[linux.git] / fs / nfsd / blocklayout.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (c) 2014-2016 Christoph Hellwig.
 */
#include <linux/exportfs.h>
#include <linux/iomap.h>
#include <linux/slab.h>
#include <linux/pr.h>

#include <linux/nfsd/debug.h>

#include "blocklayoutxdr.h"
#include "pnfs.h"
#include "filecache.h"
#include "vfs.h"

#define NFSDDBG_FACILITY	NFSDDBG_PNFS


static __be32
nfsd4_block_proc_layoutget(struct inode *inode, const struct svc_fh *fhp,
		struct nfsd4_layoutget *args)
{
	struct nfsd4_layout_seg *seg = &args->lg_seg;
	struct super_block *sb = inode->i_sb;
	u32 block_size = i_blocksize(inode);
	struct pnfs_block_extent *bex;
	struct iomap iomap;
	u32 device_generation = 0;
	int error;

	if (seg->offset & (block_size - 1)) {
		dprintk("pnfsd: I/O misaligned\n");
		goto out_layoutunavailable;
	}

	/*
	 * Some clients barf on non-zero block numbers for NONE or INVALID
	 * layouts, so make sure to zero the whole structure.
	 */
	error = -ENOMEM;
	bex = kzalloc(sizeof(*bex), GFP_KERNEL);
	if (!bex)
		goto out_error;
	args->lg_content = bex;

	error = sb->s_export_op->map_blocks(inode, seg->offset, seg->length,
					    &iomap, seg->iomode != IOMODE_READ,
					    &device_generation);
	if (error) {
		if (error == -ENXIO)
			goto out_layoutunavailable;
		goto out_error;
	}

	if (iomap.length < args->lg_minlength) {
		dprintk("pnfsd: extent smaller than minlength\n");
		goto out_layoutunavailable;
	}

	switch (iomap.type) {
	case IOMAP_MAPPED:
		if (seg->iomode == IOMODE_READ)
			bex->es = PNFS_BLOCK_READ_DATA;
		else
			bex->es = PNFS_BLOCK_READWRITE_DATA;
		bex->soff = iomap.addr;
		break;
	case IOMAP_UNWRITTEN:
		if (seg->iomode & IOMODE_RW) {
			/*
			 * Crack monkey special case from section 2.3.1.
			 */
			if (args->lg_minlength == 0) {
				dprintk("pnfsd: no soup for you!\n");
				goto out_layoutunavailable;
			}

			bex->es = PNFS_BLOCK_INVALID_DATA;
			bex->soff = iomap.addr;
			break;
		}
		fallthrough;
	case IOMAP_HOLE:
		if (seg->iomode == IOMODE_READ) {
			bex->es = PNFS_BLOCK_NONE_DATA;
			break;
		}
		fallthrough;
	case IOMAP_DELALLOC:
	default:
		WARN(1, "pnfsd: filesystem returned %d extent\n", iomap.type);
		goto out_layoutunavailable;
	}

	error = nfsd4_set_deviceid(&bex->vol_id, fhp, device_generation);
	if (error)
		goto out_error;
	bex->foff = iomap.offset;
	bex->len = iomap.length;

	seg->offset = iomap.offset;
	seg->length = iomap.length;

	dprintk("GET: 0x%llx:0x%llx %d\n", bex->foff, bex->len, bex->es);
	return 0;

out_error:
	seg->length = 0;
	return nfserrno(error);
out_layoutunavailable:
	seg->length = 0;
	return nfserr_layoutunavailable;
}

static __be32
nfsd4_block_commit_blocks(struct inode *inode, struct nfsd4_layoutcommit *lcp,
		struct iomap *iomaps, int nr_iomaps)
{
	struct timespec64 mtime = inode_get_mtime(inode);
	loff_t new_size = lcp->lc_last_wr + 1;
	struct iattr iattr = { .ia_valid = 0 };
	int error;

	if (lcp->lc_mtime.tv_nsec == UTIME_NOW ||
	    timespec64_compare(&lcp->lc_mtime, &mtime) < 0)
		lcp->lc_mtime = current_time(inode);
	iattr.ia_valid |= ATTR_ATIME | ATTR_CTIME | ATTR_MTIME;
	iattr.ia_atime = iattr.ia_ctime = iattr.ia_mtime = lcp->lc_mtime;

	if (new_size > i_size_read(inode)) {
		iattr.ia_valid |= ATTR_SIZE;
		iattr.ia_size = new_size;
	}

	error = inode->i_sb->s_export_op->commit_blocks(inode, iomaps,
			nr_iomaps, &iattr);
	kfree(iomaps);
	return nfserrno(error);
}

#ifdef CONFIG_NFSD_BLOCKLAYOUT
static int
nfsd4_block_get_device_info_simple(struct super_block *sb,
		struct nfsd4_getdeviceinfo *gdp)
{
	struct pnfs_block_deviceaddr *dev;
	struct pnfs_block_volume *b;

	dev = kzalloc(sizeof(struct pnfs_block_deviceaddr) +
		      sizeof(struct pnfs_block_volume), GFP_KERNEL);
	if (!dev)
		return -ENOMEM;
	gdp->gd_device = dev;

	dev->nr_volumes = 1;
	b = &dev->volumes[0];

	b->type = PNFS_BLOCK_VOLUME_SIMPLE;
	b->simple.sig_len = PNFS_BLOCK_UUID_LEN;
	return sb->s_export_op->get_uuid(sb, b->simple.sig, &b->simple.sig_len,
			&b->simple.offset);
}

static __be32
nfsd4_block_proc_getdeviceinfo(struct super_block *sb,
		struct svc_rqst *rqstp,
		struct nfs4_client *clp,
		struct nfsd4_getdeviceinfo *gdp)
{
	if (bdev_is_partition(sb->s_bdev))
		return nfserr_inval;
	return nfserrno(nfsd4_block_get_device_info_simple(sb, gdp));
}

static __be32
nfsd4_block_proc_layoutcommit(struct inode *inode,
		struct nfsd4_layoutcommit *lcp)
{
	struct iomap *iomaps;
	int nr_iomaps;

	nr_iomaps = nfsd4_block_decode_layoutupdate(lcp->lc_up_layout,
			lcp->lc_up_len, &iomaps, i_blocksize(inode));
	if (nr_iomaps < 0)
		return nfserrno(nr_iomaps);

	return nfsd4_block_commit_blocks(inode, lcp, iomaps, nr_iomaps);
}

const struct nfsd4_layout_ops bl_layout_ops = {
	/*
	 * Pretend that we send notification to the client.  This is a blatant
	 * lie to force recent Linux clients to cache our device IDs.
	 * We rarely ever change the device ID, so the harm of leaking deviceids
	 * for a while isn't too bad.  Unfortunately RFC5661 is a complete mess
	 * in this regard, but I filed errata 4119 for this a while ago, and
	 * hopefully the Linux client will eventually start caching deviceids
	 * without this again.
	 */
	.notify_types		=
			NOTIFY_DEVICEID4_DELETE | NOTIFY_DEVICEID4_CHANGE,
	.proc_getdeviceinfo	= nfsd4_block_proc_getdeviceinfo,
	.encode_getdeviceinfo	= nfsd4_block_encode_getdeviceinfo,
	.proc_layoutget		= nfsd4_block_proc_layoutget,
	.encode_layoutget	= nfsd4_block_encode_layoutget,
	.proc_layoutcommit	= nfsd4_block_proc_layoutcommit,
};
#endif /* CONFIG_NFSD_BLOCKLAYOUT */

#ifdef CONFIG_NFSD_SCSILAYOUT
#define NFSD_MDS_PR_KEY		0x0100000000000000ULL

/*
 * We use the client ID as a unique key for the reservations.
 * This allows us to easily fence a client when recalls fail.
 */
static u64 nfsd4_scsi_pr_key(struct nfs4_client *clp)
{
	return ((u64)clp->cl_clientid.cl_boot << 32) | clp->cl_clientid.cl_id;
}

static const u8 designator_types[] = {
	PS_DESIGNATOR_EUI64,
	PS_DESIGNATOR_NAA,
};

static int
nfsd4_block_get_unique_id(struct gendisk *disk, struct pnfs_block_volume *b)
{
	int ret, i;

	for (i = 0; i < ARRAY_SIZE(designator_types); i++) {
		u8 type = designator_types[i];

		ret = disk->fops->get_unique_id(disk, b->scsi.designator, type);
		if (ret > 0) {
			b->scsi.code_set = PS_CODE_SET_BINARY;
			b->scsi.designator_type = type;
			b->scsi.designator_len = ret;
			return 0;
		}
	}

	return -EINVAL;
}

static int
nfsd4_block_get_device_info_scsi(struct super_block *sb,
		struct nfs4_client *clp,
		struct nfsd4_getdeviceinfo *gdp)
{
	struct pnfs_block_deviceaddr *dev;
	struct pnfs_block_volume *b;
	const struct pr_ops *ops;
	int ret;

	dev = kzalloc(sizeof(struct pnfs_block_deviceaddr) +
		      sizeof(struct pnfs_block_volume), GFP_KERNEL);
	if (!dev)
		return -ENOMEM;
	gdp->gd_device = dev;

	dev->nr_volumes = 1;
	b = &dev->volumes[0];

	b->type = PNFS_BLOCK_VOLUME_SCSI;
	b->scsi.pr_key = nfsd4_scsi_pr_key(clp);

	ret = nfsd4_block_get_unique_id(sb->s_bdev->bd_disk, b);
	if (ret < 0)
		goto out_free_dev;

	ret = -EINVAL;
	ops = sb->s_bdev->bd_disk->fops->pr_ops;
	if (!ops) {
		pr_err("pNFS: device %s does not support PRs.\n",
			sb->s_id);
		goto out_free_dev;
	}

	ret = ops->pr_register(sb->s_bdev, 0, NFSD_MDS_PR_KEY, true);
	if (ret) {
		pr_err("pNFS: failed to register key for device %s.\n",
			sb->s_id);
		goto out_free_dev;
	}

	ret = ops->pr_reserve(sb->s_bdev, NFSD_MDS_PR_KEY,
			PR_EXCLUSIVE_ACCESS_REG_ONLY, 0);
	if (ret) {
		pr_err("pNFS: failed to reserve device %s.\n",
			sb->s_id);
		goto out_free_dev;
	}

	return 0;

out_free_dev:
	kfree(dev);
	gdp->gd_device = NULL;
	return ret;
}

static __be32
nfsd4_scsi_proc_getdeviceinfo(struct super_block *sb,
		struct svc_rqst *rqstp,
		struct nfs4_client *clp,
		struct nfsd4_getdeviceinfo *gdp)
{
	if (bdev_is_partition(sb->s_bdev))
		return nfserr_inval;
	return nfserrno(nfsd4_block_get_device_info_scsi(sb, clp, gdp));
}
static __be32
nfsd4_scsi_proc_layoutcommit(struct inode *inode,
		struct nfsd4_layoutcommit *lcp)
{
	struct iomap *iomaps;
	int nr_iomaps;

	nr_iomaps = nfsd4_scsi_decode_layoutupdate(lcp->lc_up_layout,
			lcp->lc_up_len, &iomaps, i_blocksize(inode));
	if (nr_iomaps < 0)
		return nfserrno(nr_iomaps);

	return nfsd4_block_commit_blocks(inode, lcp, iomaps, nr_iomaps);
}

static void
nfsd4_scsi_fence_client(struct nfs4_layout_stateid *ls)
{
	struct nfs4_client *clp = ls->ls_stid.sc_client;
	struct block_device *bdev = ls->ls_file->nf_file->f_path.mnt->mnt_sb->s_bdev;

	bdev->bd_disk->fops->pr_ops->pr_preempt(bdev, NFSD_MDS_PR_KEY,
			nfsd4_scsi_pr_key(clp), 0, true);
}

const struct nfsd4_layout_ops scsi_layout_ops = {
	/*
	 * Pretend that we send notification to the client.  This is a blatant
	 * lie to force recent Linux clients to cache our device IDs.
	 * We rarely ever change the device ID, so the harm of leaking deviceids
	 * for a while isn't too bad.  Unfortunately RFC5661 is a complete mess
	 * in this regard, but I filed errata 4119 for this a while ago, and
	 * hopefully the Linux client will eventually start caching deviceids
	 * without this again.
	 */
	.notify_types		=
			NOTIFY_DEVICEID4_DELETE | NOTIFY_DEVICEID4_CHANGE,
	.proc_getdeviceinfo	= nfsd4_scsi_proc_getdeviceinfo,
	.encode_getdeviceinfo	= nfsd4_block_encode_getdeviceinfo,
	.proc_layoutget		= nfsd4_block_proc_layoutget,
	.encode_layoutget	= nfsd4_block_encode_layoutget,
	.proc_layoutcommit	= nfsd4_scsi_proc_layoutcommit,
	.fence_client		= nfsd4_scsi_fence_client,
};
#endif /* CONFIG_NFSD_SCSILAYOUT */
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
8650b8a0	2	/*
f99d4fbd	3	* Copyright (c) 2014-2016 Christoph Hellwig.
8650b8a0 CH	4	*/
8650b8a0 CH	5	#include <linux/exportfs.h>
199a31c6	6	#include <linux/iomap.h>
8650b8a0	7	#include <linux/slab.h>
f99d4fbd	8	#include <linux/pr.h>
8650b8a0 CH	9
	10	#include <linux/nfsd/debug.h>
	11
	12	#include "blocklayoutxdr.h"
	13	#include "pnfs.h"
eb82dd39	14	#include "filecache.h"
cb12fae1	15	#include "vfs.h"
8650b8a0 CH	16
	17	#define NFSDDBG_FACILITY NFSDDBG_PNFS
	18
	19
8650b8a0 CH	20	static __be32
	21	nfsd4_block_proc_layoutget(struct inode inode, const struct svc_fh fhp,
	22	struct nfsd4_layoutget *args)
	23	{
	24	struct nfsd4_layout_seg *seg = &args->lg_seg;
	25	struct super_block *sb = inode->i_sb;
93407472	26	u32 block_size = i_blocksize(inode);
8650b8a0 CH	27	struct pnfs_block_extent *bex;
	28	struct iomap iomap;
	29	u32 device_generation = 0;
	30	int error;
	31
8650b8a0 CH	32	if (seg->offset & (block_size - 1)) {
	33	dprintk("pnfsd: I/O misaligned\n");
	34	goto out_layoutunavailable;
	35	}
	36
	37	/*
	38	* Some clients barf on non-zero block numbers for NONE or INVALID
	39	* layouts, so make sure to zero the whole structure.
	40	*/
	41	error = -ENOMEM;
	42	bex = kzalloc(sizeof(*bex), GFP_KERNEL);
	43	if (!bex)
	44	goto out_error;
	45	args->lg_content = bex;
	46
	47	error = sb->s_export_op->map_blocks(inode, seg->offset, seg->length,
	48	&iomap, seg->iomode != IOMODE_READ,
	49	&device_generation);
	50	if (error) {
	51	if (error == -ENXIO)
	52	goto out_layoutunavailable;
	53	goto out_error;
	54	}
	55
	56	if (iomap.length < args->lg_minlength) {
	57	dprintk("pnfsd: extent smaller than minlength\n");
	58	goto out_layoutunavailable;
	59	}
	60
	61	switch (iomap.type) {
	62	case IOMAP_MAPPED:
	63	if (seg->iomode == IOMODE_READ)
	64	bex->es = PNFS_BLOCK_READ_DATA;
	65	else
	66	bex->es = PNFS_BLOCK_READWRITE_DATA;
19fe5f64	67	bex->soff = iomap.addr;
8650b8a0 CH	68	break;
	69	case IOMAP_UNWRITTEN:
	70	if (seg->iomode & IOMODE_RW) {
	71	/*
	72	* Crack monkey special case from section 2.3.1.
	73	*/
	74	if (args->lg_minlength == 0) {
	75	dprintk("pnfsd: no soup for you!\n");
	76	goto out_layoutunavailable;
	77	}
	78
	79	bex->es = PNFS_BLOCK_INVALID_DATA;
19fe5f64	80	bex->soff = iomap.addr;
8650b8a0 CH	81	break;
8650b8a0 CH	82	}
df561f66	83	fallthrough;
8650b8a0 CH	84	case IOMAP_HOLE:
	85	if (seg->iomode == IOMODE_READ) {
	86	bex->es = PNFS_BLOCK_NONE_DATA;
	87	break;
	88	}
df561f66	89	fallthrough;
8650b8a0 CH	90	case IOMAP_DELALLOC:
	91	default:
	92	WARN(1, "pnfsd: filesystem returned %d extent\n", iomap.type);
	93	goto out_layoutunavailable;
	94	}
	95
	96	error = nfsd4_set_deviceid(&bex->vol_id, fhp, device_generation);
	97	if (error)
	98	goto out_error;
	99	bex->foff = iomap.offset;
	100	bex->len = iomap.length;
	101
	102	seg->offset = iomap.offset;
	103	seg->length = iomap.length;
	104
85369523	105	dprintk("GET: 0x%llx:0x%llx %d\n", bex->foff, bex->len, bex->es);
8650b8a0 CH	106	return 0;
	107
	108	out_error:
	109	seg->length = 0;
	110	return nfserrno(error);
	111	out_layoutunavailable:
	112	seg->length = 0;
	113	return nfserr_layoutunavailable;
	114	}
	115
	116	static __be32
368248ee CH	117	nfsd4_block_commit_blocks(struct inode inode, struct nfsd4_layoutcommit lcp,
368248ee CH	118	struct iomap *iomaps, int nr_iomaps)
8650b8a0	119	{
11fec9b9	120	struct timespec64 mtime = inode_get_mtime(inode);
8650b8a0 CH	121	loff_t new_size = lcp->lc_last_wr + 1;
8650b8a0 CH	122	struct iattr iattr = { .ia_valid = 0 };
8650b8a0 CH	123	int error;
8650b8a0 CH	124
8650b8a0	125	if (lcp->lc_mtime.tv_nsec == UTIME_NOW \|\|
11fec9b9	126	timespec64_compare(&lcp->lc_mtime, &mtime) < 0)
bdba5368	127	lcp->lc_mtime = current_time(inode);
8650b8a0	128	iattr.ia_valid \|= ATTR_ATIME \| ATTR_CTIME \| ATTR_MTIME;
bdba5368	129	iattr.ia_atime = iattr.ia_ctime = iattr.ia_mtime = lcp->lc_mtime;
8650b8a0 CH	130
	131	if (new_size > i_size_read(inode)) {
	132	iattr.ia_valid \|= ATTR_SIZE;
	133	iattr.ia_size = new_size;
	134	}
	135
	136	error = inode->i_sb->s_export_op->commit_blocks(inode, iomaps,
	137	nr_iomaps, &iattr);
	138	kfree(iomaps);
	139	return nfserrno(error);
	140	}
	141
368248ee CH	142	#ifdef CONFIG_NFSD_BLOCKLAYOUT
	143	static int
	144	nfsd4_block_get_device_info_simple(struct super_block *sb,
	145	struct nfsd4_getdeviceinfo *gdp)
	146	{
	147	struct pnfs_block_deviceaddr *dev;
	148	struct pnfs_block_volume *b;
	149
	150	dev = kzalloc(sizeof(struct pnfs_block_deviceaddr) +
	151	sizeof(struct pnfs_block_volume), GFP_KERNEL);
	152	if (!dev)
	153	return -ENOMEM;
	154	gdp->gd_device = dev;
	155
	156	dev->nr_volumes = 1;
	157	b = &dev->volumes[0];
	158
	159	b->type = PNFS_BLOCK_VOLUME_SIMPLE;
	160	b->simple.sig_len = PNFS_BLOCK_UUID_LEN;
	161	return sb->s_export_op->get_uuid(sb, b->simple.sig, &b->simple.sig_len,
	162	&b->simple.offset);
	163	}
	164
	165	static __be32
	166	nfsd4_block_proc_getdeviceinfo(struct super_block *sb,
d7c920d1	167	struct svc_rqst *rqstp,
f99d4fbd	168	struct nfs4_client *clp,
368248ee CH	169	struct nfsd4_getdeviceinfo *gdp)
368248ee CH	170	{
fa01b1e9	171	if (bdev_is_partition(sb->s_bdev))
368248ee CH	172	return nfserr_inval;
	173	return nfserrno(nfsd4_block_get_device_info_simple(sb, gdp));
	174	}
	175
	176	static __be32
	177	nfsd4_block_proc_layoutcommit(struct inode *inode,
	178	struct nfsd4_layoutcommit *lcp)
	179	{
	180	struct iomap *iomaps;
	181	int nr_iomaps;
	182
	183	nr_iomaps = nfsd4_block_decode_layoutupdate(lcp->lc_up_layout,
93407472	184	lcp->lc_up_len, &iomaps, i_blocksize(inode));
368248ee CH	185	if (nr_iomaps < 0)
	186	return nfserrno(nr_iomaps);
	187
	188	return nfsd4_block_commit_blocks(inode, lcp, iomaps, nr_iomaps);
	189	}
	190
8650b8a0	191	const struct nfsd4_layout_ops bl_layout_ops = {
40cdc7a5 CH	192	/*
	193	* Pretend that we send notification to the client. This is a blatant
	194	* lie to force recent Linux clients to cache our device IDs.
	195	* We rarely ever change the device ID, so the harm of leaking deviceids
	196	* for a while isn't too bad. Unfortunately RFC5661 is a complete mess
	197	* in this regard, but I filed errata 4119 for this a while ago, and
	198	* hopefully the Linux client will eventually start caching deviceids
	199	* without this again.
	200	*/
	201	.notify_types =
	202	NOTIFY_DEVICEID4_DELETE \| NOTIFY_DEVICEID4_CHANGE,
8650b8a0 CH	203	.proc_getdeviceinfo = nfsd4_block_proc_getdeviceinfo,
	204	.encode_getdeviceinfo = nfsd4_block_encode_getdeviceinfo,
	205	.proc_layoutget = nfsd4_block_proc_layoutget,
	206	.encode_layoutget = nfsd4_block_encode_layoutget,
	207	.proc_layoutcommit = nfsd4_block_proc_layoutcommit,
	208	};
368248ee	209	#endif /* CONFIG_NFSD_BLOCKLAYOUT */
f99d4fbd CH	210
f99d4fbd CH	211	#ifdef CONFIG_NFSD_SCSILAYOUT
eee93016	212	#define NFSD_MDS_PR_KEY 0x0100000000000000ULL
f99d4fbd CH	213
	214	/*
	215	* We use the client ID as a unique key for the reservations.
	216	* This allows us to easily fence a client when recalls fail.
	217	*/
	218	static u64 nfsd4_scsi_pr_key(struct nfs4_client *clp)
	219	{
	220	return ((u64)clp->cl_clientid.cl_boot << 32) \| clp->cl_clientid.cl_id;
	221	}
	222
8c6aabd1 CH	223	static const u8 designator_types[] = {
	224	PS_DESIGNATOR_EUI64,
	225	PS_DESIGNATOR_NAA,
	226	};
	227
	228	static int
	229	nfsd4_block_get_unique_id(struct gendisk disk, struct pnfs_block_volume b)
	230	{
	231	int ret, i;
	232
	233	for (i = 0; i < ARRAY_SIZE(designator_types); i++) {
	234	u8 type = designator_types[i];
	235
	236	ret = disk->fops->get_unique_id(disk, b->scsi.designator, type);
	237	if (ret > 0) {
	238	b->scsi.code_set = PS_CODE_SET_BINARY;
	239	b->scsi.designator_type = type;
	240	b->scsi.designator_len = ret;
	241	return 0;
	242	}
	243	}
	244
	245	return -EINVAL;
	246	}
	247
f99d4fbd CH	248	static int
	249	nfsd4_block_get_device_info_scsi(struct super_block *sb,
	250	struct nfs4_client *clp,
	251	struct nfsd4_getdeviceinfo *gdp)
	252	{
	253	struct pnfs_block_deviceaddr *dev;
	254	struct pnfs_block_volume *b;
	255	const struct pr_ops *ops;
8c6aabd1	256	int ret;
f99d4fbd CH	257
	258	dev = kzalloc(sizeof(struct pnfs_block_deviceaddr) +
	259	sizeof(struct pnfs_block_volume), GFP_KERNEL);
	260	if (!dev)
	261	return -ENOMEM;
	262	gdp->gd_device = dev;
	263
	264	dev->nr_volumes = 1;
	265	b = &dev->volumes[0];
	266
	267	b->type = PNFS_BLOCK_VOLUME_SCSI;
	268	b->scsi.pr_key = nfsd4_scsi_pr_key(clp);
	269
8c6aabd1 CH	270	ret = nfsd4_block_get_unique_id(sb->s_bdev->bd_disk, b);
	271	if (ret < 0)
	272	goto out_free_dev;
f99d4fbd	273
8c6aabd1	274	ret = -EINVAL;
f99d4fbd CH	275	ops = sb->s_bdev->bd_disk->fops->pr_ops;
	276	if (!ops) {
	277	pr_err("pNFS: device %s does not support PRs.\n",
	278	sb->s_id);
8c6aabd1	279	goto out_free_dev;
f99d4fbd CH	280	}
f99d4fbd CH	281
8c6aabd1 CH	282	ret = ops->pr_register(sb->s_bdev, 0, NFSD_MDS_PR_KEY, true);
8c6aabd1 CH	283	if (ret) {
f99d4fbd CH	284	pr_err("pNFS: failed to register key for device %s.\n",
f99d4fbd CH	285	sb->s_id);
8c6aabd1	286	goto out_free_dev;
f99d4fbd CH	287	}
f99d4fbd CH	288
8c6aabd1	289	ret = ops->pr_reserve(sb->s_bdev, NFSD_MDS_PR_KEY,
f99d4fbd	290	PR_EXCLUSIVE_ACCESS_REG_ONLY, 0);
8c6aabd1	291	if (ret) {
f99d4fbd CH	292	pr_err("pNFS: failed to reserve device %s.\n",
f99d4fbd CH	293	sb->s_id);
8c6aabd1	294	goto out_free_dev;
f99d4fbd CH	295	}
	296
	297	return 0;
8c6aabd1 CH	298
	299	out_free_dev:
	300	kfree(dev);
15a8b55d	301	gdp->gd_device = NULL;
8c6aabd1	302	return ret;
f99d4fbd CH	303	}
	304
	305	static __be32
	306	nfsd4_scsi_proc_getdeviceinfo(struct super_block *sb,
d7c920d1	307	struct svc_rqst *rqstp,
f99d4fbd CH	308	struct nfs4_client *clp,
	309	struct nfsd4_getdeviceinfo *gdp)
	310	{
fa01b1e9	311	if (bdev_is_partition(sb->s_bdev))
f99d4fbd CH	312	return nfserr_inval;
	313	return nfserrno(nfsd4_block_get_device_info_scsi(sb, clp, gdp));
	314	}
	315	static __be32
	316	nfsd4_scsi_proc_layoutcommit(struct inode *inode,
	317	struct nfsd4_layoutcommit *lcp)
	318	{
	319	struct iomap *iomaps;
	320	int nr_iomaps;
	321
	322	nr_iomaps = nfsd4_scsi_decode_layoutupdate(lcp->lc_up_layout,
93407472	323	lcp->lc_up_len, &iomaps, i_blocksize(inode));
f99d4fbd CH	324	if (nr_iomaps < 0)
	325	return nfserrno(nr_iomaps);
	326
	327	return nfsd4_block_commit_blocks(inode, lcp, iomaps, nr_iomaps);
	328	}
	329
	330	static void
	331	nfsd4_scsi_fence_client(struct nfs4_layout_stateid *ls)
	332	{
	333	struct nfs4_client *clp = ls->ls_stid.sc_client;
eb82dd39	334	struct block_device *bdev = ls->ls_file->nf_file->f_path.mnt->mnt_sb->s_bdev;
f99d4fbd CH	335
	336	bdev->bd_disk->fops->pr_ops->pr_preempt(bdev, NFSD_MDS_PR_KEY,
	337	nfsd4_scsi_pr_key(clp), 0, true);
	338	}
	339
	340	const struct nfsd4_layout_ops scsi_layout_ops = {
	341	/*
	342	* Pretend that we send notification to the client. This is a blatant
	343	* lie to force recent Linux clients to cache our device IDs.
	344	* We rarely ever change the device ID, so the harm of leaking deviceids
	345	* for a while isn't too bad. Unfortunately RFC5661 is a complete mess
	346	* in this regard, but I filed errata 4119 for this a while ago, and
	347	* hopefully the Linux client will eventually start caching deviceids
	348	* without this again.
	349	*/
	350	.notify_types =
	351	NOTIFY_DEVICEID4_DELETE \| NOTIFY_DEVICEID4_CHANGE,
	352	.proc_getdeviceinfo = nfsd4_scsi_proc_getdeviceinfo,
	353	.encode_getdeviceinfo = nfsd4_block_encode_getdeviceinfo,
	354	.proc_layoutget = nfsd4_block_proc_layoutget,
	355	.encode_layoutget = nfsd4_block_encode_layoutget,
	356	.proc_layoutcommit = nfsd4_scsi_proc_layoutcommit,
	357	.fence_client = nfsd4_scsi_fence_client,
	358	};
	359	#endif /* CONFIG_NFSD_SCSILAYOUT */