;
+	if (bcnt > MAXATADATA)
+		bcnt = MAXATADATA;
+
+	/* initialize the headers & frame */
+	h = (struct aoe_hdr *) f->data;
+	ah = (struct aoe_atahdr *) (h+1);
+	f->ndata = sizeof *h + sizeof *ah;
+	memset(h, 0, f->ndata);
+	f->tag = aoehdr_atainit(d, h);
+	f->waited = 0;
+	f->buf = buf;
+	f->bufaddr = buf->bufaddr;
+
+	/* set up ata header */
+	ah->scnt = bcnt >> 9;
+	ah->lba0 = sector;
+	ah->lba1 = sector >>= 8;
+	ah->lba2 = sector >>= 8;
+	ah->lba3 = sector >>= 8;
+	if (d->flags & DEVFL_EXT) {
+		ah->aflags |= AOEAFL_EXT;
+		ah->lba4 = sector >>= 8;
+		ah->lba5 = sector >>= 8;
+	} else {
+		extbit = 0;
+		ah->lba3 &= 0x0f;
+		ah->lba3 |= 0xe0;	/* LBA bit + obsolete 0xa0 */
+	}
+
+	if (bio_data_dir(buf->bio) == WRITE) {
+		ah->aflags |= AOEAFL_WRITE;
+		f->writedatalen = bcnt;
+	} else {
+		writebit = 0;
+		f->writedatalen = 0;
+	}
+
+	ah->cmdstat = WIN_READ | writebit | extbit;
+
+	/* mark all tracking fields and load out */
+	buf->nframesout += 1;
+	buf->bufaddr += bcnt;
+	buf->bv_resid -= bcnt;
+/* printk(KERN_INFO "aoe: bv_resid=%ld\n", buf->bv_resid); */
+	buf->resid -= bcnt;
+	buf->sector += bcnt >> 9;
+	if (buf->resid == 0) {
+		d->inprocess = NULL;
+	} else if (buf->bv_resid == 0) {
+		buf->bv++;
+		buf->bv_resid = buf->bv->bv_len;
+		buf->bufaddr = page_address(buf->bv->bv_page) + buf->bv->bv_offset;
+	}
+
+	skb = skb_prepare(d, f);
+	if (skb) {
+		skb->next = d->skblist;
+		d->skblist = skb;
+	}
+}
+
+/* enters with d->lock held */
+void
+aoecmd_work(struct aoedev *d)
+{
+	struct frame *f;
+	struct buf *buf;
+loop:
+	f = getframe(d, FREETAG);
+	if (f == NULL)
+		return;
+	if (d->inprocess == NULL) {
+		if (list_empty(&d->bufq))
+			return;
+		buf = container_of(d->bufq.next, struct buf, bufs);
+		list_del(d->bufq.next);
+/*printk(KERN_INFO "aoecmd_work: bi_size=%ld\n", buf->bio->bi_size); */
+		d->inprocess = buf;
+	}
+	aoecmd_ata_rw(d, f);
+	goto loop;
+}
+
+static void
+rexmit(struct aoedev *d, struct frame *f)
+{
+	struct sk_buff *skb;
+	struct aoe_hdr *h;
+	char buf[128];
+	u32 n;
+	u32 net_tag;
+
+	n = newtag(d);
+
+	snprintf(buf, sizeof buf,
+		"%15s e%ld.%ld oldtag=%08x@%08lx newtag=%08x\n",
+		"retransmit",
+		d->aoemajor, d->aoeminor, f->tag, jiffies, n);
+	aoechr_error(buf);
+
+	h = (struct aoe_hdr *) f->data;
+	f->tag = n;
+	net_tag = __cpu_to_be32(n);
+	memcpy(h->tag, &net_tag, sizeof net_tag);
+
+	skb = skb_prepare(d, f);
+	if (skb) {
+		skb->next = d->skblist;
+		d->skblist = skb;
+	}
+}
+
+static int
+tsince(int tag)
+{
+	int n;
+
+	n = jiffies & 0xffff;
+	n -= tag & 0xffff;
+	if (n < 0)
+		n += 1<<16;
+	return n;
+}
+
+static void
+rexmit_timer(ulong vp)
+{
+	struct aoedev *d;
+	struct frame *f, *e;
+	struct sk_buff *sl;
+	register long timeout;
+	ulong flags, n;
+
+	d = (struct aoedev *) vp;
+	sl = NULL;
+
+	/* timeout is always ~150% of the moving average */
+	timeout = d->rttavg;
+	timeout += timeout >> 1;
+
+	spin_lock_irqsave(&d->lock, flags);
+
+	if (d->flags & DEVFL_TKILL) {
+tdie:		spin_unlock_irqrestore(&d->lock, flags);
+		return;
+	}
+	f = d->frames;
+	e = f + d->nframes;
+	for (; f<e; f++) {
+		if (f->tag != FREETAG && tsince(f->tag) >= timeout) {
+			n = f->waited += timeout;
+			n /= HZ;
+			if (n > MAXWAIT) { /* waited too long.  device failure. */
+				aoedev_downdev(d);
+				goto tdie;
+			}
+			rexmit(d, f);
+		}
+	}
+
+	sl = d->skblist;
+	d->skblist = NULL;
+	if (sl) {
+		n = d->rttavg <<= 1;
+		if (n > MAXTIMER)
+			d->rttavg = MAXTIMER;
+	}
+
+	d->timer.expires = jiffies + TIMERTICK;
+	add_timer(&d->timer);
+
+	spin_unlock_irqrestore(&d->lock, flags);
+
+	aoenet_xmit(sl);
+}
+
+static void
+ataid_complete(struct aoedev *d, unsigned char *id)
+{
+	u64 ssize;
+	u16 n;
+
+	/* word 83: command set supported */
+	n = __le16_to_cpu(*((u16 *) &id[83<<1]));
+
+	/* word 86: command set/feature enabled */
+	n |= __le16_to_cpu(*((u16 *) &id[86<<1]));
+
+	if (n & (1<<10)) {	/* bit 10: LBA 48 */
+		d->flags |= DEVFL_EXT;
+
+		/* word 100: number lba48 sectors */
+		ssize = __le64_to_cpu(*((u64 *) &id[100<<1]));
+
+		/* set as in ide-disk.c:init_idedisk_capacity */
+		d->geo.cylinders = ssize;
+		d->geo.cylinders /= (255 * 63);
+		d->geo.heads = 255;
+		d->geo.sectors = 63;
+	} else {
+		d->flags &= ~DEVFL_EXT;
+
+		/* number lba28 sectors */
+		ssize = __le32_to_cpu(*((u32 *) &id[60<<1]));
+
+		/* NOTE: obsolete in ATA 6 */
+		d->geo.cylinders = __le16_to_cpu(*((u16 *) &id[54<<1]));
+		d->geo.heads = __le16_to_cpu(*((u16 *) &id[55<<1]));
+		d->geo.sectors = __le16_to_cpu(*((u16 *) &id[56<<1]));
+	}
+	d->ssize = ssize;
+	d->geo.start = 0;
+	if (d->gd != NULL) {
+		d->gd->capacity = ssize;
+		d->flags |= DEVFL_UP;
+		return;
+	}
+	if (d->flags & DEVFL_WORKON) {
+		printk(KERN_INFO "aoe: ataid_complete: can't schedule work, it's already on!  "
+			"(This really shouldn't happen).\n");
+		return;
+	}
+	INIT_WORK(&d->work, aoeblk_gdalloc, d);
+	schedule_work(&d->work);
+	d->flags |= DEVFL_WORKON;
+}
+
+static void
+calc_rttavg(struct aoedev *d, int rtt)
+{
+	register long n;
+
+	n = rtt;
+	if (n < MINTIMER)
+		n = MINTIMER;
+	else if (n > MAXTIMER)
+		n = MAXTIMER;
+
+	/* g == .25; cf. Congestion Avoidance and Control, Jacobson & Karels; 1988 */
+	n -= d->rttavg;
+	d->rttavg += n >> 2;
+}
+
+void
+aoecmd_ata_rsp(struct sk_buff *skb)
+{
+	struct aoedev *d;
+	struct aoe_hdr *hin;
+	struct aoe_atahdr *ahin, *ahout;
+	struct frame *f;
+	struct buf *buf;
+	struct sk_buff *sl;
+	register long n;
+	ulong flags;
+	char ebuf[128];
+	
+	hin = (struct aoe_hdr *) skb->mac.raw;
+	d = aoedev_bymac(hin->src);
+	if (d == NULL) {
+		snprintf(ebuf, sizeof ebuf, "aoecmd_ata_rsp: ata response "
+			"for unknown device %d.%d\n",
+			 __be16_to_cpu(*((u16 *) hin->major)),
+			hin->minor);
+		aoechr_error(ebuf);
+		return;
+	}
+
+	spin_lock_irqsave(&d->lock, flags);
+
+	f = getframe(d, __be32_to_cpu(*((u32 *) hin->tag)));
+	if (f == NULL) {
+		spin_unlock_irqrestore(&d->lock, flags);
+		snprintf(ebuf, sizeof ebuf,
+			"%15s e%d.%d    tag=%08x@%08lx\n",
+			"unexpected rsp",
+			__be16_to_cpu(*((u16 *) hin->major)),
+			hin->minor,
+			__be32_to_cpu(*((u32 *) hin->tag)),
+			jiffies);
+		aoechr_error(ebuf);
+		return;
+	}
+
+	calc_rttavg(d, tsince(f->tag));
+
+	ahin = (struct aoe_atahdr *) (hin+1);
+	ahout = (struct aoe_atahdr *) (f->data + sizeof(struct aoe_hdr));
+	buf = f->buf;
+
+	if (ahin->cmdstat & 0xa9) {	/* these bits cleared on success */
+		printk(KERN_CRIT "aoe: aoecmd_ata_rsp: ata error cmd=%2.2Xh "
+			"stat=%2.2Xh from e%ld.%ld\n", 
+			ahout->cmdstat, ahin->cmdstat,
+			d->aoemajor, d->aoeminor);
+		if (buf)
+			buf->flags |= BUFFL_FAIL;
+	} else {
+		switch (ahout->cmdstat) {
+		case WIN_READ:
+		case WIN_READ_EXT:
+			n = ahout->scnt << 9;
+			if (skb->len - sizeof *hin - sizeof *ahin < n) {
+				printk(KERN_CRIT "aoe: aoecmd_ata_rsp: runt "
+					"ata data size in read.  skb->len=%d\n",
+					skb->len);
+				/* fail frame f?  just returning will rexmit. */
+				spin_unlock_irqrestore(&d->lock, flags);
+				return;
+			}
+			memcpy(f->bufaddr, ahin+1, n);
+		case WIN_WRITE:
+		case WIN_WRITE_EXT:
+			break;
+		case WIN_IDENTIFY:
+			if (skb->len - sizeof *hin - sizeof *ahin < 512) {
+				printk(KERN_INFO "aoe: aoecmd_ata_rsp: runt data size "
+					"in ataid.  skb->len=%d\n", skb->len);
+				spin_unlock_irqrestore(&d->lock, flags);
+				return;
+			}
+			ataid_complete(d, (char *) (ahin+1));
+			/* d->flags |= DEVFL_WC_UPDATE; */
+			break;
+		default:
+			printk(KERN_INFO "aoe: aoecmd_ata_rsp: unrecognized "
+			       "outbound ata command %2.2Xh for %d.%d\n", 
+			       ahout->cmdstat,
+			       __be16_to_cpu(*((u16 *) hin->major)),
+			       hin->minor);
+		}
+	}
+
+	if (buf) {
+		buf->nframesout -= 1;
+		if (buf->nframesout == 0 && buf->resid == 0) {
+			n = (buf->flags & BUFFL_FAIL) ? -EIO : 0;
+			bio_endio(buf->bio, buf->bio->bi_size, n);
+			mempool_free(buf, d->bufpool);
+		}
+	}
+
+	f->buf = NULL;
+	f->tag = FREETAG;
+
+	aoecmd_work(d);
+
+	sl = d->skblist;
+	d->skblist = NULL;
+
+	spin_unlock_irqrestore(&d->lock, flags);
+
+	aoenet_xmit(sl);
+}
+
+void
+aoecmd_cfg(ushort aoemajor, unsigned char aoeminor)
+{
+	struct aoe_hdr *h;
+	struct aoe_cfghdr *ch;
+	struct sk_buff *skb, *sl;