[linux-2.6-block.git] / drivers / block / aoe / aoeblk.c

/* Copyright (c) 2013 Coraid, Inc.  See COPYING for GPL terms. */
/*
 * aoeblk.c
 * block device routines
 */

#include <linux/kernel.h>
#include <linux/hdreg.h>
#include <linux/blkdev.h>
#include <linux/backing-dev.h>
#include <linux/fs.h>
#include <linux/ioctl.h>
#include <linux/slab.h>
#include <linux/ratelimit.h>
#include <linux/genhd.h>
#include <linux/netdevice.h>
#include <linux/mutex.h>
#include <linux/export.h>
#include <linux/moduleparam.h>
#include <linux/debugfs.h>
#include <scsi/sg.h>
#include "aoe.h"

static DEFINE_MUTEX(aoeblk_mutex);
static struct kmem_cache *buf_pool_cache;
static struct dentry *aoe_debugfs_dir;

/* GPFS needs a larger value than the default. */
static int aoe_maxsectors;
module_param(aoe_maxsectors, int, 0644);
MODULE_PARM_DESC(aoe_maxsectors,
	"When nonzero, set the maximum number of sectors per I/O request");

static ssize_t aoedisk_show_state(struct device *dev,
				  struct device_attribute *attr, char *page)
{
	struct gendisk *disk = dev_to_disk(dev);
	struct aoedev *d = disk->private_data;

	return snprintf(page, PAGE_SIZE,
			"%s%s\n",
			(d->flags & DEVFL_UP) ? "up" : "down",
			(d->flags & DEVFL_KICKME) ? ",kickme" :
			(d->nopen && !(d->flags & DEVFL_UP)) ? ",closewait" : "");
	/* I'd rather see nopen exported so we can ditch closewait */
}
static ssize_t aoedisk_show_mac(struct device *dev,
				struct device_attribute *attr, char *page)
{
	struct gendisk *disk = dev_to_disk(dev);
	struct aoedev *d = disk->private_data;
	struct aoetgt *t = d->targets[0];

	if (t == NULL)
		return snprintf(page, PAGE_SIZE, "none\n");
	return snprintf(page, PAGE_SIZE, "%pm\n", t->addr);
}
static ssize_t aoedisk_show_netif(struct device *dev,
				  struct device_attribute *attr, char *page)
{
	struct gendisk *disk = dev_to_disk(dev);
	struct aoedev *d = disk->private_data;
	struct net_device *nds[8], **nd, **nnd, **ne;
	struct aoetgt **t, **te;
	struct aoeif *ifp, *e;
	char *p;

	memset(nds, 0, sizeof nds);
	nd = nds;
	ne = nd + ARRAY_SIZE(nds);
	t = d->targets;
	te = t + d->ntargets;
	for (; t < te && *t; t++) {
		ifp = (*t)->ifs;
		e = ifp + NAOEIFS;
		for (; ifp < e && ifp->nd; ifp++) {
			for (nnd = nds; nnd < nd; nnd++)
				if (*nnd == ifp->nd)
					break;
			if (nnd == nd && nd != ne)
				*nd++ = ifp->nd;
		}
	}

	ne = nd;
	nd = nds;
	if (*nd == NULL)
		return snprintf(page, PAGE_SIZE, "none\n");
	for (p = page; nd < ne; nd++)
		p += snprintf(p, PAGE_SIZE - (p-page), "%s%s",
			p == page ? "" : ",", (*nd)->name);
	p += snprintf(p, PAGE_SIZE - (p-page), "\n");
	return p-page;
}
/* firmware version */
static ssize_t aoedisk_show_fwver(struct device *dev,
				  struct device_attribute *attr, char *page)
{
	struct gendisk *disk = dev_to_disk(dev);
	struct aoedev *d = disk->private_data;

	return snprintf(page, PAGE_SIZE, "0x%04x\n", (unsigned int) d->fw_ver);
}
static ssize_t aoedisk_show_payload(struct device *dev,
				    struct device_attribute *attr, char *page)
{
	struct gendisk *disk = dev_to_disk(dev);
	struct aoedev *d = disk->private_data;

	return snprintf(page, PAGE_SIZE, "%lu\n", d->maxbcnt);
}

static int aoedisk_debugfs_show(struct seq_file *s, void *ignored)
{
	struct aoedev *d;
	struct aoetgt **t, **te;
	struct aoeif *ifp, *ife;
	unsigned long flags;
	char c;

	d = s->private;
	seq_printf(s, "rttavg: %d rttdev: %d\n",
		d->rttavg >> RTTSCALE,
		d->rttdev >> RTTDSCALE);
	seq_printf(s, "nskbpool: %d\n", skb_queue_len(&d->skbpool));
	seq_printf(s, "kicked: %ld\n", d->kicked);
	seq_printf(s, "maxbcnt: %ld\n", d->maxbcnt);
	seq_printf(s, "ref: %ld\n", d->ref);

	spin_lock_irqsave(&d->lock, flags);
	t = d->targets;
	te = t + d->ntargets;
	for (; t < te && *t; t++) {
		c = '\t';
		seq_printf(s, "falloc: %ld\n", (*t)->falloc);
		seq_printf(s, "ffree: %p\n",
			list_empty(&(*t)->ffree) ? NULL : (*t)->ffree.next);
		seq_printf(s, "%pm:%d:%d:%d\n", (*t)->addr, (*t)->nout,
			(*t)->maxout, (*t)->nframes);
		seq_printf(s, "\tssthresh:%d\n", (*t)->ssthresh);
		seq_printf(s, "\ttaint:%d\n", (*t)->taint);
		seq_printf(s, "\tr:%d\n", (*t)->rpkts);
		seq_printf(s, "\tw:%d\n", (*t)->wpkts);
		ifp = (*t)->ifs;
		ife = ifp + ARRAY_SIZE((*t)->ifs);
		for (; ifp->nd && ifp < ife; ifp++) {
			seq_printf(s, "%c%s", c, ifp->nd->name);
			c = ',';
		}
		seq_puts(s, "\n");
	}
	spin_unlock_irqrestore(&d->lock, flags);

	return 0;
}

static int aoe_debugfs_open(struct inode *inode, struct file *file)
{
	return single_open(file, aoedisk_debugfs_show, inode->i_private);
}

static DEVICE_ATTR(state, S_IRUGO, aoedisk_show_state, NULL);
static DEVICE_ATTR(mac, S_IRUGO, aoedisk_show_mac, NULL);
static DEVICE_ATTR(netif, S_IRUGO, aoedisk_show_netif, NULL);
static struct device_attribute dev_attr_firmware_version = {
	.attr = { .name = "firmware-version", .mode = S_IRUGO },
	.show = aoedisk_show_fwver,
};
static DEVICE_ATTR(payload, S_IRUGO, aoedisk_show_payload, NULL);

static struct attribute *aoe_attrs[] = {
	&dev_attr_state.attr,
	&dev_attr_mac.attr,
	&dev_attr_netif.attr,
	&dev_attr_firmware_version.attr,
	&dev_attr_payload.attr,
	NULL,
};

static const struct attribute_group attr_group = {
	.attrs = aoe_attrs,
};

static const struct file_operations aoe_debugfs_fops = {
	.open = aoe_debugfs_open,
	.read = seq_read,
	.llseek = seq_lseek,
	.release = single_release,
};

static void
aoedisk_add_debugfs(struct aoedev *d)
{
	struct dentry *entry;
	char *p;

	if (aoe_debugfs_dir == NULL)
		return;
	p = strchr(d->gd->disk_name, '/');
	if (p == NULL)
		p = d->gd->disk_name;
	else
		p++;
	BUG_ON(*p == '\0');
	entry = debugfs_create_file(p, 0444, aoe_debugfs_dir, d,
				    &aoe_debugfs_fops);
	if (IS_ERR_OR_NULL(entry)) {
		pr_info("aoe: cannot create debugfs file for %s\n",
			d->gd->disk_name);
		return;
	}
	BUG_ON(d->debugfs);
	d->debugfs = entry;
}
void
aoedisk_rm_debugfs(struct aoedev *d)
{
	debugfs_remove(d->debugfs);
	d->debugfs = NULL;
}

static int
aoedisk_add_sysfs(struct aoedev *d)
{
	return sysfs_create_group(&disk_to_dev(d->gd)->kobj, &attr_group);
}
void
aoedisk_rm_sysfs(struct aoedev *d)
{
	sysfs_remove_group(&disk_to_dev(d->gd)->kobj, &attr_group);
}

static int
aoeblk_open(struct block_device *bdev, fmode_t mode)
{
	struct aoedev *d = bdev->bd_disk->private_data;
	ulong flags;

	if (!virt_addr_valid(d)) {
		pr_crit("aoe: invalid device pointer in %s\n",
			__func__);
		WARN_ON(1);
		return -ENODEV;
	}
	if (!(d->flags & DEVFL_UP) || d->flags & DEVFL_TKILL)
		return -ENODEV;

	mutex_lock(&aoeblk_mutex);
	spin_lock_irqsave(&d->lock, flags);
	if (d->flags & DEVFL_UP && !(d->flags & DEVFL_TKILL)) {
		d->nopen++;
		spin_unlock_irqrestore(&d->lock, flags);
		mutex_unlock(&aoeblk_mutex);
		return 0;
	}
	spin_unlock_irqrestore(&d->lock, flags);
	mutex_unlock(&aoeblk_mutex);
	return -ENODEV;
}

static void
aoeblk_release(struct gendisk *disk, fmode_t mode)
{
	struct aoedev *d = disk->private_data;
	ulong flags;

	spin_lock_irqsave(&d->lock, flags);

	if (--d->nopen == 0) {
		spin_unlock_irqrestore(&d->lock, flags);
		aoecmd_cfg(d->aoemajor, d->aoeminor);
		return;
	}
	spin_unlock_irqrestore(&d->lock, flags);
}

static void
aoeblk_request(struct request_queue *q)
{
	struct aoedev *d;
	struct request *rq;

	d = q->queuedata;
	if ((d->flags & DEVFL_UP) == 0) {
		pr_info_ratelimited("aoe: device %ld.%d is not up\n",
			d->aoemajor, d->aoeminor);
		while ((rq = blk_peek_request(q))) {
			blk_start_request(rq);
			aoe_end_request(d, rq, 1);
		}
		return;
	}
	aoecmd_work(d);
}

static int
aoeblk_getgeo(struct block_device *bdev, struct hd_geometry *geo)
{
	struct aoedev *d = bdev->bd_disk->private_data;

	if ((d->flags & DEVFL_UP) == 0) {
		printk(KERN_ERR "aoe: disk not up\n");
		return -ENODEV;
	}

	geo->cylinders = d->geo.cylinders;
	geo->heads = d->geo.heads;
	geo->sectors = d->geo.sectors;
	return 0;
}

static int
aoeblk_ioctl(struct block_device *bdev, fmode_t mode, uint cmd, ulong arg)
{
	struct aoedev *d;

	if (!arg)
		return -EINVAL;

	d = bdev->bd_disk->private_data;
	if ((d->flags & DEVFL_UP) == 0) {
		pr_err("aoe: disk not up\n");
		return -ENODEV;
	}

	if (cmd == HDIO_GET_IDENTITY) {
		if (!copy_to_user((void __user *) arg, &d->ident,
			sizeof(d->ident)))
			return 0;
		return -EFAULT;
	}

	/* udev calls scsi_id, which uses SG_IO, resulting in noise */
	if (cmd != SG_IO)
		pr_info("aoe: unknown ioctl 0x%x\n", cmd);

	return -ENOTTY;
}

static const struct block_device_operations aoe_bdops = {
	.open = aoeblk_open,
	.release = aoeblk_release,
	.ioctl = aoeblk_ioctl,
	.getgeo = aoeblk_getgeo,
	.owner = THIS_MODULE,
};

/* alloc_disk and add_disk can sleep */
void
aoeblk_gdalloc(void *vp)
{
	struct aoedev *d = vp;
	struct gendisk *gd;
	mempool_t *mp;
	struct request_queue *q;
	enum { KB = 1024, MB = KB * KB, READ_AHEAD = 2 * MB, };
	ulong flags;
	int late = 0;

	spin_lock_irqsave(&d->lock, flags);
	if (d->flags & DEVFL_GDALLOC
	&& !(d->flags & DEVFL_TKILL)
	&& !(d->flags & DEVFL_GD_NOW))
		d->flags |= DEVFL_GD_NOW;
	else
		late = 1;
	spin_unlock_irqrestore(&d->lock, flags);
	if (late)
		return;

	gd = alloc_disk(AOE_PARTITIONS);
	if (gd == NULL) {
		pr_err("aoe: cannot allocate disk structure for %ld.%d\n",
			d->aoemajor, d->aoeminor);
		goto err;
	}

	mp = mempool_create(MIN_BUFS, mempool_alloc_slab, mempool_free_slab,
		buf_pool_cache);
	if (mp == NULL) {
		printk(KERN_ERR "aoe: cannot allocate bufpool for %ld.%d\n",
			d->aoemajor, d->aoeminor);
		goto err_disk;
	}
	q = blk_init_queue(aoeblk_request, &d->lock);
	if (q == NULL) {
		pr_err("aoe: cannot allocate block queue for %ld.%d\n",
			d->aoemajor, d->aoeminor);
		goto err_mempool;
	}

	spin_lock_irqsave(&d->lock, flags);
	WARN_ON(!(d->flags & DEVFL_GD_NOW));
	WARN_ON(!(d->flags & DEVFL_GDALLOC));
	WARN_ON(d->flags & DEVFL_TKILL);
	WARN_ON(d->gd);
	WARN_ON(d->flags & DEVFL_UP);
	blk_queue_max_hw_sectors(q, BLK_DEF_MAX_SECTORS);
	q->backing_dev_info->name = "aoe";
	q->backing_dev_info->ra_pages = READ_AHEAD / PAGE_SIZE;
	d->bufpool = mp;
	d->blkq = gd->queue = q;
	q->queuedata = d;
	d->gd = gd;
	if (aoe_maxsectors)
		blk_queue_max_hw_sectors(q, aoe_maxsectors);
	gd->major = AOE_MAJOR;
	gd->first_minor = d->sysminor;
	gd->fops = &aoe_bdops;
	gd->private_data = d;
	set_capacity(gd, d->ssize);
	snprintf(gd->disk_name, sizeof gd->disk_name, "etherd/e%ld.%d",
		d->aoemajor, d->aoeminor);

	d->flags &= ~DEVFL_GDALLOC;
	d->flags |= DEVFL_UP;

	spin_unlock_irqrestore(&d->lock, flags);

	add_disk(gd);
	aoedisk_add_sysfs(d);
	aoedisk_add_debugfs(d);

	spin_lock_irqsave(&d->lock, flags);
	WARN_ON(!(d->flags & DEVFL_GD_NOW));
	d->flags &= ~DEVFL_GD_NOW;
	spin_unlock_irqrestore(&d->lock, flags);
	return;

err_mempool:
	mempool_destroy(mp);
err_disk:
	put_disk(gd);
err:
	spin_lock_irqsave(&d->lock, flags);
	d->flags &= ~DEVFL_GD_NOW;
	schedule_work(&d->work);
	spin_unlock_irqrestore(&d->lock, flags);
}

void
aoeblk_exit(void)
{
	debugfs_remove_recursive(aoe_debugfs_dir);
	aoe_debugfs_dir = NULL;
	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);
	if (buf_pool_cache == NULL)
		return -ENOMEM;
	aoe_debugfs_dir = debugfs_create_dir("aoe", NULL);
	if (IS_ERR_OR_NULL(aoe_debugfs_dir)) {
		pr_info("aoe: cannot create debugfs directory\n");
		aoe_debugfs_dir = NULL;
	}
	return 0;
}
Commit	Line	Data
ec345120	1	/* Copyright (c) 2013 Coraid, Inc. See COPYING for GPL terms. */
1da177e4 LT	2	/*
	3	* aoeblk.c
	4	* block device routines
	5	*/
	6
027b180d	7	#include <linux/kernel.h>
1da177e4 LT	8	#include <linux/hdreg.h>
1da177e4 LT	9	#include <linux/blkdev.h>
43cbe2cb	10	#include <linux/backing-dev.h>
1da177e4 LT	11	#include <linux/fs.h>
1da177e4 LT	12	#include <linux/ioctl.h>
5a0e3ad6	13	#include <linux/slab.h>
027b180d	14	#include <linux/ratelimit.h>
1da177e4 LT	15	#include <linux/genhd.h>
1da177e4 LT	16	#include <linux/netdevice.h>
2a48fc0a	17	#include <linux/mutex.h>
d5decd3b	18	#include <linux/export.h>
aa304fde	19	#include <linux/moduleparam.h>
190519cd	20	#include <linux/debugfs.h>
667be1e7	21	#include <scsi/sg.h>
1da177e4 LT	22	#include "aoe.h"
1da177e4 LT	23
2a48fc0a	24	static DEFINE_MUTEX(aoeblk_mutex);
e18b890b	25	static struct kmem_cache *buf_pool_cache;
190519cd	26	static struct dentry *aoe_debugfs_dir;
1da177e4	27
aa304fde EC	28	/* GPFS needs a larger value than the default. */
	29	static int aoe_maxsectors;
	30	module_param(aoe_maxsectors, int, 0644);
	31	MODULE_PARM_DESC(aoe_maxsectors,
	32	"When nonzero, set the maximum number of sectors per I/O request");
	33
edfaa7c3 KS	34	static ssize_t aoedisk_show_state(struct device *dev,
edfaa7c3 KS	35	struct device_attribute attr, char page)
1da177e4	36	{
edfaa7c3	37	struct gendisk *disk = dev_to_disk(dev);
1da177e4 LT	38	struct aoedev *d = disk->private_data;
	39
	40	return snprintf(page, PAGE_SIZE,
	41	"%s%s\n",
	42	(d->flags & DEVFL_UP) ? "up" : "down",
68e0d42f	43	(d->flags & DEVFL_KICKME) ? ",kickme" :
3ae1c24e EC	44	(d->nopen && !(d->flags & DEVFL_UP)) ? ",closewait" : "");
3ae1c24e EC	45	/* I'd rather see nopen exported so we can ditch closewait */
1da177e4	46	}
edfaa7c3 KS	47	static ssize_t aoedisk_show_mac(struct device *dev,
edfaa7c3 KS	48	struct device_attribute attr, char page)
1da177e4	49	{
edfaa7c3	50	struct gendisk *disk = dev_to_disk(dev);
1da177e4	51	struct aoedev *d = disk->private_data;
68e0d42f	52	struct aoetgt *t = d->targets[0];
1da177e4	53
68e0d42f EC	54	if (t == NULL)
68e0d42f EC	55	return snprintf(page, PAGE_SIZE, "none\n");
411c41ee	56	return snprintf(page, PAGE_SIZE, "%pm\n", t->addr);
1da177e4	57	}
edfaa7c3 KS	58	static ssize_t aoedisk_show_netif(struct device *dev,
edfaa7c3 KS	59	struct device_attribute attr, char page)
1da177e4	60	{
edfaa7c3	61	struct gendisk *disk = dev_to_disk(dev);
1da177e4	62	struct aoedev *d = disk->private_data;
68e0d42f EC	63	struct net_device nds[8], nd, nnd, *ne;
	64	struct aoetgt t, te;
	65	struct aoeif ifp, e;
	66	char *p;
	67
	68	memset(nds, 0, sizeof nds);
	69	nd = nds;
	70	ne = nd + ARRAY_SIZE(nds);
	71	t = d->targets;
71114ec4	72	te = t + d->ntargets;
68e0d42f EC	73	for (; t < te && *t; t++) {
	74	ifp = (*t)->ifs;
	75	e = ifp + NAOEIFS;
	76	for (; ifp < e && ifp->nd; ifp++) {
	77	for (nnd = nds; nnd < nd; nnd++)
	78	if (*nnd == ifp->nd)
	79	break;
	80	if (nnd == nd && nd != ne)
	81	*nd++ = ifp->nd;
	82	}
	83	}
1da177e4	84
68e0d42f EC	85	ne = nd;
	86	nd = nds;
	87	if (*nd == NULL)
	88	return snprintf(page, PAGE_SIZE, "none\n");
	89	for (p = page; nd < ne; nd++)
	90	p += snprintf(p, PAGE_SIZE - (p-page), "%s%s",
	91	p == page ? "" : ",", (*nd)->name);
	92	p += snprintf(p, PAGE_SIZE - (p-page), "\n");
	93	return p-page;
1da177e4	94	}
4613ed27	95	/* firmware version */
edfaa7c3 KS	96	static ssize_t aoedisk_show_fwver(struct device *dev,
edfaa7c3 KS	97	struct device_attribute attr, char page)
4613ed27	98	{
edfaa7c3	99	struct gendisk *disk = dev_to_disk(dev);
4613ed27 EC	100	struct aoedev *d = disk->private_data;
	101
	102	return snprintf(page, PAGE_SIZE, "0x%04x\n", (unsigned int) d->fw_ver);
	103	}
90a2508d EC	104	static ssize_t aoedisk_show_payload(struct device *dev,
	105	struct device_attribute attr, char page)
	106	{
	107	struct gendisk *disk = dev_to_disk(dev);
	108	struct aoedev *d = disk->private_data;
	109
	110	return snprintf(page, PAGE_SIZE, "%lu\n", d->maxbcnt);
	111	}
1da177e4	112
1cf94797 EC	113	static int aoedisk_debugfs_show(struct seq_file s, void ignored)
	114	{
	115	struct aoedev *d;
2256c1c5 EC	116	struct aoetgt t, te;
2256c1c5 EC	117	struct aoeif ifp, ife;
1cf94797	118	unsigned long flags;
2256c1c5	119	char c;
1cf94797 EC	120
1cf94797 EC	121	d = s->private;
2256c1c5 EC	122	seq_printf(s, "rttavg: %d rttdev: %d\n",
	123	d->rttavg >> RTTSCALE,
	124	d->rttdev >> RTTDSCALE);
	125	seq_printf(s, "nskbpool: %d\n", skb_queue_len(&d->skbpool));
	126	seq_printf(s, "kicked: %ld\n", d->kicked);
	127	seq_printf(s, "maxbcnt: %ld\n", d->maxbcnt);
	128	seq_printf(s, "ref: %ld\n", d->ref);
	129
1cf94797	130	spin_lock_irqsave(&d->lock, flags);
2256c1c5 EC	131	t = d->targets;
	132	te = t + d->ntargets;
	133	for (; t < te && *t; t++) {
	134	c = '\t';
	135	seq_printf(s, "falloc: %ld\n", (*t)->falloc);
	136	seq_printf(s, "ffree: %p\n",
	137	list_empty(&(t)->ffree) ? NULL : (t)->ffree.next);
	138	seq_printf(s, "%pm:%d:%d:%d\n", (t)->addr, (t)->nout,
	139	(t)->maxout, (t)->nframes);
	140	seq_printf(s, "\tssthresh:%d\n", (*t)->ssthresh);
	141	seq_printf(s, "\ttaint:%d\n", (*t)->taint);
	142	seq_printf(s, "\tr:%d\n", (*t)->rpkts);
	143	seq_printf(s, "\tw:%d\n", (*t)->wpkts);
	144	ifp = (*t)->ifs;
	145	ife = ifp + ARRAY_SIZE((*t)->ifs);
	146	for (; ifp->nd && ifp < ife; ifp++) {
	147	seq_printf(s, "%c%s", c, ifp->nd->name);
	148	c = ',';
	149	}
	150	seq_puts(s, "\n");
	151	}
1cf94797 EC	152	spin_unlock_irqrestore(&d->lock, flags);
	153
	154	return 0;
	155	}
	156
	157	static int aoe_debugfs_open(struct inode inode, struct file file)
	158	{
	159	return single_open(file, aoedisk_debugfs_show, inode->i_private);
	160	}
	161
edfaa7c3 KS	162	static DEVICE_ATTR(state, S_IRUGO, aoedisk_show_state, NULL);
	163	static DEVICE_ATTR(mac, S_IRUGO, aoedisk_show_mac, NULL);
	164	static DEVICE_ATTR(netif, S_IRUGO, aoedisk_show_netif, NULL);
	165	static struct device_attribute dev_attr_firmware_version = {
01e8ef11	166	.attr = { .name = "firmware-version", .mode = S_IRUGO },
edfaa7c3	167	.show = aoedisk_show_fwver,
4613ed27	168	};
90a2508d	169	static DEVICE_ATTR(payload, S_IRUGO, aoedisk_show_payload, NULL);
1da177e4	170
4ca5224f	171	static struct attribute *aoe_attrs[] = {
edfaa7c3 KS	172	&dev_attr_state.attr,
	173	&dev_attr_mac.attr,
	174	&dev_attr_netif.attr,
	175	&dev_attr_firmware_version.attr,
90a2508d	176	&dev_attr_payload.attr,
edfaa7c3	177	NULL,
4ca5224f GKH	178	};
	179
	180	static const struct attribute_group attr_group = {
	181	.attrs = aoe_attrs,
	182	};
	183
1cf94797 EC	184	static const struct file_operations aoe_debugfs_fops = {
	185	.open = aoe_debugfs_open,
	186	.read = seq_read,
	187	.llseek = seq_lseek,
	188	.release = single_release,
	189	};
e8866cf2 EC	190
	191	static void
	192	aoedisk_add_debugfs(struct aoedev *d)
	193	{
	194	struct dentry *entry;
	195	char *p;
	196
	197	if (aoe_debugfs_dir == NULL)
	198	return;
	199	p = strchr(d->gd->disk_name, '/');
	200	if (p == NULL)
	201	p = d->gd->disk_name;
	202	else
	203	p++;
	204	BUG_ON(*p == '\0');
	205	entry = debugfs_create_file(p, 0444, aoe_debugfs_dir, d,
	206	&aoe_debugfs_fops);
	207	if (IS_ERR_OR_NULL(entry)) {
	208	pr_info("aoe: cannot create debugfs file for %s\n",
	209	d->gd->disk_name);
	210	return;
	211	}
	212	BUG_ON(d->debugfs);
	213	d->debugfs = entry;
	214	}
	215	void
	216	aoedisk_rm_debugfs(struct aoedev *d)
	217	{
e8866cf2 EC	218	debugfs_remove(d->debugfs);
	219	d->debugfs = NULL;
	220	}
	221
4ca5224f	222	static int
1da177e4 LT	223	aoedisk_add_sysfs(struct aoedev *d)
1da177e4 LT	224	{
ed9e1982	225	return sysfs_create_group(&disk_to_dev(d->gd)->kobj, &attr_group);
1da177e4 LT	226	}
	227	void
	228	aoedisk_rm_sysfs(struct aoedev *d)
	229	{
ed9e1982	230	sysfs_remove_group(&disk_to_dev(d->gd)->kobj, &attr_group);
1da177e4 LT	231	}
	232
	233	static int
94562c17	234	aoeblk_open(struct block_device *bdev, fmode_t mode)
1da177e4	235	{
94562c17	236	struct aoedev *d = bdev->bd_disk->private_data;
1da177e4 LT	237	ulong flags;
1da177e4 LT	238
e52a2932 EC	239	if (!virt_addr_valid(d)) {
	240	pr_crit("aoe: invalid device pointer in %s\n",
	241	__func__);
	242	WARN_ON(1);
	243	return -ENODEV;
	244	}
	245	if (!(d->flags & DEVFL_UP) \|\| d->flags & DEVFL_TKILL)
	246	return -ENODEV;
	247
2a48fc0a	248	mutex_lock(&aoeblk_mutex);
1da177e4	249	spin_lock_irqsave(&d->lock, flags);
e52a2932	250	if (d->flags & DEVFL_UP && !(d->flags & DEVFL_TKILL)) {
1da177e4 LT	251	d->nopen++;
1da177e4 LT	252	spin_unlock_irqrestore(&d->lock, flags);
2a48fc0a	253	mutex_unlock(&aoeblk_mutex);
1da177e4 LT	254	return 0;
	255	}
	256	spin_unlock_irqrestore(&d->lock, flags);
2a48fc0a	257	mutex_unlock(&aoeblk_mutex);
1da177e4 LT	258	return -ENODEV;
	259	}
	260
db2a144b	261	static void
94562c17	262	aoeblk_release(struct gendisk *disk, fmode_t mode)
1da177e4	263	{
94562c17	264	struct aoedev *d = disk->private_data;
1da177e4 LT	265	ulong flags;
1da177e4 LT	266
1da177e4 LT	267	spin_lock_irqsave(&d->lock, flags);
1da177e4 LT	268
5f7702fd	269	if (--d->nopen == 0) {
1da177e4 LT	270	spin_unlock_irqrestore(&d->lock, flags);
1da177e4 LT	271	aoecmd_cfg(d->aoemajor, d->aoeminor);
db2a144b	272	return;
1da177e4 LT	273	}
1da177e4 LT	274	spin_unlock_irqrestore(&d->lock, flags);
1da177e4 LT	275	}
1da177e4 LT	276
5a7bbad2	277	static void
69cf2d85	278	aoeblk_request(struct request_queue *q)
1da177e4 LT	279	{
1da177e4 LT	280	struct aoedev *d;
69cf2d85	281	struct request *rq;
1da177e4	282
69cf2d85	283	d = q->queuedata;
1da177e4	284	if ((d->flags & DEVFL_UP) == 0) {
027b180d	285	pr_info_ratelimited("aoe: device %ld.%d is not up\n",
a12c93f0	286	d->aoemajor, d->aoeminor);
69cf2d85 EC	287	while ((rq = blk_peek_request(q))) {
	288	blk_start_request(rq);
	289	aoe_end_request(d, rq, 1);
	290	}
5a7bbad2	291	return;
1da177e4	292	}
3ae1c24e	293	aoecmd_work(d);
1da177e4 LT	294	}
1da177e4 LT	295
1da177e4	296	static int
a885c8c4	297	aoeblk_getgeo(struct block_device bdev, struct hd_geometry geo)
1da177e4	298	{
a885c8c4	299	struct aoedev *d = bdev->bd_disk->private_data;
1da177e4	300
1da177e4	301	if ((d->flags & DEVFL_UP) == 0) {
a12c93f0	302	printk(KERN_ERR "aoe: disk not up\n");
1da177e4 LT	303	return -ENODEV;
	304	}
	305
a885c8c4 CH	306	geo->cylinders = d->geo.cylinders;
	307	geo->heads = d->geo.heads;
	308	geo->sectors = d->geo.sectors;
	309	return 0;
1da177e4 LT	310	}
1da177e4 LT	311
667be1e7 EC	312	static int
	313	aoeblk_ioctl(struct block_device *bdev, fmode_t mode, uint cmd, ulong arg)
	314	{
	315	struct aoedev *d;
	316
	317	if (!arg)
	318	return -EINVAL;
	319
	320	d = bdev->bd_disk->private_data;
	321	if ((d->flags & DEVFL_UP) == 0) {
	322	pr_err("aoe: disk not up\n");
	323	return -ENODEV;
	324	}
	325
	326	if (cmd == HDIO_GET_IDENTITY) {
	327	if (!copy_to_user((void __user *) arg, &d->ident,
	328	sizeof(d->ident)))
	329	return 0;
	330	return -EFAULT;
	331	}
	332
	333	/* udev calls scsi_id, which uses SG_IO, resulting in noise */
	334	if (cmd != SG_IO)
	335	pr_info("aoe: unknown ioctl 0x%x\n", cmd);
	336
	337	return -ENOTTY;
	338	}
	339
83d5cde4	340	static const struct block_device_operations aoe_bdops = {
94562c17 AV	341	.open = aoeblk_open,
94562c17 AV	342	.release = aoeblk_release,
667be1e7	343	.ioctl = aoeblk_ioctl,
a885c8c4	344	.getgeo = aoeblk_getgeo,
1da177e4 LT	345	.owner = THIS_MODULE,
	346	};
	347
	348	/* alloc_disk and add_disk can sleep */
	349	void
	350	aoeblk_gdalloc(void *vp)
	351	{
	352	struct aoedev *d = vp;
	353	struct gendisk *gd;
69cf2d85 EC	354	mempool_t *mp;
	355	struct request_queue *q;
	356	enum { KB = 1024, MB = KB * KB, READ_AHEAD = 2 * MB, };
1da177e4	357	ulong flags;
e52a2932 EC	358	int late = 0;
	359
	360	spin_lock_irqsave(&d->lock, flags);
	361	if (d->flags & DEVFL_GDALLOC
	362	&& !(d->flags & DEVFL_TKILL)
	363	&& !(d->flags & DEVFL_GD_NOW))
	364	d->flags \|= DEVFL_GD_NOW;
	365	else
	366	late = 1;
	367	spin_unlock_irqrestore(&d->lock, flags);
	368	if (late)
	369	return;
1da177e4 LT	370
	371	gd = alloc_disk(AOE_PARTITIONS);
	372	if (gd == NULL) {
69cf2d85	373	pr_err("aoe: cannot allocate disk structure for %ld.%d\n",
6bb6285f	374	d->aoemajor, d->aoeminor);
43cbe2cb	375	goto err;
1da177e4 LT	376	}
1da177e4 LT	377
69cf2d85 EC	378	mp = mempool_create(MIN_BUFS, mempool_alloc_slab, mempool_free_slab,
	379	buf_pool_cache);
	380	if (mp == NULL) {
1d75981a	381	printk(KERN_ERR "aoe: cannot allocate bufpool for %ld.%d\n",
6bb6285f	382	d->aoemajor, d->aoeminor);
43cbe2cb	383	goto err_disk;
1da177e4	384	}
69cf2d85 EC	385	q = blk_init_queue(aoeblk_request, &d->lock);
	386	if (q == NULL) {
	387	pr_err("aoe: cannot allocate block queue for %ld.%d\n",
	388	d->aoemajor, d->aoeminor);
0a41409c	389	goto err_mempool;
69cf2d85	390	}
1da177e4	391
43cbe2cb	392	spin_lock_irqsave(&d->lock, flags);
e52a2932 EC	393	WARN_ON(!(d->flags & DEVFL_GD_NOW));
	394	WARN_ON(!(d->flags & DEVFL_GDALLOC));
	395	WARN_ON(d->flags & DEVFL_TKILL);
	396	WARN_ON(d->gd);
	397	WARN_ON(d->flags & DEVFL_UP);
30e2bc08	398	blk_queue_max_hw_sectors(q, BLK_DEF_MAX_SECTORS);
dc3b17cc JK	399	q->backing_dev_info->name = "aoe";
dc3b17cc JK	400	q->backing_dev_info->ra_pages = READ_AHEAD / PAGE_SIZE;
69cf2d85 EC	401	d->bufpool = mp;
	402	d->blkq = gd->queue = q;
	403	q->queuedata = d;
	404	d->gd = gd;
aa304fde EC	405	if (aoe_maxsectors)
aa304fde EC	406	blk_queue_max_hw_sectors(q, aoe_maxsectors);
1da177e4	407	gd->major = AOE_MAJOR;
0c966214	408	gd->first_minor = d->sysminor;
1da177e4 LT	409	gd->fops = &aoe_bdops;
1da177e4 LT	410	gd->private_data = d;
80795aef	411	set_capacity(gd, d->ssize);
68e0d42f	412	snprintf(gd->disk_name, sizeof gd->disk_name, "etherd/e%ld.%d",
1da177e4 LT	413	d->aoemajor, d->aoeminor);
1da177e4 LT	414
3ae1c24e	415	d->flags &= ~DEVFL_GDALLOC;
1da177e4 LT	416	d->flags \|= DEVFL_UP;
	417
	418	spin_unlock_irqrestore(&d->lock, flags);
	419
	420	add_disk(gd);
	421	aoedisk_add_sysfs(d);
e8866cf2	422	aoedisk_add_debugfs(d);
e52a2932 EC	423
	424	spin_lock_irqsave(&d->lock, flags);
	425	WARN_ON(!(d->flags & DEVFL_GD_NOW));
	426	d->flags &= ~DEVFL_GD_NOW;
	427	spin_unlock_irqrestore(&d->lock, flags);
43cbe2cb AM	428	return;
	429
	430	err_mempool:
0a41409c	431	mempool_destroy(mp);
43cbe2cb AM	432	err_disk:
	433	put_disk(gd);
	434	err:
	435	spin_lock_irqsave(&d->lock, flags);
e52a2932 EC	436	d->flags &= ~DEVFL_GD_NOW;
e52a2932 EC	437	schedule_work(&d->work);
43cbe2cb	438	spin_unlock_irqrestore(&d->lock, flags);
1da177e4 LT	439	}
	440
	441	void
	442	aoeblk_exit(void)
	443	{
190519cd EC	444	debugfs_remove_recursive(aoe_debugfs_dir);
190519cd EC	445	aoe_debugfs_dir = NULL;
1da177e4 LT	446	kmem_cache_destroy(buf_pool_cache);
	447	}
	448
	449	int __init
	450	aoeblk_init(void)
	451	{
20c2df83	452	buf_pool_cache = kmem_cache_create("aoe_bufs",
1da177e4	453	sizeof(struct buf),
20c2df83	454	0, 0, NULL);
1da177e4 LT	455	if (buf_pool_cache == NULL)
1da177e4 LT	456	return -ENOMEM;
190519cd EC	457	aoe_debugfs_dir = debugfs_create_dir("aoe", NULL);
	458	if (IS_ERR_OR_NULL(aoe_debugfs_dir)) {
	459	pr_info("aoe: cannot create debugfs directory\n");
	460	aoe_debugfs_dir = NULL;
	461	}
1da177e4 LT	462	return 0;
	463	}
	464