[deliverable/linux.git] / fs / nfs / nfs4filelayoutdev.c

/*
 *  Device operations for the pnfs nfs4 file layout driver.
 *
 *  Copyright (c) 2002
 *  The Regents of the University of Michigan
 *  All Rights Reserved
 *
 *  Dean Hildebrand <dhildebz@umich.edu>
 *  Garth Goodson   <Garth.Goodson@netapp.com>
 *
 *  Permission is granted to use, copy, create derivative works, and
 *  redistribute this software and such derivative works for any purpose,
 *  so long as the name of the University of Michigan is not used in
 *  any advertising or publicity pertaining to the use or distribution
 *  of this software without specific, written prior authorization. If
 *  the above copyright notice or any other identification of the
 *  University of Michigan is included in any copy of any portion of
 *  this software, then the disclaimer below must also be included.
 *
 *  This software is provided as is, without representation or warranty
 *  of any kind either express or implied, including without limitation
 *  the implied warranties of merchantability, fitness for a particular
 *  purpose, or noninfringement.  The Regents of the University of
 *  Michigan shall not be liable for any damages, including special,
 *  indirect, incidental, or consequential damages, with respect to any
 *  claim arising out of or in connection with the use of the software,
 *  even if it has been or is hereafter advised of the possibility of
 *  such damages.
 */

#include <linux/nfs_fs.h>
#include <linux/vmalloc.h>

#include "internal.h"
#include "nfs4filelayout.h"

#define NFSDBG_FACILITY		NFSDBG_PNFS_LD

/*
 * Device ID RCU cache. A device ID is unique per client ID and layout type.
 */
#define NFS4_FL_DEVICE_ID_HASH_BITS	5
#define NFS4_FL_DEVICE_ID_HASH_SIZE	(1 << NFS4_FL_DEVICE_ID_HASH_BITS)
#define NFS4_FL_DEVICE_ID_HASH_MASK	(NFS4_FL_DEVICE_ID_HASH_SIZE - 1)

static inline u32
nfs4_fl_deviceid_hash(struct nfs4_deviceid *id)
{
	unsigned char *cptr = (unsigned char *)id->data;
	unsigned int nbytes = NFS4_DEVICEID4_SIZE;
	u32 x = 0;

	while (nbytes--) {
		x *= 37;
		x += *cptr++;
	}
	return x & NFS4_FL_DEVICE_ID_HASH_MASK;
}

static struct hlist_head filelayout_deviceid_cache[NFS4_FL_DEVICE_ID_HASH_SIZE];
static DEFINE_SPINLOCK(filelayout_deviceid_lock);

/*
 * Data server cache
 *
 * Data servers can be mapped to different device ids.
 * nfs4_pnfs_ds reference counting
 *   - set to 1 on allocation
 *   - incremented when a device id maps a data server already in the cache.
 *   - decremented when deviceid is removed from the cache.
 */
DEFINE_SPINLOCK(nfs4_ds_cache_lock);
static LIST_HEAD(nfs4_data_server_cache);

/* Debug routines */
void
print_ds(struct nfs4_pnfs_ds *ds)
{
	if (ds == NULL) {
		printk("%s NULL device\n", __func__);
		return;
	}
	printk("        ip_addr %x port %hu\n"
		"        ref count %d\n"
		"        client %p\n"
		"        cl_exchange_flags %x\n",
		ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
		atomic_read(&ds->ds_count), ds->ds_clp,
		ds->ds_clp ? ds->ds_clp->cl_exchange_flags : 0);
}

void
print_ds_list(struct nfs4_file_layout_dsaddr *dsaddr)
{
	int i;

	ifdebug(FACILITY) {
		printk("%s dsaddr->ds_num %d\n", __func__,
		       dsaddr->ds_num);
		for (i = 0; i < dsaddr->ds_num; i++)
			print_ds(dsaddr->ds_list[i]);
	}
}

void print_deviceid(struct nfs4_deviceid *id)
{
	u32 *p = (u32 *)id;

	dprintk("%s: device id= [%x%x%x%x]\n", __func__,
		p[0], p[1], p[2], p[3]);
}

/* nfs4_ds_cache_lock is held */
static struct nfs4_pnfs_ds *
_data_server_lookup_locked(u32 ip_addr, u32 port)
{
	struct nfs4_pnfs_ds *ds;

	dprintk("_data_server_lookup: ip_addr=%x port=%hu\n",
			ntohl(ip_addr), ntohs(port));

	list_for_each_entry(ds, &nfs4_data_server_cache, ds_node) {
		if (ds->ds_ip_addr == ip_addr &&
		    ds->ds_port == port) {
			return ds;
		}
	}
	return NULL;
}

/*
 * Create an rpc connection to the nfs4_pnfs_ds data server
 * Currently only support IPv4
 */
static int
nfs4_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds)
{
	struct nfs_client *clp;
	struct sockaddr_in sin;
	int status = 0;

	dprintk("--> %s ip:port %x:%hu au_flavor %d\n", __func__,
		ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
		mds_srv->nfs_client->cl_rpcclient->cl_auth->au_flavor);

	sin.sin_family = AF_INET;
	sin.sin_addr.s_addr = ds->ds_ip_addr;
	sin.sin_port = ds->ds_port;

	clp = nfs4_set_ds_client(mds_srv->nfs_client, (struct sockaddr *)&sin,
				 sizeof(sin), IPPROTO_TCP);
	if (IS_ERR(clp)) {
		status = PTR_ERR(clp);
		goto out;
	}

	if ((clp->cl_exchange_flags & EXCHGID4_FLAG_MASK_PNFS) != 0) {
		if (!is_ds_client(clp)) {
			status = -ENODEV;
			goto out_put;
		}
		ds->ds_clp = clp;
		dprintk("%s [existing] ip=%x, port=%hu\n", __func__,
			ntohl(ds->ds_ip_addr), ntohs(ds->ds_port));
		goto out;
	}

	/*
	 * Do not set NFS_CS_CHECK_LEASE_TIME instead set the DS lease to
	 * be equal to the MDS lease. Renewal is scheduled in create_session.
	 */
	spin_lock(&mds_srv->nfs_client->cl_lock);
	clp->cl_lease_time = mds_srv->nfs_client->cl_lease_time;
	spin_unlock(&mds_srv->nfs_client->cl_lock);
	clp->cl_last_renewal = jiffies;

	/* New nfs_client */
	status = nfs4_init_ds_session(clp);
	if (status)
		goto out_put;

	ds->ds_clp = clp;
	dprintk("%s [new] ip=%x, port=%hu\n", __func__, ntohl(ds->ds_ip_addr),
		ntohs(ds->ds_port));
out:
	return status;
out_put:
	nfs_put_client(clp);
	goto out;
}

static void
destroy_ds(struct nfs4_pnfs_ds *ds)
{
	dprintk("--> %s\n", __func__);
	ifdebug(FACILITY)
		print_ds(ds);

	if (ds->ds_clp)
		nfs_put_client(ds->ds_clp);
	kfree(ds);
}

static void
nfs4_fl_free_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
{
	struct nfs4_pnfs_ds *ds;
	int i;

	print_deviceid(&dsaddr->deviceid);

	for (i = 0; i < dsaddr->ds_num; i++) {
		ds = dsaddr->ds_list[i];
		if (ds != NULL) {
			if (atomic_dec_and_lock(&ds->ds_count,
						&nfs4_ds_cache_lock)) {
				list_del_init(&ds->ds_node);
				spin_unlock(&nfs4_ds_cache_lock);
				destroy_ds(ds);
			}
		}
	}
	kfree(dsaddr->stripe_indices);
	kfree(dsaddr);
}

static struct nfs4_pnfs_ds *
nfs4_pnfs_ds_add(struct inode *inode, u32 ip_addr, u32 port, gfp_t gfp_flags)
{
	struct nfs4_pnfs_ds *tmp_ds, *ds;

	ds = kzalloc(sizeof(*tmp_ds), gfp_flags);
	if (!ds)
		goto out;

	spin_lock(&nfs4_ds_cache_lock);
	tmp_ds = _data_server_lookup_locked(ip_addr, port);
	if (tmp_ds == NULL) {
		ds->ds_ip_addr = ip_addr;
		ds->ds_port = port;
		atomic_set(&ds->ds_count, 1);
		INIT_LIST_HEAD(&ds->ds_node);
		ds->ds_clp = NULL;
		list_add(&ds->ds_node, &nfs4_data_server_cache);
		dprintk("%s add new data server ip 0x%x\n", __func__,
			ds->ds_ip_addr);
	} else {
		kfree(ds);
		atomic_inc(&tmp_ds->ds_count);
		dprintk("%s data server found ip 0x%x, inc'ed ds_count to %d\n",
			__func__, tmp_ds->ds_ip_addr,
			atomic_read(&tmp_ds->ds_count));
		ds = tmp_ds;
	}
	spin_unlock(&nfs4_ds_cache_lock);
out:
	return ds;
}

/*
 * Currently only support ipv4, and one multi-path address.
 */
static struct nfs4_pnfs_ds *
decode_and_add_ds(struct xdr_stream *streamp, struct inode *inode, gfp_t gfp_flags)
{
	struct nfs4_pnfs_ds *ds = NULL;
	char *buf;
	const char *ipend, *pstr;
	u32 ip_addr, port;
	int nlen, rlen, i;
	int tmp[2];
	__be32 *p;

	/* r_netid */
	p = xdr_inline_decode(streamp, 4);
	if (unlikely(!p))
		goto out_err;
	nlen = be32_to_cpup(p++);

	p = xdr_inline_decode(streamp, nlen);
	if (unlikely(!p))
		goto out_err;

	/* Check that netid is "tcp" */
	if (nlen != 3 ||  memcmp((char *)p, "tcp", 3)) {
		dprintk("%s: ERROR: non ipv4 TCP r_netid\n", __func__);
		goto out_err;
	}

	/* r_addr */
	p = xdr_inline_decode(streamp, 4);
	if (unlikely(!p))
		goto out_err;
	rlen = be32_to_cpup(p);

	p = xdr_inline_decode(streamp, rlen);
	if (unlikely(!p))
		goto out_err;

	/* ipv6 length plus port is legal */
	if (rlen > INET6_ADDRSTRLEN + 8) {
		dprintk("%s: Invalid address, length %d\n", __func__,
			rlen);
		goto out_err;
	}
	buf = kmalloc(rlen + 1, gfp_flags);
	if (!buf) {
		dprintk("%s: Not enough memory\n", __func__);
		goto out_err;
	}
	buf[rlen] = '\0';
	memcpy(buf, p, rlen);

	/* replace the port dots with dashes for the in4_pton() delimiter*/
	for (i = 0; i < 2; i++) {
		char *res = strrchr(buf, '.');
		if (!res) {
			dprintk("%s: Failed finding expected dots in port\n",
				__func__);
			goto out_free;
		}
		*res = '-';
	}

	/* Currently only support ipv4 address */
	if (in4_pton(buf, rlen, (u8 *)&ip_addr, '-', &ipend) == 0) {
		dprintk("%s: Only ipv4 addresses supported\n", __func__);
		goto out_free;
	}

	/* port */
	pstr = ipend;
	sscanf(pstr, "-%d-%d", &tmp[0], &tmp[1]);
	port = htons((tmp[0] << 8) | (tmp[1]));

	ds = nfs4_pnfs_ds_add(inode, ip_addr, port, gfp_flags);
	dprintk("%s: Decoded address and port %s\n", __func__, buf);
out_free:
	kfree(buf);
out_err:
	return ds;
}

/* Decode opaque device data and return the result */
static struct nfs4_file_layout_dsaddr*
decode_device(struct inode *ino, struct pnfs_device *pdev, gfp_t gfp_flags)
{
	int i;
	u32 cnt, num;
	u8 *indexp;
	__be32 *p;
	u8 *stripe_indices;
	u8 max_stripe_index;
	struct nfs4_file_layout_dsaddr *dsaddr = NULL;
	struct xdr_stream stream;
	struct xdr_buf buf = {
		.pages = pdev->pages,
		.page_len = pdev->pglen,
		.buflen = pdev->pglen,
		.len = pdev->pglen,
	};
	struct page *scratch;

	/* set up xdr stream */
	scratch = alloc_page(gfp_flags);
	if (!scratch)
		goto out_err;

	xdr_init_decode(&stream, &buf, NULL);
	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);

	/* Get the stripe count (number of stripe index) */
	p = xdr_inline_decode(&stream, 4);
	if (unlikely(!p))
		goto out_err_free_scratch;

	cnt = be32_to_cpup(p);
	dprintk("%s stripe count  %d\n", __func__, cnt);
	if (cnt > NFS4_PNFS_MAX_STRIPE_CNT) {
		printk(KERN_WARNING "%s: stripe count %d greater than "
		       "supported maximum %d\n", __func__,
			cnt, NFS4_PNFS_MAX_STRIPE_CNT);
		goto out_err_free_scratch;
	}

	/* read stripe indices */
	stripe_indices = kcalloc(cnt, sizeof(u8), gfp_flags);
	if (!stripe_indices)
		goto out_err_free_scratch;

	p = xdr_inline_decode(&stream, cnt << 2);
	if (unlikely(!p))
		goto out_err_free_stripe_indices;

	indexp = &stripe_indices[0];
	max_stripe_index = 0;
	for (i = 0; i < cnt; i++) {
		*indexp = be32_to_cpup(p++);
		max_stripe_index = max(max_stripe_index, *indexp);
		indexp++;
	}

	/* Check the multipath list count */
	p = xdr_inline_decode(&stream, 4);
	if (unlikely(!p))
		goto out_err_free_stripe_indices;

	num = be32_to_cpup(p);
	dprintk("%s ds_num %u\n", __func__, num);
	if (num > NFS4_PNFS_MAX_MULTI_CNT) {
		printk(KERN_WARNING "%s: multipath count %d greater than "
			"supported maximum %d\n", __func__,
			num, NFS4_PNFS_MAX_MULTI_CNT);
		goto out_err_free_stripe_indices;
	}

	/* validate stripe indices are all < num */
	if (max_stripe_index >= num) {
		printk(KERN_WARNING "%s: stripe index %u >= num ds %u\n",
			__func__, max_stripe_index, num);
		goto out_err_free_stripe_indices;
	}

	dsaddr = kzalloc(sizeof(*dsaddr) +
			(sizeof(struct nfs4_pnfs_ds *) * (num - 1)),
			gfp_flags);
	if (!dsaddr)
		goto out_err_free_stripe_indices;

	dsaddr->stripe_count = cnt;
	dsaddr->stripe_indices = stripe_indices;
	stripe_indices = NULL;
	dsaddr->ds_num = num;

	memcpy(&dsaddr->deviceid, &pdev->dev_id, sizeof(pdev->dev_id));

	for (i = 0; i < dsaddr->ds_num; i++) {
		int j;
		u32 mp_count;

		p = xdr_inline_decode(&stream, 4);
		if (unlikely(!p))
			goto out_err_free_deviceid;

		mp_count = be32_to_cpup(p); /* multipath count */
		if (mp_count > 1) {
			printk(KERN_WARNING
			       "%s: Multipath count %d not supported, "
			       "skipping all greater than 1\n", __func__,
				mp_count);
		}
		for (j = 0; j < mp_count; j++) {
			if (j == 0) {
				dsaddr->ds_list[i] = decode_and_add_ds(&stream,
					ino, gfp_flags);
				if (dsaddr->ds_list[i] == NULL)
					goto out_err_free_deviceid;
			} else {
				u32 len;
				/* skip extra multipath */

				/* read len, skip */
				p = xdr_inline_decode(&stream, 4);
				if (unlikely(!p))
					goto out_err_free_deviceid;
				len = be32_to_cpup(p);

				p = xdr_inline_decode(&stream, len);
				if (unlikely(!p))
					goto out_err_free_deviceid;

				/* read len, skip */
				p = xdr_inline_decode(&stream, 4);
				if (unlikely(!p))
					goto out_err_free_deviceid;
				len = be32_to_cpup(p);

				p = xdr_inline_decode(&stream, len);
				if (unlikely(!p))
					goto out_err_free_deviceid;
			}
		}
	}

	__free_page(scratch);
	return dsaddr;

out_err_free_deviceid:
	nfs4_fl_free_deviceid(dsaddr);
	/* stripe_indicies was part of dsaddr */
	goto out_err_free_scratch;
out_err_free_stripe_indices:
	kfree(stripe_indices);
out_err_free_scratch:
	__free_page(scratch);
out_err:
	dprintk("%s ERROR: returning NULL\n", __func__);
	return NULL;
}

/*
 * Decode the opaque device specified in 'dev' and add it to the cache of
 * available devices.
 */
static struct nfs4_file_layout_dsaddr *
decode_and_add_device(struct inode *inode, struct pnfs_device *dev, gfp_t gfp_flags)
{
	struct nfs4_file_layout_dsaddr *d, *new;
	long hash;

	new = decode_device(inode, dev, gfp_flags);
	if (!new) {
		printk(KERN_WARNING "%s: Could not decode or add device\n",
			__func__);
		return NULL;
	}

	spin_lock(&filelayout_deviceid_lock);
	d = nfs4_fl_find_get_deviceid(&new->deviceid);
	if (d) {
		spin_unlock(&filelayout_deviceid_lock);
		nfs4_fl_free_deviceid(new);
		return d;
	}

	INIT_HLIST_NODE(&new->node);
	atomic_set(&new->ref, 1);
	hash = nfs4_fl_deviceid_hash(&new->deviceid);
	hlist_add_head_rcu(&new->node, &filelayout_deviceid_cache[hash]);
	spin_unlock(&filelayout_deviceid_lock);

	return new;
}

/*
 * Retrieve the information for dev_id, add it to the list
 * of available devices, and return it.
 */
struct nfs4_file_layout_dsaddr *
get_device_info(struct inode *inode, struct nfs4_deviceid *dev_id, gfp_t gfp_flags)
{
	struct pnfs_device *pdev = NULL;
	u32 max_resp_sz;
	int max_pages;
	struct page **pages = NULL;
	struct nfs4_file_layout_dsaddr *dsaddr = NULL;
	int rc, i;
	struct nfs_server *server = NFS_SERVER(inode);

	/*
	 * Use the session max response size as the basis for setting
	 * GETDEVICEINFO's maxcount
	 */
	max_resp_sz = server->nfs_client->cl_session->fc_attrs.max_resp_sz;
	max_pages = max_resp_sz >> PAGE_SHIFT;
	dprintk("%s inode %p max_resp_sz %u max_pages %d\n",
		__func__, inode, max_resp_sz, max_pages);

	pdev = kzalloc(sizeof(struct pnfs_device), gfp_flags);
	if (pdev == NULL)
		return NULL;

	pages = kzalloc(max_pages * sizeof(struct page *), gfp_flags);
	if (pages == NULL) {
		kfree(pdev);
		return NULL;
	}
	for (i = 0; i < max_pages; i++) {
		pages[i] = alloc_page(gfp_flags);
		if (!pages[i])
			goto out_free;
	}

	memcpy(&pdev->dev_id, dev_id, sizeof(*dev_id));
	pdev->layout_type = LAYOUT_NFSV4_1_FILES;
	pdev->pages = pages;
	pdev->pgbase = 0;
	pdev->pglen = PAGE_SIZE * max_pages;
	pdev->mincount = 0;

	rc = nfs4_proc_getdeviceinfo(server, pdev);
	dprintk("%s getdevice info returns %d\n", __func__, rc);
	if (rc)
		goto out_free;

	/*
	 * Found new device, need to decode it and then add it to the
	 * list of known devices for this mountpoint.
	 */
	dsaddr = decode_and_add_device(inode, pdev, gfp_flags);
out_free:
	for (i = 0; i < max_pages; i++)
		__free_page(pages[i]);
	kfree(pages);
	kfree(pdev);
	dprintk("<-- %s dsaddr %p\n", __func__, dsaddr);
	return dsaddr;
}

void
nfs4_fl_put_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
{
	if (atomic_dec_and_lock(&dsaddr->ref, &filelayout_deviceid_lock)) {
		hlist_del_rcu(&dsaddr->node);
		spin_unlock(&filelayout_deviceid_lock);

		synchronize_rcu();
		nfs4_fl_free_deviceid(dsaddr);
	}
}

struct nfs4_file_layout_dsaddr *
nfs4_fl_find_get_deviceid(struct nfs4_deviceid *id)
{
	struct nfs4_file_layout_dsaddr *d;
	struct hlist_node *n;
	long hash = nfs4_fl_deviceid_hash(id);


	rcu_read_lock();
	hlist_for_each_entry_rcu(d, n, &filelayout_deviceid_cache[hash], node) {
		if (!memcmp(&d->deviceid, id, sizeof(*id))) {
			if (!atomic_inc_not_zero(&d->ref))
				goto fail;
			rcu_read_unlock();
			return d;
		}
	}
fail:
	rcu_read_unlock();
	return NULL;
}

/*
 * Want res = (offset - layout->pattern_offset)/ layout->stripe_unit
 * Then: ((res + fsi) % dsaddr->stripe_count)
 */
u32
nfs4_fl_calc_j_index(struct pnfs_layout_segment *lseg, loff_t offset)
{
	struct nfs4_filelayout_segment *flseg = FILELAYOUT_LSEG(lseg);
	u64 tmp;

	tmp = offset - flseg->pattern_offset;
	do_div(tmp, flseg->stripe_unit);
	tmp += flseg->first_stripe_index;
	return do_div(tmp, flseg->dsaddr->stripe_count);
}

u32
nfs4_fl_calc_ds_index(struct pnfs_layout_segment *lseg, u32 j)
{
	return FILELAYOUT_LSEG(lseg)->dsaddr->stripe_indices[j];
}

struct nfs_fh *
nfs4_fl_select_ds_fh(struct pnfs_layout_segment *lseg, u32 j)
{
	struct nfs4_filelayout_segment *flseg = FILELAYOUT_LSEG(lseg);
	u32 i;

	if (flseg->stripe_type == STRIPE_SPARSE) {
		if (flseg->num_fh == 1)
			i = 0;
		else if (flseg->num_fh == 0)
			/* Use the MDS OPEN fh set in nfs_read_rpcsetup */
			return NULL;
		else
			i = nfs4_fl_calc_ds_index(lseg, j);
	} else
		i = j;
	return flseg->fh_array[i];
}

static void
filelayout_mark_devid_negative(struct nfs4_file_layout_dsaddr *dsaddr,
			       int err, u32 ds_addr)
{
	u32 *p = (u32 *)&dsaddr->deviceid;

	printk(KERN_ERR "NFS: data server %x connection error %d."
		" Deviceid [%x%x%x%x] marked out of use.\n",
		ds_addr, err, p[0], p[1], p[2], p[3]);

	spin_lock(&filelayout_deviceid_lock);
	dsaddr->flags |= NFS4_DEVICE_ID_NEG_ENTRY;
	spin_unlock(&filelayout_deviceid_lock);
}

struct nfs4_pnfs_ds *
nfs4_fl_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx)
{
	struct nfs4_file_layout_dsaddr *dsaddr = FILELAYOUT_LSEG(lseg)->dsaddr;
	struct nfs4_pnfs_ds *ds = dsaddr->ds_list[ds_idx];

	if (ds == NULL) {
		printk(KERN_ERR "%s: No data server for offset index %d\n",
			__func__, ds_idx);
		return NULL;
	}

	if (!ds->ds_clp) {
		struct nfs_server *s = NFS_SERVER(lseg->pls_layout->plh_inode);
		int err;

		if (dsaddr->flags & NFS4_DEVICE_ID_NEG_ENTRY) {
			/* Already tried to connect, don't try again */
			dprintk("%s Deviceid marked out of use\n", __func__);
			return NULL;
		}
		err = nfs4_ds_connect(s, ds);
		if (err) {
			filelayout_mark_devid_negative(dsaddr, err,
						       ntohl(ds->ds_ip_addr));
			return NULL;
		}
	}
	return ds;
}
Commit	Line	Data
16b374ca AA	1	/*
	2	* Device operations for the pnfs nfs4 file layout driver.
	3	*
	4	* Copyright (c) 2002
	5	* The Regents of the University of Michigan
	6	* All Rights Reserved
	7	*
	8	* Dean Hildebrand <dhildebz@umich.edu>
	9	* Garth Goodson <Garth.Goodson@netapp.com>
	10	*
	11	* Permission is granted to use, copy, create derivative works, and
	12	* redistribute this software and such derivative works for any purpose,
	13	* so long as the name of the University of Michigan is not used in
	14	* any advertising or publicity pertaining to the use or distribution
	15	* of this software without specific, written prior authorization. If
	16	* the above copyright notice or any other identification of the
	17	* University of Michigan is included in any copy of any portion of
	18	* this software, then the disclaimer below must also be included.
	19	*
	20	* This software is provided as is, without representation or warranty
	21	* of any kind either express or implied, including without limitation
	22	* the implied warranties of merchantability, fitness for a particular
	23	* purpose, or noninfringement. The Regents of the University of
	24	* Michigan shall not be liable for any damages, including special,
	25	* indirect, incidental, or consequential damages, with respect to any
	26	* claim arising out of or in connection with the use of the software,
	27	* even if it has been or is hereafter advised of the possibility of
	28	* such damages.
	29	*/
	30
	31	#include <linux/nfs_fs.h>
	32	#include <linux/vmalloc.h>
	33
	34	#include "internal.h"
	35	#include "nfs4filelayout.h"
	36
	37	#define NFSDBG_FACILITY NFSDBG_PNFS_LD
	38
ea8eecdd CH	39	/*
	40	* Device ID RCU cache. A device ID is unique per client ID and layout type.
	41	*/
	42	#define NFS4_FL_DEVICE_ID_HASH_BITS 5
	43	#define NFS4_FL_DEVICE_ID_HASH_SIZE (1 << NFS4_FL_DEVICE_ID_HASH_BITS)
	44	#define NFS4_FL_DEVICE_ID_HASH_MASK (NFS4_FL_DEVICE_ID_HASH_SIZE - 1)
	45
	46	static inline u32
	47	nfs4_fl_deviceid_hash(struct nfs4_deviceid *id)
	48	{
	49	unsigned char cptr = (unsigned char )id->data;
	50	unsigned int nbytes = NFS4_DEVICEID4_SIZE;
	51	u32 x = 0;
	52
	53	while (nbytes--) {
	54	x *= 37;
	55	x += *cptr++;
	56	}
	57	return x & NFS4_FL_DEVICE_ID_HASH_MASK;
	58	}
	59
	60	static struct hlist_head filelayout_deviceid_cache[NFS4_FL_DEVICE_ID_HASH_SIZE];
	61	static DEFINE_SPINLOCK(filelayout_deviceid_lock);
	62
16b374ca AA	63	/*
	64	* Data server cache
	65	*
	66	* Data servers can be mapped to different device ids.
	67	* nfs4_pnfs_ds reference counting
	68	* - set to 1 on allocation
	69	* - incremented when a device id maps a data server already in the cache.
	70	* - decremented when deviceid is removed from the cache.
	71	*/
	72	DEFINE_SPINLOCK(nfs4_ds_cache_lock);
	73	static LIST_HEAD(nfs4_data_server_cache);
	74
	75	/* Debug routines */
	76	void
	77	print_ds(struct nfs4_pnfs_ds *ds)
	78	{
	79	if (ds == NULL) {
	80	printk("%s NULL device\n", __func__);
	81	return;
	82	}
	83	printk(" ip_addr %x port %hu\n"
	84	" ref count %d\n"
	85	" client %p\n"
	86	" cl_exchange_flags %x\n",
	87	ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
	88	atomic_read(&ds->ds_count), ds->ds_clp,
	89	ds->ds_clp ? ds->ds_clp->cl_exchange_flags : 0);
	90	}
	91
	92	void
	93	print_ds_list(struct nfs4_file_layout_dsaddr *dsaddr)
	94	{
	95	int i;
	96
	97	ifdebug(FACILITY) {
	98	printk("%s dsaddr->ds_num %d\n", __func__,
	99	dsaddr->ds_num);
	100	for (i = 0; i < dsaddr->ds_num; i++)
	101	print_ds(dsaddr->ds_list[i]);
	102	}
	103	}
	104
	105	void print_deviceid(struct nfs4_deviceid *id)
	106	{
	107	u32 p = (u32 )id;
	108
	109	dprintk("%s: device id= [%x%x%x%x]\n", __func__,
	110	p[0], p[1], p[2], p[3]);
	111	}
	112
	113	/* nfs4_ds_cache_lock is held */
	114	static struct nfs4_pnfs_ds *
	115	_data_server_lookup_locked(u32 ip_addr, u32 port)
	116	{
	117	struct nfs4_pnfs_ds *ds;
	118
	119	dprintk("_data_server_lookup: ip_addr=%x port=%hu\n",
	120	ntohl(ip_addr), ntohs(port));
	121
	122	list_for_each_entry(ds, &nfs4_data_server_cache, ds_node) {
	123	if (ds->ds_ip_addr == ip_addr &&
	124	ds->ds_port == port) {
	125	return ds;
	126	}
127	}
128	return NULL;
129	}
130
d83217c1 AA	131	/*
	132	* Create an rpc connection to the nfs4_pnfs_ds data server
	133	* Currently only support IPv4
	134	*/
	135	static int
	136	nfs4_ds_connect(struct nfs_server mds_srv, struct nfs4_pnfs_ds ds)
	137	{
	138	struct nfs_client *clp;
	139	struct sockaddr_in sin;
	140	int status = 0;
	141
	142	dprintk("--> %s ip:port %x:%hu au_flavor %d\n", __func__,
	143	ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
	144	mds_srv->nfs_client->cl_rpcclient->cl_auth->au_flavor);
	145
	146	sin.sin_family = AF_INET;
	147	sin.sin_addr.s_addr = ds->ds_ip_addr;
	148	sin.sin_port = ds->ds_port;
	149
	150	clp = nfs4_set_ds_client(mds_srv->nfs_client, (struct sockaddr *)&sin,
	151	sizeof(sin), IPPROTO_TCP);
	152	if (IS_ERR(clp)) {
	153	status = PTR_ERR(clp);
	154	goto out;
	155	}
	156
	157	if ((clp->cl_exchange_flags & EXCHGID4_FLAG_MASK_PNFS) != 0) {
	158	if (!is_ds_client(clp)) {
	159	status = -ENODEV;
	160	goto out_put;
	161	}
	162	ds->ds_clp = clp;
	163	dprintk("%s [existing] ip=%x, port=%hu\n", __func__,
	164	ntohl(ds->ds_ip_addr), ntohs(ds->ds_port));
	165	goto out;
	166	}
	167
	168	/*
	169	* Do not set NFS_CS_CHECK_LEASE_TIME instead set the DS lease to
	170	* be equal to the MDS lease. Renewal is scheduled in create_session.
	171	*/
	172	spin_lock(&mds_srv->nfs_client->cl_lock);
	173	clp->cl_lease_time = mds_srv->nfs_client->cl_lease_time;
	174	spin_unlock(&mds_srv->nfs_client->cl_lock);
	175	clp->cl_last_renewal = jiffies;
	176
	177	/* New nfs_client */
	178	status = nfs4_init_ds_session(clp);
	179	if (status)
	180	goto out_put;
	181
	182	ds->ds_clp = clp;
	183	dprintk("%s [new] ip=%x, port=%hu\n", __func__, ntohl(ds->ds_ip_addr),
	184	ntohs(ds->ds_port));
	185	out:
	186	return status;
	187	out_put:
	188	nfs_put_client(clp);
	189	goto out;
	190	}
	191
16b374ca AA	192	static void
	193	destroy_ds(struct nfs4_pnfs_ds *ds)
	194	{
	195	dprintk("--> %s\n", __func__);
	196	ifdebug(FACILITY)
	197	print_ds(ds);
	198
	199	if (ds->ds_clp)
	200	nfs_put_client(ds->ds_clp);
	201	kfree(ds);
	202	}
	203
	204	static void
	205	nfs4_fl_free_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
	206	{
	207	struct nfs4_pnfs_ds *ds;
	208	int i;
	209
ea8eecdd	210	print_deviceid(&dsaddr->deviceid);
16b374ca AA	211
	212	for (i = 0; i < dsaddr->ds_num; i++) {
	213	ds = dsaddr->ds_list[i];
	214	if (ds != NULL) {
	215	if (atomic_dec_and_lock(&ds->ds_count,
	216	&nfs4_ds_cache_lock)) {
	217	list_del_init(&ds->ds_node);
	218	spin_unlock(&nfs4_ds_cache_lock);
	219	destroy_ds(ds);
	220	}
	221	}
	222	}
	223	kfree(dsaddr->stripe_indices);
	224	kfree(dsaddr);
	225	}
	226
16b374ca	227	static struct nfs4_pnfs_ds *
a75b9df9	228	nfs4_pnfs_ds_add(struct inode *inode, u32 ip_addr, u32 port, gfp_t gfp_flags)
16b374ca AA	229	{
	230	struct nfs4_pnfs_ds tmp_ds, ds;
	231
a75b9df9	232	ds = kzalloc(sizeof(*tmp_ds), gfp_flags);
16b374ca AA	233	if (!ds)
	234	goto out;
	235
	236	spin_lock(&nfs4_ds_cache_lock);
	237	tmp_ds = _data_server_lookup_locked(ip_addr, port);
	238	if (tmp_ds == NULL) {
	239	ds->ds_ip_addr = ip_addr;
	240	ds->ds_port = port;
	241	atomic_set(&ds->ds_count, 1);
	242	INIT_LIST_HEAD(&ds->ds_node);
	243	ds->ds_clp = NULL;
	244	list_add(&ds->ds_node, &nfs4_data_server_cache);
	245	dprintk("%s add new data server ip 0x%x\n", __func__,
	246	ds->ds_ip_addr);
	247	} else {
	248	kfree(ds);
	249	atomic_inc(&tmp_ds->ds_count);
	250	dprintk("%s data server found ip 0x%x, inc'ed ds_count to %d\n",
	251	__func__, tmp_ds->ds_ip_addr,
	252	atomic_read(&tmp_ds->ds_count));
	253	ds = tmp_ds;
	254	}
	255	spin_unlock(&nfs4_ds_cache_lock);
	256	out:
	257	return ds;
	258	}
	259
	260	/*
	261	* Currently only support ipv4, and one multi-path address.
	262	*/
	263	static struct nfs4_pnfs_ds *
a75b9df9	264	decode_and_add_ds(struct xdr_stream streamp, struct inode inode, gfp_t gfp_flags)
16b374ca AA	265	{
	266	struct nfs4_pnfs_ds *ds = NULL;
	267	char *buf;
	268	const char ipend, pstr;
	269	u32 ip_addr, port;
	270	int nlen, rlen, i;
	271	int tmp[2];
35124a09	272	__be32 *p;
16b374ca AA	273
16b374ca AA	274	/* r_netid */
35124a09 WAA	275	p = xdr_inline_decode(streamp, 4);
	276	if (unlikely(!p))
	277	goto out_err;
16b374ca	278	nlen = be32_to_cpup(p++);
16b374ca	279
35124a09 WAA	280	p = xdr_inline_decode(streamp, nlen);
	281	if (unlikely(!p))
	282	goto out_err;
16b374ca AA	283
16b374ca AA	284	/* Check that netid is "tcp" */
35124a09	285	if (nlen != 3 \|\| memcmp((char *)p, "tcp", 3)) {
16b374ca AA	286	dprintk("%s: ERROR: non ipv4 TCP r_netid\n", __func__);
	287	goto out_err;
	288	}
	289
35124a09 WAA	290	/* r_addr */
	291	p = xdr_inline_decode(streamp, 4);
	292	if (unlikely(!p))
	293	goto out_err;
	294	rlen = be32_to_cpup(p);
	295
	296	p = xdr_inline_decode(streamp, rlen);
	297	if (unlikely(!p))
	298	goto out_err;
	299
16b374ca AA	300	/* ipv6 length plus port is legal */
16b374ca AA	301	if (rlen > INET6_ADDRSTRLEN + 8) {
ad3d2eed	302	dprintk("%s: Invalid address, length %d\n", __func__,
16b374ca AA	303	rlen);
	304	goto out_err;
	305	}
a75b9df9	306	buf = kmalloc(rlen + 1, gfp_flags);
b9f81057 SF	307	if (!buf) {
	308	dprintk("%s: Not enough memory\n", __func__);
	309	goto out_err;
	310	}
16b374ca	311	buf[rlen] = '\0';
35124a09	312	memcpy(buf, p, rlen);
16b374ca AA	313
	314	/* replace the port dots with dashes for the in4_pton() delimiter*/
	315	for (i = 0; i < 2; i++) {
	316	char *res = strrchr(buf, '.');
ad3d2eed JJ	317	if (!res) {
	318	dprintk("%s: Failed finding expected dots in port\n",
	319	__func__);
	320	goto out_free;
	321	}
16b374ca AA	322	*res = '-';
	323	}
	324
	325	/* Currently only support ipv4 address */
	326	if (in4_pton(buf, rlen, (u8 *)&ip_addr, '-', &ipend) == 0) {
	327	dprintk("%s: Only ipv4 addresses supported\n", __func__);
	328	goto out_free;
	329	}
	330
	331	/* port */
	332	pstr = ipend;
	333	sscanf(pstr, "-%d-%d", &tmp[0], &tmp[1]);
	334	port = htons((tmp[0] << 8) \| (tmp[1]));
	335
a75b9df9	336	ds = nfs4_pnfs_ds_add(inode, ip_addr, port, gfp_flags);
ad3d2eed	337	dprintk("%s: Decoded address and port %s\n", __func__, buf);
16b374ca AA	338	out_free:
	339	kfree(buf);
	340	out_err:
	341	return ds;
	342	}
	343
	344	/* Decode opaque device data and return the result */
	345	static struct nfs4_file_layout_dsaddr*
a75b9df9	346	decode_device(struct inode ino, struct pnfs_device pdev, gfp_t gfp_flags)
16b374ca	347	{
35124a09	348	int i;
16b374ca AA	349	u32 cnt, num;
16b374ca AA	350	u8 *indexp;
35124a09 WAA	351	__be32 *p;
	352	u8 *stripe_indices;
	353	u8 max_stripe_index;
	354	struct nfs4_file_layout_dsaddr *dsaddr = NULL;
	355	struct xdr_stream stream;
	356	struct xdr_buf buf = {
	357	.pages = pdev->pages,
	358	.page_len = pdev->pglen,
	359	.buflen = pdev->pglen,
	360	.len = pdev->pglen,
	361	};
	362	struct page *scratch;
	363
	364	/* set up xdr stream */
a75b9df9	365	scratch = alloc_page(gfp_flags);
35124a09 WAA	366	if (!scratch)
	367	goto out_err;
	368
	369	xdr_init_decode(&stream, &buf, NULL);
	370	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
16b374ca AA	371
16b374ca AA	372	/* Get the stripe count (number of stripe index) */
35124a09 WAA	373	p = xdr_inline_decode(&stream, 4);
	374	if (unlikely(!p))
	375	goto out_err_free_scratch;
	376
	377	cnt = be32_to_cpup(p);
16b374ca AA	378	dprintk("%s stripe count %d\n", __func__, cnt);
	379	if (cnt > NFS4_PNFS_MAX_STRIPE_CNT) {
	380	printk(KERN_WARNING "%s: stripe count %d greater than "
	381	"supported maximum %d\n", __func__,
	382	cnt, NFS4_PNFS_MAX_STRIPE_CNT);
35124a09 WAA	383	goto out_err_free_scratch;
	384	}
	385
	386	/* read stripe indices */
a75b9df9	387	stripe_indices = kcalloc(cnt, sizeof(u8), gfp_flags);
35124a09 WAA	388	if (!stripe_indices)
	389	goto out_err_free_scratch;
	390
	391	p = xdr_inline_decode(&stream, cnt << 2);
	392	if (unlikely(!p))
	393	goto out_err_free_stripe_indices;
	394
	395	indexp = &stripe_indices[0];
	396	max_stripe_index = 0;
	397	for (i = 0; i < cnt; i++) {
	398	*indexp = be32_to_cpup(p++);
	399	max_stripe_index = max(max_stripe_index, *indexp);
	400	indexp++;
16b374ca AA	401	}
	402
	403	/* Check the multipath list count */
35124a09 WAA	404	p = xdr_inline_decode(&stream, 4);
	405	if (unlikely(!p))
	406	goto out_err_free_stripe_indices;
	407
	408	num = be32_to_cpup(p);
16b374ca AA	409	dprintk("%s ds_num %u\n", __func__, num);
	410	if (num > NFS4_PNFS_MAX_MULTI_CNT) {
	411	printk(KERN_WARNING "%s: multipath count %d greater than "
	412	"supported maximum %d\n", __func__,
	413	num, NFS4_PNFS_MAX_MULTI_CNT);
35124a09	414	goto out_err_free_stripe_indices;
16b374ca	415	}
35124a09 WAA	416
	417	/* validate stripe indices are all < num */
	418	if (max_stripe_index >= num) {
	419	printk(KERN_WARNING "%s: stripe index %u >= num ds %u\n",
	420	__func__, max_stripe_index, num);
	421	goto out_err_free_stripe_indices;
	422	}
	423
16b374ca AA	424	dsaddr = kzalloc(sizeof(*dsaddr) +
16b374ca AA	425	(sizeof(struct nfs4_pnfs_ds ) (num - 1)),
a75b9df9	426	gfp_flags);
16b374ca	427	if (!dsaddr)
35124a09	428	goto out_err_free_stripe_indices;
16b374ca AA	429
16b374ca AA	430	dsaddr->stripe_count = cnt;
35124a09 WAA	431	dsaddr->stripe_indices = stripe_indices;
35124a09 WAA	432	stripe_indices = NULL;
16b374ca AA	433	dsaddr->ds_num = num;
16b374ca AA	434
ea8eecdd	435	memcpy(&dsaddr->deviceid, &pdev->dev_id, sizeof(pdev->dev_id));
16b374ca	436
16b374ca AA	437	for (i = 0; i < dsaddr->ds_num; i++) {
16b374ca AA	438	int j;
35124a09 WAA	439	u32 mp_count;
	440
	441	p = xdr_inline_decode(&stream, 4);
	442	if (unlikely(!p))
	443	goto out_err_free_deviceid;
16b374ca	444
35124a09 WAA	445	mp_count = be32_to_cpup(p); /* multipath count */
35124a09 WAA	446	if (mp_count > 1) {
16b374ca AA	447	printk(KERN_WARNING
	448	"%s: Multipath count %d not supported, "
	449	"skipping all greater than 1\n", __func__,
35124a09	450	mp_count);
16b374ca	451	}
35124a09	452	for (j = 0; j < mp_count; j++) {
16b374ca	453	if (j == 0) {
35124a09	454	dsaddr->ds_list[i] = decode_and_add_ds(&stream,
a75b9df9	455	ino, gfp_flags);
16b374ca	456	if (dsaddr->ds_list[i] == NULL)
35124a09	457	goto out_err_free_deviceid;
16b374ca AA	458	} else {
	459	u32 len;
	460	/* skip extra multipath */
35124a09 WAA	461
	462	/* read len, skip */
	463	p = xdr_inline_decode(&stream, 4);
	464	if (unlikely(!p))
	465	goto out_err_free_deviceid;
	466	len = be32_to_cpup(p);
	467
	468	p = xdr_inline_decode(&stream, len);
	469	if (unlikely(!p))
	470	goto out_err_free_deviceid;
	471
	472	/* read len, skip */
	473	p = xdr_inline_decode(&stream, 4);
	474	if (unlikely(!p))
	475	goto out_err_free_deviceid;
	476	len = be32_to_cpup(p);
	477
	478	p = xdr_inline_decode(&stream, len);
	479	if (unlikely(!p))
	480	goto out_err_free_deviceid;
16b374ca AA	481	}
	482	}
	483	}
35124a09 WAA	484
35124a09 WAA	485	__free_page(scratch);
16b374ca AA	486	return dsaddr;
16b374ca AA	487
35124a09	488	out_err_free_deviceid:
16b374ca	489	nfs4_fl_free_deviceid(dsaddr);
35124a09 WAA	490	/* stripe_indicies was part of dsaddr */
	491	goto out_err_free_scratch;
	492	out_err_free_stripe_indices:
	493	kfree(stripe_indices);
	494	out_err_free_scratch:
	495	__free_page(scratch);
16b374ca AA	496	out_err:
	497	dprintk("%s ERROR: returning NULL\n", __func__);
	498	return NULL;
	499	}
	500
	501	/*
ea8eecdd CH	502	* Decode the opaque device specified in 'dev' and add it to the cache of
ea8eecdd CH	503	* available devices.
16b374ca	504	*/
ea8eecdd	505	static struct nfs4_file_layout_dsaddr *
a75b9df9	506	decode_and_add_device(struct inode inode, struct pnfs_device dev, gfp_t gfp_flags)
16b374ca	507	{
ea8eecdd CH	508	struct nfs4_file_layout_dsaddr d, new;
ea8eecdd CH	509	long hash;
16b374ca	510
a75b9df9	511	new = decode_device(inode, dev, gfp_flags);
ea8eecdd	512	if (!new) {
16b374ca AA	513	printk(KERN_WARNING "%s: Could not decode or add device\n",
	514	__func__);
	515	return NULL;
	516	}
	517
ea8eecdd CH	518	spin_lock(&filelayout_deviceid_lock);
	519	d = nfs4_fl_find_get_deviceid(&new->deviceid);
	520	if (d) {
	521	spin_unlock(&filelayout_deviceid_lock);
	522	nfs4_fl_free_deviceid(new);
	523	return d;
	524	}
	525
	526	INIT_HLIST_NODE(&new->node);
	527	atomic_set(&new->ref, 1);
	528	hash = nfs4_fl_deviceid_hash(&new->deviceid);
	529	hlist_add_head_rcu(&new->node, &filelayout_deviceid_cache[hash]);
	530	spin_unlock(&filelayout_deviceid_lock);
16b374ca	531
ea8eecdd	532	return new;
16b374ca AA	533	}
	534
	535	/*
	536	* Retrieve the information for dev_id, add it to the list
	537	* of available devices, and return it.
	538	*/
	539	struct nfs4_file_layout_dsaddr *
a75b9df9	540	get_device_info(struct inode inode, struct nfs4_deviceid dev_id, gfp_t gfp_flags)
16b374ca AA	541	{
	542	struct pnfs_device *pdev = NULL;
	543	u32 max_resp_sz;
	544	int max_pages;
	545	struct page **pages = NULL;
	546	struct nfs4_file_layout_dsaddr *dsaddr = NULL;
	547	int rc, i;
	548	struct nfs_server *server = NFS_SERVER(inode);
	549
	550	/*
	551	* Use the session max response size as the basis for setting
	552	* GETDEVICEINFO's maxcount
	553	*/
	554	max_resp_sz = server->nfs_client->cl_session->fc_attrs.max_resp_sz;
	555	max_pages = max_resp_sz >> PAGE_SHIFT;
	556	dprintk("%s inode %p max_resp_sz %u max_pages %d\n",
	557	__func__, inode, max_resp_sz, max_pages);
	558
a75b9df9	559	pdev = kzalloc(sizeof(struct pnfs_device), gfp_flags);
16b374ca AA	560	if (pdev == NULL)
	561	return NULL;
	562
a75b9df9	563	pages = kzalloc(max_pages * sizeof(struct page *), gfp_flags);
16b374ca AA	564	if (pages == NULL) {
	565	kfree(pdev);
	566	return NULL;
	567	}
	568	for (i = 0; i < max_pages; i++) {
a75b9df9	569	pages[i] = alloc_page(gfp_flags);
16b374ca AA	570	if (!pages[i])
	571	goto out_free;
	572	}
	573
16b374ca AA	574	memcpy(&pdev->dev_id, dev_id, sizeof(*dev_id));
	575	pdev->layout_type = LAYOUT_NFSV4_1_FILES;
	576	pdev->pages = pages;
	577	pdev->pgbase = 0;
	578	pdev->pglen = PAGE_SIZE * max_pages;
	579	pdev->mincount = 0;
	580
	581	rc = nfs4_proc_getdeviceinfo(server, pdev);
	582	dprintk("%s getdevice info returns %d\n", __func__, rc);
	583	if (rc)
	584	goto out_free;
	585
	586	/*
	587	* Found new device, need to decode it and then add it to the
	588	* list of known devices for this mountpoint.
	589	*/
a75b9df9	590	dsaddr = decode_and_add_device(inode, pdev, gfp_flags);
16b374ca	591	out_free:
16b374ca AA	592	for (i = 0; i < max_pages; i++)
	593	__free_page(pages[i]);
	594	kfree(pages);
	595	kfree(pdev);
	596	dprintk("<-- %s dsaddr %p\n", __func__, dsaddr);
	597	return dsaddr;
	598	}
	599
ea8eecdd CH	600	void
ea8eecdd CH	601	nfs4_fl_put_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
16b374ca	602	{
ea8eecdd CH	603	if (atomic_dec_and_lock(&dsaddr->ref, &filelayout_deviceid_lock)) {
	604	hlist_del_rcu(&dsaddr->node);
	605	spin_unlock(&filelayout_deviceid_lock);
	606
	607	synchronize_rcu();
	608	nfs4_fl_free_deviceid(dsaddr);
	609	}
	610	}
16b374ca	611
ea8eecdd CH	612	struct nfs4_file_layout_dsaddr *
	613	nfs4_fl_find_get_deviceid(struct nfs4_deviceid *id)
	614	{
	615	struct nfs4_file_layout_dsaddr *d;
	616	struct hlist_node *n;
	617	long hash = nfs4_fl_deviceid_hash(id);
	618
	619
	620	rcu_read_lock();
	621	hlist_for_each_entry_rcu(d, n, &filelayout_deviceid_cache[hash], node) {
	622	if (!memcmp(&d->deviceid, id, sizeof(*id))) {
	623	if (!atomic_inc_not_zero(&d->ref))
	624	goto fail;
	625	rcu_read_unlock();
	626	return d;
	627	}
	628	}
	629	fail:
	630	rcu_read_unlock();
	631	return NULL;
16b374ca	632	}
cfe7f412 FI	633
	634	/*
	635	* Want res = (offset - layout->pattern_offset)/ layout->stripe_unit
	636	* Then: ((res + fsi) % dsaddr->stripe_count)
	637	*/
	638	u32
	639	nfs4_fl_calc_j_index(struct pnfs_layout_segment *lseg, loff_t offset)
	640	{
	641	struct nfs4_filelayout_segment *flseg = FILELAYOUT_LSEG(lseg);
	642	u64 tmp;
	643
	644	tmp = offset - flseg->pattern_offset;
	645	do_div(tmp, flseg->stripe_unit);
	646	tmp += flseg->first_stripe_index;
	647	return do_div(tmp, flseg->dsaddr->stripe_count);
	648	}
	649
	650	u32
	651	nfs4_fl_calc_ds_index(struct pnfs_layout_segment *lseg, u32 j)
	652	{
	653	return FILELAYOUT_LSEG(lseg)->dsaddr->stripe_indices[j];
	654	}
	655
	656	struct nfs_fh *
	657	nfs4_fl_select_ds_fh(struct pnfs_layout_segment *lseg, u32 j)
	658	{
	659	struct nfs4_filelayout_segment *flseg = FILELAYOUT_LSEG(lseg);
	660	u32 i;
	661
	662	if (flseg->stripe_type == STRIPE_SPARSE) {
	663	if (flseg->num_fh == 1)
	664	i = 0;
	665	else if (flseg->num_fh == 0)
	666	/* Use the MDS OPEN fh set in nfs_read_rpcsetup */
	667	return NULL;
	668	else
	669	i = nfs4_fl_calc_ds_index(lseg, j);
	670	} else
	671	i = j;
	672	return flseg->fh_array[i];
	673	}
	674
568e8c49 AA	675	static void
	676	filelayout_mark_devid_negative(struct nfs4_file_layout_dsaddr *dsaddr,
	677	int err, u32 ds_addr)
	678	{
	679	u32 p = (u32 )&dsaddr->deviceid;
	680
	681	printk(KERN_ERR "NFS: data server %x connection error %d."
	682	" Deviceid [%x%x%x%x] marked out of use.\n",
	683	ds_addr, err, p[0], p[1], p[2], p[3]);
	684
	685	spin_lock(&filelayout_deviceid_lock);
	686	dsaddr->flags \|= NFS4_DEVICE_ID_NEG_ENTRY;
	687	spin_unlock(&filelayout_deviceid_lock);
	688	}
	689
cfe7f412 FI	690	struct nfs4_pnfs_ds *
	691	nfs4_fl_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx)
	692	{
	693	struct nfs4_file_layout_dsaddr *dsaddr = FILELAYOUT_LSEG(lseg)->dsaddr;
	694	struct nfs4_pnfs_ds *ds = dsaddr->ds_list[ds_idx];
	695
	696	if (ds == NULL) {
	697	printk(KERN_ERR "%s: No data server for offset index %d\n",
	698	__func__, ds_idx);
	699	return NULL;
	700	}
	701
	702	if (!ds->ds_clp) {
568e8c49	703	struct nfs_server *s = NFS_SERVER(lseg->pls_layout->plh_inode);
cfe7f412 FI	704	int err;
cfe7f412 FI	705
568e8c49 AA	706	if (dsaddr->flags & NFS4_DEVICE_ID_NEG_ENTRY) {
	707	/* Already tried to connect, don't try again */
	708	dprintk("%s Deviceid marked out of use\n", __func__);
	709	return NULL;
	710	}
	711	err = nfs4_ds_connect(s, ds);
cfe7f412	712	if (err) {
568e8c49 AA	713	filelayout_mark_devid_negative(dsaddr, err,
568e8c49 AA	714	ntohl(ds->ds_ip_addr));
cfe7f412 FI	715	return NULL;
	716	}
	717	}
	718	return ds;
	719	}