[linux-block.git] / drivers / dax / super.c

/*
 * Copyright(c) 2017 Intel Corporation. All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * 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.
 */
#include <linux/pagemap.h>
#include <linux/module.h>
#include <linux/mount.h>
#include <linux/magic.h>
#include <linux/cdev.h>
#include <linux/hash.h>
#include <linux/slab.h>
#include <linux/fs.h>

static int nr_dax = CONFIG_NR_DEV_DAX;
module_param(nr_dax, int, S_IRUGO);
MODULE_PARM_DESC(nr_dax, "max number of dax device instances");

static dev_t dax_devt;
DEFINE_STATIC_SRCU(dax_srcu);
static struct vfsmount *dax_mnt;
static DEFINE_IDA(dax_minor_ida);
static struct kmem_cache *dax_cache __read_mostly;
static struct super_block *dax_superblock __read_mostly;

#define DAX_HASH_SIZE (PAGE_SIZE / sizeof(struct hlist_head))
static struct hlist_head dax_host_list[DAX_HASH_SIZE];
static DEFINE_SPINLOCK(dax_host_lock);

int dax_read_lock(void)
{
	return srcu_read_lock(&dax_srcu);
}
EXPORT_SYMBOL_GPL(dax_read_lock);

void dax_read_unlock(int id)
{
	srcu_read_unlock(&dax_srcu, id);
}
EXPORT_SYMBOL_GPL(dax_read_unlock);

/**
 * struct dax_device - anchor object for dax services
 * @inode: core vfs
 * @cdev: optional character interface for "device dax"
 * @host: optional name for lookups where the device path is not available
 * @private: dax driver private data
 * @alive: !alive + rcu grace period == no new operations / mappings
 */
struct dax_device {
	struct hlist_node list;
	struct inode inode;
	struct cdev cdev;
	const char *host;
	void *private;
	bool alive;
};

bool dax_alive(struct dax_device *dax_dev)
{
	lockdep_assert_held(&dax_srcu);
	return dax_dev->alive;
}
EXPORT_SYMBOL_GPL(dax_alive);

static int dax_host_hash(const char *host)
{
	return hashlen_hash(hashlen_string("DAX", host)) % DAX_HASH_SIZE;
}

/*
 * Note, rcu is not protecting the liveness of dax_dev, rcu is ensuring
 * that any fault handlers or operations that might have seen
 * dax_alive(), have completed.  Any operations that start after
 * synchronize_srcu() has run will abort upon seeing !dax_alive().
 */
void kill_dax(struct dax_device *dax_dev)
{
	if (!dax_dev)
		return;

	dax_dev->alive = false;

	synchronize_srcu(&dax_srcu);

	spin_lock(&dax_host_lock);
	hlist_del_init(&dax_dev->list);
	spin_unlock(&dax_host_lock);

	dax_dev->private = NULL;
}
EXPORT_SYMBOL_GPL(kill_dax);

static struct inode *dax_alloc_inode(struct super_block *sb)
{
	struct dax_device *dax_dev;

	dax_dev = kmem_cache_alloc(dax_cache, GFP_KERNEL);
	return &dax_dev->inode;
}

static struct dax_device *to_dax_dev(struct inode *inode)
{
	return container_of(inode, struct dax_device, inode);
}

static void dax_i_callback(struct rcu_head *head)
{
	struct inode *inode = container_of(head, struct inode, i_rcu);
	struct dax_device *dax_dev = to_dax_dev(inode);

	kfree(dax_dev->host);
	dax_dev->host = NULL;
	ida_simple_remove(&dax_minor_ida, MINOR(inode->i_rdev));
	kmem_cache_free(dax_cache, dax_dev);
}

static void dax_destroy_inode(struct inode *inode)
{
	struct dax_device *dax_dev = to_dax_dev(inode);

	WARN_ONCE(dax_dev->alive,
			"kill_dax() must be called before final iput()\n");
	call_rcu(&inode->i_rcu, dax_i_callback);
}

static const struct super_operations dax_sops = {
	.statfs = simple_statfs,
	.alloc_inode = dax_alloc_inode,
	.destroy_inode = dax_destroy_inode,
	.drop_inode = generic_delete_inode,
};

static struct dentry *dax_mount(struct file_system_type *fs_type,
		int flags, const char *dev_name, void *data)
{
	return mount_pseudo(fs_type, "dax:", &dax_sops, NULL, DAXFS_MAGIC);
}

static struct file_system_type dax_fs_type = {
	.name = "dax",
	.mount = dax_mount,
	.kill_sb = kill_anon_super,
};

static int dax_test(struct inode *inode, void *data)
{
	dev_t devt = *(dev_t *) data;

	return inode->i_rdev == devt;
}

static int dax_set(struct inode *inode, void *data)
{
	dev_t devt = *(dev_t *) data;

	inode->i_rdev = devt;
	return 0;
}

static struct dax_device *dax_dev_get(dev_t devt)
{
	struct dax_device *dax_dev;
	struct inode *inode;

	inode = iget5_locked(dax_superblock, hash_32(devt + DAXFS_MAGIC, 31),
			dax_test, dax_set, &devt);

	if (!inode)
		return NULL;

	dax_dev = to_dax_dev(inode);
	if (inode->i_state & I_NEW) {
		dax_dev->alive = true;
		inode->i_cdev = &dax_dev->cdev;
		inode->i_mode = S_IFCHR;
		inode->i_flags = S_DAX;
		mapping_set_gfp_mask(&inode->i_data, GFP_USER);
		unlock_new_inode(inode);
	}

	return dax_dev;
}

static void dax_add_host(struct dax_device *dax_dev, const char *host)
{
	int hash;

	/*
	 * Unconditionally init dax_dev since it's coming from a
	 * non-zeroed slab cache
	 */
	INIT_HLIST_NODE(&dax_dev->list);
	dax_dev->host = host;
	if (!host)
		return;

	hash = dax_host_hash(host);
	spin_lock(&dax_host_lock);
	hlist_add_head(&dax_dev->list, &dax_host_list[hash]);
	spin_unlock(&dax_host_lock);
}

struct dax_device *alloc_dax(void *private, const char *__host)
{
	struct dax_device *dax_dev;
	const char *host;
	dev_t devt;
	int minor;

	host = kstrdup(__host, GFP_KERNEL);
	if (__host && !host)
		return NULL;

	minor = ida_simple_get(&dax_minor_ida, 0, nr_dax, GFP_KERNEL);
	if (minor < 0)
		goto err_minor;

	devt = MKDEV(MAJOR(dax_devt), minor);
	dax_dev = dax_dev_get(devt);
	if (!dax_dev)
		goto err_dev;

	dax_add_host(dax_dev, host);
	dax_dev->private = private;
	return dax_dev;

 err_dev:
	ida_simple_remove(&dax_minor_ida, minor);
 err_minor:
	kfree(host);
	return NULL;
}
EXPORT_SYMBOL_GPL(alloc_dax);

void put_dax(struct dax_device *dax_dev)
{
	if (!dax_dev)
		return;
	iput(&dax_dev->inode);
}
EXPORT_SYMBOL_GPL(put_dax);

/**
 * dax_get_by_host() - temporary lookup mechanism for filesystem-dax
 * @host: alternate name for the device registered by a dax driver
 */
struct dax_device *dax_get_by_host(const char *host)
{
	struct dax_device *dax_dev, *found = NULL;
	int hash, id;

	if (!host)
		return NULL;

	hash = dax_host_hash(host);

	id = dax_read_lock();
	spin_lock(&dax_host_lock);
	hlist_for_each_entry(dax_dev, &dax_host_list[hash], list) {
		if (!dax_alive(dax_dev)
				|| strcmp(host, dax_dev->host) != 0)
			continue;

		if (igrab(&dax_dev->inode))
			found = dax_dev;
		break;
	}
	spin_unlock(&dax_host_lock);
	dax_read_unlock(id);

	return found;
}
EXPORT_SYMBOL_GPL(dax_get_by_host);

/**
 * inode_dax: convert a public inode into its dax_dev
 * @inode: An inode with i_cdev pointing to a dax_dev
 *
 * Note this is not equivalent to to_dax_dev() which is for private
 * internal use where we know the inode filesystem type == dax_fs_type.
 */
struct dax_device *inode_dax(struct inode *inode)
{
	struct cdev *cdev = inode->i_cdev;

	return container_of(cdev, struct dax_device, cdev);
}
EXPORT_SYMBOL_GPL(inode_dax);

struct inode *dax_inode(struct dax_device *dax_dev)
{
	return &dax_dev->inode;
}
EXPORT_SYMBOL_GPL(dax_inode);

void *dax_get_private(struct dax_device *dax_dev)
{
	return dax_dev->private;
}
EXPORT_SYMBOL_GPL(dax_get_private);

static void init_once(void *_dax_dev)
{
	struct dax_device *dax_dev = _dax_dev;
	struct inode *inode = &dax_dev->inode;

	inode_init_once(inode);
}

static int __dax_fs_init(void)
{
	int rc;

	dax_cache = kmem_cache_create("dax_cache", sizeof(struct dax_device), 0,
			(SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
			 SLAB_MEM_SPREAD|SLAB_ACCOUNT),
			init_once);
	if (!dax_cache)
		return -ENOMEM;

	rc = register_filesystem(&dax_fs_type);
	if (rc)
		goto err_register_fs;

	dax_mnt = kern_mount(&dax_fs_type);
	if (IS_ERR(dax_mnt)) {
		rc = PTR_ERR(dax_mnt);
		goto err_mount;
	}
	dax_superblock = dax_mnt->mnt_sb;

	return 0;

 err_mount:
	unregister_filesystem(&dax_fs_type);
 err_register_fs:
	kmem_cache_destroy(dax_cache);

	return rc;
}

static void __dax_fs_exit(void)
{
	kern_unmount(dax_mnt);
	unregister_filesystem(&dax_fs_type);
	kmem_cache_destroy(dax_cache);
}

static int __init dax_fs_init(void)
{
	int rc;

	rc = __dax_fs_init();
	if (rc)
		return rc;

	nr_dax = max(nr_dax, 256);
	rc = alloc_chrdev_region(&dax_devt, 0, nr_dax, "dax");
	if (rc)
		__dax_fs_exit();
	return rc;
}

static void __exit dax_fs_exit(void)
{
	unregister_chrdev_region(dax_devt, nr_dax);
	ida_destroy(&dax_minor_ida);
	__dax_fs_exit();
}

MODULE_AUTHOR("Intel Corporation");
MODULE_LICENSE("GPL v2");
subsys_initcall(dax_fs_init);
module_exit(dax_fs_exit);
Commit	Line	Data
7b6be844 DW	1	/*
	2	* Copyright(c) 2017 Intel Corporation. All rights reserved.
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of version 2 of the GNU General Public License as
	6	* published by the Free Software Foundation.
	7	*
	8	* This program is distributed in the hope that it will be useful, but
	9	* WITHOUT ANY WARRANTY; without even the implied warranty of
	10	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	11	* General Public License for more details.
	12	*/
	13	#include <linux/pagemap.h>
	14	#include <linux/module.h>
	15	#include <linux/mount.h>
	16	#include <linux/magic.h>
	17	#include <linux/cdev.h>
	18	#include <linux/hash.h>
	19	#include <linux/slab.h>
	20	#include <linux/fs.h>
	21
	22	static int nr_dax = CONFIG_NR_DEV_DAX;
	23	module_param(nr_dax, int, S_IRUGO);
	24	MODULE_PARM_DESC(nr_dax, "max number of dax device instances");
	25
	26	static dev_t dax_devt;
	27	DEFINE_STATIC_SRCU(dax_srcu);
	28	static struct vfsmount *dax_mnt;
	29	static DEFINE_IDA(dax_minor_ida);
	30	static struct kmem_cache *dax_cache __read_mostly;
	31	static struct super_block *dax_superblock __read_mostly;
	32
72058005 DW	33	#define DAX_HASH_SIZE (PAGE_SIZE / sizeof(struct hlist_head))
	34	static struct hlist_head dax_host_list[DAX_HASH_SIZE];
	35	static DEFINE_SPINLOCK(dax_host_lock);
	36
7b6be844 DW	37	int dax_read_lock(void)
	38	{
	39	return srcu_read_lock(&dax_srcu);
	40	}
	41	EXPORT_SYMBOL_GPL(dax_read_lock);
	42
	43	void dax_read_unlock(int id)
	44	{
	45	srcu_read_unlock(&dax_srcu, id);
	46	}
	47	EXPORT_SYMBOL_GPL(dax_read_unlock);
	48
	49	/**
	50	* struct dax_device - anchor object for dax services
	51	* @inode: core vfs
	52	* @cdev: optional character interface for "device dax"
72058005	53	* @host: optional name for lookups where the device path is not available
7b6be844 DW	54	* @private: dax driver private data
	55	* @alive: !alive + rcu grace period == no new operations / mappings
	56	*/
	57	struct dax_device {
72058005	58	struct hlist_node list;
7b6be844 DW	59	struct inode inode;
7b6be844 DW	60	struct cdev cdev;
72058005	61	const char *host;
7b6be844 DW	62	void *private;
	63	bool alive;
	64	};
	65
	66	bool dax_alive(struct dax_device *dax_dev)
	67	{
	68	lockdep_assert_held(&dax_srcu);
	69	return dax_dev->alive;
	70	}
	71	EXPORT_SYMBOL_GPL(dax_alive);
	72
72058005 DW	73	static int dax_host_hash(const char *host)
	74	{
	75	return hashlen_hash(hashlen_string("DAX", host)) % DAX_HASH_SIZE;
	76	}
	77
7b6be844 DW	78	/*
	79	* Note, rcu is not protecting the liveness of dax_dev, rcu is ensuring
	80	* that any fault handlers or operations that might have seen
	81	* dax_alive(), have completed. Any operations that start after
	82	* synchronize_srcu() has run will abort upon seeing !dax_alive().
	83	*/
	84	void kill_dax(struct dax_device *dax_dev)
	85	{
	86	if (!dax_dev)
	87	return;
	88
	89	dax_dev->alive = false;
72058005	90
7b6be844	91	synchronize_srcu(&dax_srcu);
72058005 DW	92
	93	spin_lock(&dax_host_lock);
	94	hlist_del_init(&dax_dev->list);
	95	spin_unlock(&dax_host_lock);
	96
7b6be844 DW	97	dax_dev->private = NULL;
	98	}
	99	EXPORT_SYMBOL_GPL(kill_dax);
	100
	101	static struct inode dax_alloc_inode(struct super_block sb)
	102	{
	103	struct dax_device *dax_dev;
	104
	105	dax_dev = kmem_cache_alloc(dax_cache, GFP_KERNEL);
	106	return &dax_dev->inode;
	107	}
	108
	109	static struct dax_device to_dax_dev(struct inode inode)
	110	{
	111	return container_of(inode, struct dax_device, inode);
	112	}
	113
	114	static void dax_i_callback(struct rcu_head *head)
	115	{
	116	struct inode *inode = container_of(head, struct inode, i_rcu);
	117	struct dax_device *dax_dev = to_dax_dev(inode);
	118
72058005 DW	119	kfree(dax_dev->host);
72058005 DW	120	dax_dev->host = NULL;
7b6be844 DW	121	ida_simple_remove(&dax_minor_ida, MINOR(inode->i_rdev));
	122	kmem_cache_free(dax_cache, dax_dev);
	123	}
	124
	125	static void dax_destroy_inode(struct inode *inode)
	126	{
	127	struct dax_device *dax_dev = to_dax_dev(inode);
	128
	129	WARN_ONCE(dax_dev->alive,
	130	"kill_dax() must be called before final iput()\n");
	131	call_rcu(&inode->i_rcu, dax_i_callback);
	132	}
	133
	134	static const struct super_operations dax_sops = {
	135	.statfs = simple_statfs,
	136	.alloc_inode = dax_alloc_inode,
	137	.destroy_inode = dax_destroy_inode,
	138	.drop_inode = generic_delete_inode,
	139	};
	140
	141	static struct dentry dax_mount(struct file_system_type fs_type,
	142	int flags, const char dev_name, void data)
	143	{
	144	return mount_pseudo(fs_type, "dax:", &dax_sops, NULL, DAXFS_MAGIC);
	145	}
	146
	147	static struct file_system_type dax_fs_type = {
	148	.name = "dax",
	149	.mount = dax_mount,
	150	.kill_sb = kill_anon_super,
	151	};
	152
	153	static int dax_test(struct inode inode, void data)
	154	{
	155	dev_t devt = (dev_t ) data;
	156
	157	return inode->i_rdev == devt;
	158	}
	159
	160	static int dax_set(struct inode inode, void data)
	161	{
	162	dev_t devt = (dev_t ) data;
	163
	164	inode->i_rdev = devt;
	165	return 0;
	166	}
	167
	168	static struct dax_device *dax_dev_get(dev_t devt)
	169	{
	170	struct dax_device *dax_dev;
	171	struct inode *inode;
	172
	173	inode = iget5_locked(dax_superblock, hash_32(devt + DAXFS_MAGIC, 31),
	174	dax_test, dax_set, &devt);
	175
	176	if (!inode)
	177	return NULL;
	178
	179	dax_dev = to_dax_dev(inode);
	180	if (inode->i_state & I_NEW) {
	181	dax_dev->alive = true;
	182	inode->i_cdev = &dax_dev->cdev;
	183	inode->i_mode = S_IFCHR;
	184	inode->i_flags = S_DAX;
185	mapping_set_gfp_mask(&inode->i_data, GFP_USER);
186	unlock_new_inode(inode);
187	}
188
189	return dax_dev;
190	}
191
72058005 DW	192	static void dax_add_host(struct dax_device dax_dev, const char host)
	193	{
	194	int hash;
	195
	196	/*
	197	* Unconditionally init dax_dev since it's coming from a
	198	* non-zeroed slab cache
	199	*/
	200	INIT_HLIST_NODE(&dax_dev->list);
	201	dax_dev->host = host;
	202	if (!host)
	203	return;
	204
	205	hash = dax_host_hash(host);
	206	spin_lock(&dax_host_lock);
	207	hlist_add_head(&dax_dev->list, &dax_host_list[hash]);
	208	spin_unlock(&dax_host_lock);
	209	}
	210
	211	struct dax_device alloc_dax(void private, const char *__host)
7b6be844 DW	212	{
7b6be844 DW	213	struct dax_device *dax_dev;
72058005	214	const char *host;
7b6be844 DW	215	dev_t devt;
	216	int minor;
	217
72058005 DW	218	host = kstrdup(__host, GFP_KERNEL);
	219	if (__host && !host)
	220	return NULL;
	221
7b6be844 DW	222	minor = ida_simple_get(&dax_minor_ida, 0, nr_dax, GFP_KERNEL);
7b6be844 DW	223	if (minor < 0)
72058005	224	goto err_minor;
7b6be844 DW	225
	226	devt = MKDEV(MAJOR(dax_devt), minor);
	227	dax_dev = dax_dev_get(devt);
	228	if (!dax_dev)
72058005	229	goto err_dev;
7b6be844	230
72058005	231	dax_add_host(dax_dev, host);
7b6be844 DW	232	dax_dev->private = private;
	233	return dax_dev;
	234
72058005	235	err_dev:
7b6be844	236	ida_simple_remove(&dax_minor_ida, minor);
72058005 DW	237	err_minor:
72058005 DW	238	kfree(host);
7b6be844 DW	239	return NULL;
	240	}
	241	EXPORT_SYMBOL_GPL(alloc_dax);
	242
	243	void put_dax(struct dax_device *dax_dev)
	244	{
	245	if (!dax_dev)
	246	return;
	247	iput(&dax_dev->inode);
	248	}
	249	EXPORT_SYMBOL_GPL(put_dax);
	250
72058005 DW	251	/**
	252	* dax_get_by_host() - temporary lookup mechanism for filesystem-dax
	253	* @host: alternate name for the device registered by a dax driver
	254	*/
	255	struct dax_device dax_get_by_host(const char host)
	256	{
	257	struct dax_device dax_dev, found = NULL;
	258	int hash, id;
	259
	260	if (!host)
	261	return NULL;
	262
	263	hash = dax_host_hash(host);
	264
	265	id = dax_read_lock();
	266	spin_lock(&dax_host_lock);
	267	hlist_for_each_entry(dax_dev, &dax_host_list[hash], list) {
	268	if (!dax_alive(dax_dev)
	269	\|\| strcmp(host, dax_dev->host) != 0)
	270	continue;
	271
	272	if (igrab(&dax_dev->inode))
	273	found = dax_dev;
	274	break;
	275	}
	276	spin_unlock(&dax_host_lock);
	277	dax_read_unlock(id);
	278
	279	return found;
	280	}
	281	EXPORT_SYMBOL_GPL(dax_get_by_host);
	282
7b6be844 DW	283	/**
	284	* inode_dax: convert a public inode into its dax_dev
	285	* @inode: An inode with i_cdev pointing to a dax_dev
	286	*
	287	* Note this is not equivalent to to_dax_dev() which is for private
	288	* internal use where we know the inode filesystem type == dax_fs_type.
	289	*/
	290	struct dax_device inode_dax(struct inode inode)
	291	{
	292	struct cdev *cdev = inode->i_cdev;
	293
	294	return container_of(cdev, struct dax_device, cdev);
	295	}
	296	EXPORT_SYMBOL_GPL(inode_dax);
	297
	298	struct inode dax_inode(struct dax_device dax_dev)
	299	{
	300	return &dax_dev->inode;
	301	}
	302	EXPORT_SYMBOL_GPL(dax_inode);
	303
	304	void dax_get_private(struct dax_device dax_dev)
	305	{
	306	return dax_dev->private;
	307	}
	308	EXPORT_SYMBOL_GPL(dax_get_private);
	309
	310	static void init_once(void *_dax_dev)
	311	{
	312	struct dax_device *dax_dev = _dax_dev;
	313	struct inode *inode = &dax_dev->inode;
	314
	315	inode_init_once(inode);
	316	}
	317
	318	static int __dax_fs_init(void)
	319	{
	320	int rc;
	321
	322	dax_cache = kmem_cache_create("dax_cache", sizeof(struct dax_device), 0,
	323	(SLAB_HWCACHE_ALIGN\|SLAB_RECLAIM_ACCOUNT\|
	324	SLAB_MEM_SPREAD\|SLAB_ACCOUNT),
	325	init_once);
	326	if (!dax_cache)
	327	return -ENOMEM;
	328
	329	rc = register_filesystem(&dax_fs_type);
	330	if (rc)
	331	goto err_register_fs;
	332
	333	dax_mnt = kern_mount(&dax_fs_type);
	334	if (IS_ERR(dax_mnt)) {
	335	rc = PTR_ERR(dax_mnt);
	336	goto err_mount;
	337	}
	338	dax_superblock = dax_mnt->mnt_sb;
	339
	340	return 0;
	341
	342	err_mount:
	343	unregister_filesystem(&dax_fs_type);
	344	err_register_fs:
	345	kmem_cache_destroy(dax_cache);
	346
347	return rc;
348	}
349
350	static void __dax_fs_exit(void)
351	{
352	kern_unmount(dax_mnt);
353	unregister_filesystem(&dax_fs_type);
354	kmem_cache_destroy(dax_cache);
355	}
356
357	static int __init dax_fs_init(void)
358	{
359	int rc;
360
361	rc = __dax_fs_init();
362	if (rc)
363	return rc;
364
365	nr_dax = max(nr_dax, 256);
366	rc = alloc_chrdev_region(&dax_devt, 0, nr_dax, "dax");
367	if (rc)
368	__dax_fs_exit();
369	return rc;
370	}
371
372	static void __exit dax_fs_exit(void)
373	{
374	unregister_chrdev_region(dax_devt, nr_dax);
375	ida_destroy(&dax_minor_ida);
376	__dax_fs_exit();
377	}
378
379	MODULE_AUTHOR("Intel Corporation");
380	MODULE_LICENSE("GPL v2");
381	subsys_initcall(dax_fs_init);
382	module_exit(dax_fs_exit);