[mirror_ubuntu-artful-kernel.git] / fs / nfs / flexfilelayout / flexfilelayoutdev.c

/*
 * Device operations for the pnfs nfs4 file layout driver.
 *
 * Copyright (c) 2014, Primary Data, Inc. All rights reserved.
 *
 * Tao Peng <bergwolf@primarydata.com>
 */

#include <linux/nfs_fs.h>
#include <linux/vmalloc.h>
#include <linux/module.h>
#include <linux/sunrpc/addr.h>

#include "../internal.h"
#include "../nfs4session.h"
#include "flexfilelayout.h"

#define NFSDBG_FACILITY		NFSDBG_PNFS_LD

static unsigned int dataserver_timeo = NFS_DEF_TCP_RETRANS;
static unsigned int dataserver_retrans;

static bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg);

void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
{
	if (!IS_ERR_OR_NULL(mirror_ds))
		nfs4_put_deviceid_node(&mirror_ds->id_node);
}

void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
{
	nfs4_print_deviceid(&mirror_ds->id_node.deviceid);
	nfs4_pnfs_ds_put(mirror_ds->ds);
	kfree_rcu(mirror_ds, id_node.rcu);
}

/* Decode opaque device data and construct new_ds using it */
struct nfs4_ff_layout_ds *
nfs4_ff_alloc_deviceid_node(struct nfs_server *server, struct pnfs_device *pdev,
			    gfp_t gfp_flags)
{
	struct xdr_stream stream;
	struct xdr_buf buf;
	struct page *scratch;
	struct list_head dsaddrs;
	struct nfs4_pnfs_ds_addr *da;
	struct nfs4_ff_layout_ds *new_ds = NULL;
	struct nfs4_ff_ds_version *ds_versions = NULL;
	u32 mp_count;
	u32 version_count;
	__be32 *p;
	int i, ret = -ENOMEM;

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

	new_ds = kzalloc(sizeof(struct nfs4_ff_layout_ds), gfp_flags);
	if (!new_ds)
		goto out_scratch;

	nfs4_init_deviceid_node(&new_ds->id_node,
				server,
				&pdev->dev_id);
	INIT_LIST_HEAD(&dsaddrs);

	xdr_init_decode_pages(&stream, &buf, pdev->pages, pdev->pglen);
	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);

	/* multipath count */
	p = xdr_inline_decode(&stream, 4);
	if (unlikely(!p))
		goto out_err_drain_dsaddrs;
	mp_count = be32_to_cpup(p);
	dprintk("%s: multipath ds count %d\n", __func__, mp_count);

	for (i = 0; i < mp_count; i++) {
		/* multipath ds */
		da = nfs4_decode_mp_ds_addr(server->nfs_client->cl_net,
					    &stream, gfp_flags);
		if (da)
			list_add_tail(&da->da_node, &dsaddrs);
	}
	if (list_empty(&dsaddrs)) {
		dprintk("%s: no suitable DS addresses found\n",
			__func__);
		ret = -ENOMEDIUM;
		goto out_err_drain_dsaddrs;
	}

	/* version count */
	p = xdr_inline_decode(&stream, 4);
	if (unlikely(!p))
		goto out_err_drain_dsaddrs;
	version_count = be32_to_cpup(p);
	dprintk("%s: version count %d\n", __func__, version_count);

	ds_versions = kzalloc(version_count * sizeof(struct nfs4_ff_ds_version),
			      gfp_flags);
	if (!ds_versions)
		goto out_scratch;

	for (i = 0; i < version_count; i++) {
		/* 20 = version(4) + minor_version(4) + rsize(4) + wsize(4) +
		 * tightly_coupled(4) */
		p = xdr_inline_decode(&stream, 20);
		if (unlikely(!p))
			goto out_err_drain_dsaddrs;
		ds_versions[i].version = be32_to_cpup(p++);
		ds_versions[i].minor_version = be32_to_cpup(p++);
		ds_versions[i].rsize = nfs_block_size(be32_to_cpup(p++), NULL);
		ds_versions[i].wsize = nfs_block_size(be32_to_cpup(p++), NULL);
		ds_versions[i].tightly_coupled = be32_to_cpup(p);

		if (ds_versions[i].rsize > NFS_MAX_FILE_IO_SIZE)
			ds_versions[i].rsize = NFS_MAX_FILE_IO_SIZE;
		if (ds_versions[i].wsize > NFS_MAX_FILE_IO_SIZE)
			ds_versions[i].wsize = NFS_MAX_FILE_IO_SIZE;

		if (ds_versions[i].version != 3 || ds_versions[i].minor_version != 0) {
			dprintk("%s: [%d] unsupported ds version %d-%d\n", __func__,
				i, ds_versions[i].version,
				ds_versions[i].minor_version);
			ret = -EPROTONOSUPPORT;
			goto out_err_drain_dsaddrs;
		}

		dprintk("%s: [%d] vers %u minor_ver %u rsize %u wsize %u coupled %d\n",
			__func__, i, ds_versions[i].version,
			ds_versions[i].minor_version,
			ds_versions[i].rsize,
			ds_versions[i].wsize,
			ds_versions[i].tightly_coupled);
	}

	new_ds->ds_versions = ds_versions;
	new_ds->ds_versions_cnt = version_count;

	new_ds->ds = nfs4_pnfs_ds_add(&dsaddrs, gfp_flags);
	if (!new_ds->ds)
		goto out_err_drain_dsaddrs;

	/* If DS was already in cache, free ds addrs */
	while (!list_empty(&dsaddrs)) {
		da = list_first_entry(&dsaddrs,
				      struct nfs4_pnfs_ds_addr,
				      da_node);
		list_del_init(&da->da_node);
		kfree(da->da_remotestr);
		kfree(da);
	}

	__free_page(scratch);
	return new_ds;

out_err_drain_dsaddrs:
	while (!list_empty(&dsaddrs)) {
		da = list_first_entry(&dsaddrs, struct nfs4_pnfs_ds_addr,
				      da_node);
		list_del_init(&da->da_node);
		kfree(da->da_remotestr);
		kfree(da);
	}

	kfree(ds_versions);
out_scratch:
	__free_page(scratch);
out_err:
	kfree(new_ds);

	dprintk("%s ERROR: returning %d\n", __func__, ret);
	return NULL;
}

static void ff_layout_mark_devid_invalid(struct pnfs_layout_segment *lseg,
		struct nfs4_deviceid_node *devid)
{
	nfs4_delete_deviceid(devid->ld, devid->nfs_client, &devid->deviceid);
	if (!ff_layout_has_available_ds(lseg))
		pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
				lseg);
}

static bool ff_layout_mirror_valid(struct pnfs_layout_segment *lseg,
				   struct nfs4_ff_layout_mirror *mirror,
				   bool create)
{
	if (mirror == NULL || IS_ERR(mirror->mirror_ds))
		goto outerr;
	if (mirror->mirror_ds == NULL) {
		if (create) {
			struct nfs4_deviceid_node *node;
			struct pnfs_layout_hdr *lh = lseg->pls_layout;
			struct nfs4_ff_layout_ds *mirror_ds = ERR_PTR(-ENODEV);

			node = nfs4_find_get_deviceid(NFS_SERVER(lh->plh_inode),
					&mirror->devid, lh->plh_lc_cred,
					GFP_KERNEL);
			if (node)
				mirror_ds = FF_LAYOUT_MIRROR_DS(node);

			/* check for race with another call to this function */
			if (cmpxchg(&mirror->mirror_ds, NULL, mirror_ds) &&
			    mirror_ds != ERR_PTR(-ENODEV))
				nfs4_put_deviceid_node(node);
		} else
			goto outerr;
	}
	if (mirror->mirror_ds->ds == NULL) {
		struct nfs4_deviceid_node *devid;
		devid = &mirror->mirror_ds->id_node;
		ff_layout_mark_devid_invalid(lseg, devid);
		return false;
	}
	return true;
outerr:
	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode, lseg);
	return false;
}

static void extend_ds_error(struct nfs4_ff_layout_ds_err *err,
			    u64 offset, u64 length)
{
	u64 end;

	end = max_t(u64, pnfs_end_offset(err->offset, err->length),
		    pnfs_end_offset(offset, length));
	err->offset = min_t(u64, err->offset, offset);
	err->length = end - err->offset;
}

static int
ff_ds_error_match(const struct nfs4_ff_layout_ds_err *e1,
		const struct nfs4_ff_layout_ds_err *e2)
{
	int ret;

	if (e1->opnum != e2->opnum)
		return e1->opnum < e2->opnum ? -1 : 1;
	if (e1->status != e2->status)
		return e1->status < e2->status ? -1 : 1;
	ret = memcmp(e1->stateid.data, e2->stateid.data,
			sizeof(e1->stateid.data));
	if (ret != 0)
		return ret;
	ret = memcmp(&e1->deviceid, &e2->deviceid, sizeof(e1->deviceid));
	if (ret != 0)
		return ret;
	if (pnfs_end_offset(e1->offset, e1->length) < e2->offset)
		return -1;
	if (e1->offset > pnfs_end_offset(e2->offset, e2->length))
		return 1;
	/* If ranges overlap or are contiguous, they are the same */
	return 0;
}

static void
ff_layout_add_ds_error_locked(struct nfs4_flexfile_layout *flo,
			      struct nfs4_ff_layout_ds_err *dserr)
{
	struct nfs4_ff_layout_ds_err *err, *tmp;
	struct list_head *head = &flo->error_list;
	int match;

	/* Do insertion sort w/ merges */
	list_for_each_entry_safe(err, tmp, &flo->error_list, list) {
		match = ff_ds_error_match(err, dserr);
		if (match < 0)
			continue;
		if (match > 0) {
			/* Add entry "dserr" _before_ entry "err" */
			head = &err->list;
			break;
		}
		/* Entries match, so merge "err" into "dserr" */
		extend_ds_error(dserr, err->offset, err->length);
		list_replace(&err->list, &dserr->list);
		kfree(err);
		return;
	}

	list_add_tail(&dserr->list, head);
}

int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
			     struct nfs4_ff_layout_mirror *mirror, u64 offset,
			     u64 length, int status, enum nfs_opnum4 opnum,
			     gfp_t gfp_flags)
{
	struct nfs4_ff_layout_ds_err *dserr;

	if (status == 0)
		return 0;

	if (mirror->mirror_ds == NULL)
		return -EINVAL;

	dserr = kmalloc(sizeof(*dserr), gfp_flags);
	if (!dserr)
		return -ENOMEM;

	INIT_LIST_HEAD(&dserr->list);
	dserr->offset = offset;
	dserr->length = length;
	dserr->status = status;
	dserr->opnum = opnum;
	nfs4_stateid_copy(&dserr->stateid, &mirror->stateid);
	memcpy(&dserr->deviceid, &mirror->mirror_ds->id_node.deviceid,
	       NFS4_DEVICEID4_SIZE);

	spin_lock(&flo->generic_hdr.plh_inode->i_lock);
	ff_layout_add_ds_error_locked(flo, dserr);
	spin_unlock(&flo->generic_hdr.plh_inode->i_lock);

	return 0;
}

static struct rpc_cred *
ff_layout_get_mirror_cred(struct nfs4_ff_layout_mirror *mirror, u32 iomode)
{
	struct rpc_cred *cred, __rcu **pcred;

	if (iomode == IOMODE_READ)
		pcred = &mirror->ro_cred;
	else
		pcred = &mirror->rw_cred;

	rcu_read_lock();
	do {
		cred = rcu_dereference(*pcred);
		if (!cred)
			break;

		cred = get_rpccred_rcu(cred);
	} while(!cred);
	rcu_read_unlock();
	return cred;
}

struct nfs_fh *
nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx)
{
	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, mirror_idx);
	struct nfs_fh *fh = NULL;

	if (!ff_layout_mirror_valid(lseg, mirror, false)) {
		pr_err_ratelimited("NFS: %s: No data server for mirror offset index %d\n",
			__func__, mirror_idx);
		goto out;
	}

	/* FIXME: For now assume there is only 1 version available for the DS */
	fh = &mirror->fh_versions[0];
out:
	return fh;
}

/**
 * nfs4_ff_layout_prepare_ds - prepare a DS connection for an RPC call
 * @lseg: the layout segment we're operating on
 * @ds_idx: index of the DS to use
 * @fail_return: return layout on connect failure?
 *
 * Try to prepare a DS connection to accept an RPC call. This involves
 * selecting a mirror to use and connecting the client to it if it's not
 * already connected.
 *
 * Since we only need a single functioning mirror to satisfy a read, we don't
 * want to return the layout if there is one. For writes though, any down
 * mirror should result in a LAYOUTRETURN. @fail_return is how we distinguish
 * between the two cases.
 *
 * Returns a pointer to a connected DS object on success or NULL on failure.
 */
struct nfs4_pnfs_ds *
nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
			  bool fail_return)
{
	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	struct nfs4_pnfs_ds *ds = NULL;
	struct nfs4_deviceid_node *devid;
	struct inode *ino = lseg->pls_layout->plh_inode;
	struct nfs_server *s = NFS_SERVER(ino);
	unsigned int max_payload;

	if (!ff_layout_mirror_valid(lseg, mirror, true)) {
		pr_err_ratelimited("NFS: %s: No data server for offset index %d\n",
			__func__, ds_idx);
		goto out;
	}

	devid = &mirror->mirror_ds->id_node;
	if (ff_layout_test_devid_unavailable(devid))
		goto out_fail;

	ds = mirror->mirror_ds->ds;
	/* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */
	smp_rmb();
	if (ds->ds_clp)
		goto out;

	/* FIXME: For now we assume the server sent only one version of NFS
	 * to use for the DS.
	 */
	nfs4_pnfs_ds_connect(s, ds, devid, dataserver_timeo,
			     dataserver_retrans,
			     mirror->mirror_ds->ds_versions[0].version,
			     mirror->mirror_ds->ds_versions[0].minor_version);

	/* connect success, check rsize/wsize limit */
	if (ds->ds_clp) {
		max_payload =
			nfs_block_size(rpc_max_payload(ds->ds_clp->cl_rpcclient),
				       NULL);
		if (mirror->mirror_ds->ds_versions[0].rsize > max_payload)
			mirror->mirror_ds->ds_versions[0].rsize = max_payload;
		if (mirror->mirror_ds->ds_versions[0].wsize > max_payload)
			mirror->mirror_ds->ds_versions[0].wsize = max_payload;
		goto out;
	}
	ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
				 mirror, lseg->pls_range.offset,
				 lseg->pls_range.length, NFS4ERR_NXIO,
				 OP_ILLEGAL, GFP_NOIO);
out_fail:
	if (fail_return || !ff_layout_has_available_ds(lseg))
		pnfs_error_mark_layout_for_return(ino, lseg);
	ds = NULL;
out:
	return ds;
}

struct rpc_cred *
ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg, u32 ds_idx,
		      struct rpc_cred *mdscred)
{
	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	struct rpc_cred *cred;

	if (mirror) {
		cred = ff_layout_get_mirror_cred(mirror, lseg->pls_range.iomode);
		if (!cred)
			cred = get_rpccred(mdscred);
	} else {
		cred = get_rpccred(mdscred);
	}
	return cred;
}

/**
* Find or create a DS rpc client with th MDS server rpc client auth flavor
* in the nfs_client cl_ds_clients list.
*/
struct rpc_clnt *
nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg, u32 ds_idx,
				 struct nfs_client *ds_clp, struct inode *inode)
{
	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);

	switch (mirror->mirror_ds->ds_versions[0].version) {
	case 3:
		/* For NFSv3 DS, flavor is set when creating DS connections */
		return ds_clp->cl_rpcclient;
	case 4:
		return nfs4_find_or_create_ds_client(ds_clp, inode);
	default:
		BUG();
	}
}

void ff_layout_free_ds_ioerr(struct list_head *head)
{
	struct nfs4_ff_layout_ds_err *err;

	while (!list_empty(head)) {
		err = list_first_entry(head,
				struct nfs4_ff_layout_ds_err,
				list);
		list_del(&err->list);
		kfree(err);
	}
}

/* called with inode i_lock held */
int ff_layout_encode_ds_ioerr(struct xdr_stream *xdr, const struct list_head *head)
{
	struct nfs4_ff_layout_ds_err *err;
	__be32 *p;

	list_for_each_entry(err, head, list) {
		/* offset(8) + length(8) + stateid(NFS4_STATEID_SIZE)
		 * + array length + deviceid(NFS4_DEVICEID4_SIZE)
		 * + status(4) + opnum(4)
		 */
		p = xdr_reserve_space(xdr,
				28 + NFS4_STATEID_SIZE + NFS4_DEVICEID4_SIZE);
		if (unlikely(!p))
			return -ENOBUFS;
		p = xdr_encode_hyper(p, err->offset);
		p = xdr_encode_hyper(p, err->length);
		p = xdr_encode_opaque_fixed(p, &err->stateid,
					    NFS4_STATEID_SIZE);
		/* Encode 1 error */
		*p++ = cpu_to_be32(1);
		p = xdr_encode_opaque_fixed(p, &err->deviceid,
					    NFS4_DEVICEID4_SIZE);
		*p++ = cpu_to_be32(err->status);
		*p++ = cpu_to_be32(err->opnum);
		dprintk("%s: offset %llu length %llu status %d op %d\n",
			__func__, err->offset, err->length, err->status,
			err->opnum);
	}

	return 0;
}

static
unsigned int do_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
				      const struct pnfs_layout_range *range,
				      struct list_head *head,
				      unsigned int maxnum)
{
	struct nfs4_flexfile_layout *flo = FF_LAYOUT_FROM_HDR(lo);
	struct inode *inode = lo->plh_inode;
	struct nfs4_ff_layout_ds_err *err, *n;
	unsigned int ret = 0;

	spin_lock(&inode->i_lock);
	list_for_each_entry_safe(err, n, &flo->error_list, list) {
		if (!pnfs_is_range_intersecting(err->offset,
				pnfs_end_offset(err->offset, err->length),
				range->offset,
				pnfs_end_offset(range->offset, range->length)))
			continue;
		if (!maxnum)
			break;
		list_move(&err->list, head);
		maxnum--;
		ret++;
	}
	spin_unlock(&inode->i_lock);
	return ret;
}

unsigned int ff_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
				      const struct pnfs_layout_range *range,
				      struct list_head *head,
				      unsigned int maxnum)
{
	unsigned int ret;

	ret = do_layout_fetch_ds_ioerr(lo, range, head, maxnum);
	/* If we're over the max, discard all remaining entries */
	if (ret == maxnum) {
		LIST_HEAD(discard);
		do_layout_fetch_ds_ioerr(lo, range, &discard, -1);
		ff_layout_free_ds_ioerr(&discard);
	}
	return ret;
}

static bool ff_read_layout_has_available_ds(struct pnfs_layout_segment *lseg)
{
	struct nfs4_ff_layout_mirror *mirror;
	struct nfs4_deviceid_node *devid;
	u32 idx;

	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
		mirror = FF_LAYOUT_COMP(lseg, idx);
		if (mirror) {
			if (!mirror->mirror_ds)
				return true;
			if (IS_ERR(mirror->mirror_ds))
				continue;
			devid = &mirror->mirror_ds->id_node;
			if (!ff_layout_test_devid_unavailable(devid))
				return true;
		}
	}

	return false;
}

static bool ff_rw_layout_has_available_ds(struct pnfs_layout_segment *lseg)
{
	struct nfs4_ff_layout_mirror *mirror;
	struct nfs4_deviceid_node *devid;
	u32 idx;

	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
		mirror = FF_LAYOUT_COMP(lseg, idx);
		if (!mirror || IS_ERR(mirror->mirror_ds))
			return false;
		if (!mirror->mirror_ds)
			continue;
		devid = &mirror->mirror_ds->id_node;
		if (ff_layout_test_devid_unavailable(devid))
			return false;
	}

	return FF_LAYOUT_MIRROR_COUNT(lseg) != 0;
}

static bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg)
{
	if (lseg->pls_range.iomode == IOMODE_READ)
		return  ff_read_layout_has_available_ds(lseg);
	/* Note: RW layout needs all mirrors available */
	return ff_rw_layout_has_available_ds(lseg);
}

bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg)
{
	return ff_layout_no_fallback_to_mds(lseg) ||
	       ff_layout_has_available_ds(lseg);
}

bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg)
{
	return lseg->pls_range.iomode == IOMODE_RW &&
	       ff_layout_no_read_on_rw(lseg);
}

module_param(dataserver_retrans, uint, 0644);
MODULE_PARM_DESC(dataserver_retrans, "The  number of times the NFSv4.1 client "
			"retries a request before it attempts further "
			" recovery  action.");
module_param(dataserver_timeo, uint, 0644);
MODULE_PARM_DESC(dataserver_timeo, "The time (in tenths of a second) the "
			"NFSv4.1  client  waits for a response from a "
			" data server before it retries an NFS request.");
Commit	Line	Data
d67ae825 TH	1	/*
	2	* Device operations for the pnfs nfs4 file layout driver.
	3	*
	4	* Copyright (c) 2014, Primary Data, Inc. All rights reserved.
	5	*
	6	* Tao Peng <bergwolf@primarydata.com>
	7	*/
	8
	9	#include <linux/nfs_fs.h>
	10	#include <linux/vmalloc.h>
	11	#include <linux/module.h>
	12	#include <linux/sunrpc/addr.h>
	13
	14	#include "../internal.h"
	15	#include "../nfs4session.h"
	16	#include "flexfilelayout.h"
	17
	18	#define NFSDBG_FACILITY NFSDBG_PNFS_LD
	19
15d03055 TM	20	static unsigned int dataserver_timeo = NFS_DEF_TCP_RETRANS;
15d03055 TM	21	static unsigned int dataserver_retrans;
d67ae825	22
65990d1a FI	23	static bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg);
65990d1a FI	24
d67ae825 TH	25	void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
d67ae825 TH	26	{
65990d1a	27	if (!IS_ERR_OR_NULL(mirror_ds))
d67ae825 TH	28	nfs4_put_deviceid_node(&mirror_ds->id_node);
	29	}
	30
	31	void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
	32	{
	33	nfs4_print_deviceid(&mirror_ds->id_node.deviceid);
	34	nfs4_pnfs_ds_put(mirror_ds->ds);
84a80f62	35	kfree_rcu(mirror_ds, id_node.rcu);
d67ae825 TH	36	}
	37
	38	/* Decode opaque device data and construct new_ds using it */
	39	struct nfs4_ff_layout_ds *
	40	nfs4_ff_alloc_deviceid_node(struct nfs_server server, struct pnfs_device pdev,
	41	gfp_t gfp_flags)
	42	{
	43	struct xdr_stream stream;
	44	struct xdr_buf buf;
	45	struct page *scratch;
	46	struct list_head dsaddrs;
	47	struct nfs4_pnfs_ds_addr *da;
	48	struct nfs4_ff_layout_ds *new_ds = NULL;
	49	struct nfs4_ff_ds_version *ds_versions = NULL;
	50	u32 mp_count;
	51	u32 version_count;
	52	__be32 *p;
	53	int i, ret = -ENOMEM;
	54
	55	/* set up xdr stream */
	56	scratch = alloc_page(gfp_flags);
	57	if (!scratch)
	58	goto out_err;
	59
	60	new_ds = kzalloc(sizeof(struct nfs4_ff_layout_ds), gfp_flags);
	61	if (!new_ds)
	62	goto out_scratch;
	63
	64	nfs4_init_deviceid_node(&new_ds->id_node,
	65	server,
	66	&pdev->dev_id);
	67	INIT_LIST_HEAD(&dsaddrs);
	68
	69	xdr_init_decode_pages(&stream, &buf, pdev->pages, pdev->pglen);
	70	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
	71
	72	/* multipath count */
	73	p = xdr_inline_decode(&stream, 4);
	74	if (unlikely(!p))
	75	goto out_err_drain_dsaddrs;
	76	mp_count = be32_to_cpup(p);
	77	dprintk("%s: multipath ds count %d\n", __func__, mp_count);
	78
	79	for (i = 0; i < mp_count; i++) {
	80	/* multipath ds */
	81	da = nfs4_decode_mp_ds_addr(server->nfs_client->cl_net,
	82	&stream, gfp_flags);
	83	if (da)
	84	list_add_tail(&da->da_node, &dsaddrs);
	85	}
	86	if (list_empty(&dsaddrs)) {
	87	dprintk("%s: no suitable DS addresses found\n",
	88	__func__);
	89	ret = -ENOMEDIUM;
	90	goto out_err_drain_dsaddrs;
	91	}
	92
	93	/* version count */
	94	p = xdr_inline_decode(&stream, 4);
	95	if (unlikely(!p))
	96	goto out_err_drain_dsaddrs;
	97	version_count = be32_to_cpup(p);
	98	dprintk("%s: version count %d\n", __func__, version_count);
	99
100	ds_versions = kzalloc(version_count * sizeof(struct nfs4_ff_ds_version),
101	gfp_flags);
102	if (!ds_versions)
103	goto out_scratch;
104
105	for (i = 0; i < version_count; i++) {
106	/* 20 = version(4) + minor_version(4) + rsize(4) + wsize(4) +
107	* tightly_coupled(4) */
108	p = xdr_inline_decode(&stream, 20);
109	if (unlikely(!p))
110	goto out_err_drain_dsaddrs;
111	ds_versions[i].version = be32_to_cpup(p++);
112	ds_versions[i].minor_version = be32_to_cpup(p++);
113	ds_versions[i].rsize = nfs_block_size(be32_to_cpup(p++), NULL);
114	ds_versions[i].wsize = nfs_block_size(be32_to_cpup(p++), NULL);
115	ds_versions[i].tightly_coupled = be32_to_cpup(p);
116
117	if (ds_versions[i].rsize > NFS_MAX_FILE_IO_SIZE)
118	ds_versions[i].rsize = NFS_MAX_FILE_IO_SIZE;
119	if (ds_versions[i].wsize > NFS_MAX_FILE_IO_SIZE)
120	ds_versions[i].wsize = NFS_MAX_FILE_IO_SIZE;
121
122	if (ds_versions[i].version != 3 \|\| ds_versions[i].minor_version != 0) {
123	dprintk("%s: [%d] unsupported ds version %d-%d\n", __func__,
124	i, ds_versions[i].version,
125	ds_versions[i].minor_version);
126	ret = -EPROTONOSUPPORT;
127	goto out_err_drain_dsaddrs;
128	}
129
130	dprintk("%s: [%d] vers %u minor_ver %u rsize %u wsize %u coupled %d\n",
131	__func__, i, ds_versions[i].version,
132	ds_versions[i].minor_version,
133	ds_versions[i].rsize,
134	ds_versions[i].wsize,
135	ds_versions[i].tightly_coupled);
136	}
137
138	new_ds->ds_versions = ds_versions;
139	new_ds->ds_versions_cnt = version_count;
140
141	new_ds->ds = nfs4_pnfs_ds_add(&dsaddrs, gfp_flags);
142	if (!new_ds->ds)
143	goto out_err_drain_dsaddrs;
144
145	/* If DS was already in cache, free ds addrs */
146	while (!list_empty(&dsaddrs)) {
147	da = list_first_entry(&dsaddrs,
148	struct nfs4_pnfs_ds_addr,
149	da_node);
150	list_del_init(&da->da_node);
151	kfree(da->da_remotestr);
152	kfree(da);
153	}
154
155	__free_page(scratch);
156	return new_ds;
157
158	out_err_drain_dsaddrs:
159	while (!list_empty(&dsaddrs)) {
160	da = list_first_entry(&dsaddrs, struct nfs4_pnfs_ds_addr,
161	da_node);
162	list_del_init(&da->da_node);
163	kfree(da->da_remotestr);
164	kfree(da);
165	}
166
167	kfree(ds_versions);
168	out_scratch:
169	__free_page(scratch);
170	out_err:
171	kfree(new_ds);
172
173	dprintk("%s ERROR: returning %d\n", __func__, ret);
174	return NULL;
175	}
176
889d94d4 TM	177	static void ff_layout_mark_devid_invalid(struct pnfs_layout_segment *lseg,
	178	struct nfs4_deviceid_node *devid)
	179	{
1c48cee8	180	nfs4_delete_deviceid(devid->ld, devid->nfs_client, &devid->deviceid);
889d94d4 TM	181	if (!ff_layout_has_available_ds(lseg))
	182	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
	183	lseg);
	184	}
	185
	186	static bool ff_layout_mirror_valid(struct pnfs_layout_segment *lseg,
65990d1a FI	187	struct nfs4_ff_layout_mirror *mirror,
65990d1a FI	188	bool create)
889d94d4	189	{
65990d1a FI	190	if (mirror == NULL \|\| IS_ERR(mirror->mirror_ds))
	191	goto outerr;
	192	if (mirror->mirror_ds == NULL) {
	193	if (create) {
	194	struct nfs4_deviceid_node *node;
	195	struct pnfs_layout_hdr *lh = lseg->pls_layout;
	196	struct nfs4_ff_layout_ds *mirror_ds = ERR_PTR(-ENODEV);
	197
	198	node = nfs4_find_get_deviceid(NFS_SERVER(lh->plh_inode),
	199	&mirror->devid, lh->plh_lc_cred,
	200	GFP_KERNEL);
	201	if (node)
	202	mirror_ds = FF_LAYOUT_MIRROR_DS(node);
	203
	204	/* check for race with another call to this function */
	205	if (cmpxchg(&mirror->mirror_ds, NULL, mirror_ds) &&
	206	mirror_ds != ERR_PTR(-ENODEV))
	207	nfs4_put_deviceid_node(node);
	208	} else
	209	goto outerr;
889d94d4 TM	210	}
	211	if (mirror->mirror_ds->ds == NULL) {
	212	struct nfs4_deviceid_node *devid;
	213	devid = &mirror->mirror_ds->id_node;
	214	ff_layout_mark_devid_invalid(lseg, devid);
	215	return false;
	216	}
	217	return true;
65990d1a FI	218	outerr:
	219	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode, lseg);
	220	return false;
889d94d4 TM	221	}
889d94d4 TM	222
d67ae825 TH	223	static void extend_ds_error(struct nfs4_ff_layout_ds_err *err,
	224	u64 offset, u64 length)
	225	{
	226	u64 end;
	227
17822b20 TM	228	end = max_t(u64, pnfs_end_offset(err->offset, err->length),
17822b20 TM	229	pnfs_end_offset(offset, length));
d67ae825 TH	230	err->offset = min_t(u64, err->offset, offset);
	231	err->length = end - err->offset;
	232	}
	233
b819ed4b TM	234	static int
	235	ff_ds_error_match(const struct nfs4_ff_layout_ds_err *e1,
	236	const struct nfs4_ff_layout_ds_err *e2)
d67ae825	237	{
b819ed4b TM	238	int ret;
	239
	240	if (e1->opnum != e2->opnum)
	241	return e1->opnum < e2->opnum ? -1 : 1;
	242	if (e1->status != e2->status)
	243	return e1->status < e2->status ? -1 : 1;
93b717fd TM	244	ret = memcmp(e1->stateid.data, e2->stateid.data,
93b717fd TM	245	sizeof(e1->stateid.data));
b819ed4b TM	246	if (ret != 0)
	247	return ret;
	248	ret = memcmp(&e1->deviceid, &e2->deviceid, sizeof(e1->deviceid));
	249	if (ret != 0)
	250	return ret;
17822b20	251	if (pnfs_end_offset(e1->offset, e1->length) < e2->offset)
b819ed4b	252	return -1;
17822b20	253	if (e1->offset > pnfs_end_offset(e2->offset, e2->length))
b819ed4b TM	254	return 1;
	255	/* If ranges overlap or are contiguous, they are the same */
	256	return 0;
d67ae825 TH	257	}
d67ae825 TH	258
b819ed4b	259	static void
d67ae825 TH	260	ff_layout_add_ds_error_locked(struct nfs4_flexfile_layout *flo,
	261	struct nfs4_ff_layout_ds_err *dserr)
	262	{
b819ed4b TM	263	struct nfs4_ff_layout_ds_err err, tmp;
	264	struct list_head *head = &flo->error_list;
	265	int match;
	266
	267	/* Do insertion sort w/ merges */
	268	list_for_each_entry_safe(err, tmp, &flo->error_list, list) {
	269	match = ff_ds_error_match(err, dserr);
	270	if (match < 0)
	271	continue;
	272	if (match > 0) {
	273	/* Add entry "dserr" _before_ entry "err" */
	274	head = &err->list;
d67ae825 TH	275	break;
d67ae825 TH	276	}
b819ed4b TM	277	/* Entries match, so merge "err" into "dserr" */
b819ed4b TM	278	extend_ds_error(dserr, err->offset, err->length);
cb067935	279	list_replace(&err->list, &dserr->list);
b819ed4b	280	kfree(err);
cb067935	281	return;
d67ae825 TH	282	}
d67ae825 TH	283
b819ed4b	284	list_add_tail(&dserr->list, head);
d67ae825 TH	285	}
	286
	287	int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
	288	struct nfs4_ff_layout_mirror *mirror, u64 offset,
	289	u64 length, int status, enum nfs_opnum4 opnum,
	290	gfp_t gfp_flags)
	291	{
	292	struct nfs4_ff_layout_ds_err *dserr;
d67ae825 TH	293
	294	if (status == 0)
	295	return 0;
	296
	297	if (mirror->mirror_ds == NULL)
	298	return -EINVAL;
	299
d67ae825 TH	300	dserr = kmalloc(sizeof(*dserr), gfp_flags);
	301	if (!dserr)
	302	return -ENOMEM;
	303
	304	INIT_LIST_HEAD(&dserr->list);
	305	dserr->offset = offset;
	306	dserr->length = length;
	307	dserr->status = status;
	308	dserr->opnum = opnum;
	309	nfs4_stateid_copy(&dserr->stateid, &mirror->stateid);
	310	memcpy(&dserr->deviceid, &mirror->mirror_ds->id_node.deviceid,
	311	NFS4_DEVICEID4_SIZE);
	312
	313	spin_lock(&flo->generic_hdr.plh_inode->i_lock);
b819ed4b	314	ff_layout_add_ds_error_locked(flo, dserr);
d67ae825	315	spin_unlock(&flo->generic_hdr.plh_inode->i_lock);
d67ae825 TH	316
	317	return 0;
	318	}
	319
57f3f4c0 JL	320	static struct rpc_cred *
	321	ff_layout_get_mirror_cred(struct nfs4_ff_layout_mirror *mirror, u32 iomode)
	322	{
3064b686	323	struct rpc_cred cred, __rcu *pcred;
57f3f4c0	324
3064b686 JL	325	if (iomode == IOMODE_READ)
	326	pcred = &mirror->ro_cred;
	327	else
	328	pcred = &mirror->rw_cred;
57f3f4c0 JL	329
	330	rcu_read_lock();
	331	do {
	332	cred = rcu_dereference(*pcred);
	333	if (!cred)
	334	break;
	335
	336	cred = get_rpccred_rcu(cred);
	337	} while(!cred);
	338	rcu_read_unlock();
	339	return cred;
	340	}
	341
d67ae825 TH	342	struct nfs_fh *
	343	nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx)
	344	{
	345	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, mirror_idx);
	346	struct nfs_fh *fh = NULL;
d67ae825	347
65990d1a	348	if (!ff_layout_mirror_valid(lseg, mirror, false)) {
889d94d4	349	pr_err_ratelimited("NFS: %s: No data server for mirror offset index %d\n",
d67ae825	350	__func__, mirror_idx);
d67ae825 TH	351	goto out;
	352	}
	353
	354	/* FIXME: For now assume there is only 1 version available for the DS */
	355	fh = &mirror->fh_versions[0];
	356	out:
	357	return fh;
	358	}
	359
95e2b7e9 JL	360	/**
	361	* nfs4_ff_layout_prepare_ds - prepare a DS connection for an RPC call
	362	* @lseg: the layout segment we're operating on
	363	* @ds_idx: index of the DS to use
	364	* @fail_return: return layout on connect failure?
	365	*
	366	* Try to prepare a DS connection to accept an RPC call. This involves
	367	* selecting a mirror to use and connecting the client to it if it's not
	368	* already connected.
	369	*
	370	* Since we only need a single functioning mirror to satisfy a read, we don't
	371	* want to return the layout if there is one. For writes though, any down
	372	* mirror should result in a LAYOUTRETURN. @fail_return is how we distinguish
	373	* between the two cases.
	374	*
	375	* Returns a pointer to a connected DS object on success or NULL on failure.
	376	*/
d67ae825 TH	377	struct nfs4_pnfs_ds *
	378	nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
	379	bool fail_return)
	380	{
	381	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	382	struct nfs4_pnfs_ds *ds = NULL;
	383	struct nfs4_deviceid_node *devid;
	384	struct inode *ino = lseg->pls_layout->plh_inode;
	385	struct nfs_server *s = NFS_SERVER(ino);
	386	unsigned int max_payload;
d67ae825	387
65990d1a	388	if (!ff_layout_mirror_valid(lseg, mirror, true)) {
889d94d4	389	pr_err_ratelimited("NFS: %s: No data server for offset index %d\n",
d67ae825	390	__func__, ds_idx);
d67ae825 TH	391	goto out;
	392	}
	393
	394	devid = &mirror->mirror_ds->id_node;
	395	if (ff_layout_test_devid_unavailable(devid))
3dc14735	396	goto out_fail;
d67ae825 TH	397
	398	ds = mirror->mirror_ds->ds;
	399	/* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */
	400	smp_rmb();
	401	if (ds->ds_clp)
90a0be00	402	goto out;
d67ae825 TH	403
	404	/* FIXME: For now we assume the server sent only one version of NFS
	405	* to use for the DS.
	406	*/
	407	nfs4_pnfs_ds_connect(s, ds, devid, dataserver_timeo,
	408	dataserver_retrans,
	409	mirror->mirror_ds->ds_versions[0].version,
7d38de3f	410	mirror->mirror_ds->ds_versions[0].minor_version);
d67ae825 TH	411
	412	/* connect success, check rsize/wsize limit */
	413	if (ds->ds_clp) {
	414	max_payload =
	415	nfs_block_size(rpc_max_payload(ds->ds_clp->cl_rpcclient),
	416	NULL);
	417	if (mirror->mirror_ds->ds_versions[0].rsize > max_payload)
	418	mirror->mirror_ds->ds_versions[0].rsize = max_payload;
	419	if (mirror->mirror_ds->ds_versions[0].wsize > max_payload)
	420	mirror->mirror_ds->ds_versions[0].wsize = max_payload;
3dc14735	421	goto out;
d67ae825	422	}
3dc14735 TM	423	ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
	424	mirror, lseg->pls_range.offset,
	425	lseg->pls_range.length, NFS4ERR_NXIO,
	426	OP_ILLEGAL, GFP_NOIO);
	427	out_fail:
	428	if (fail_return \|\| !ff_layout_has_available_ds(lseg))
	429	pnfs_error_mark_layout_for_return(ino, lseg);
	430	ds = NULL;
d67ae825 TH	431	out:
	432	return ds;
	433	}
	434
	435	struct rpc_cred *
	436	ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg, u32 ds_idx,
	437	struct rpc_cred *mdscred)
	438	{
	439	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
547a6376	440	struct rpc_cred *cred;
d67ae825	441
57f3f4c0 JL	442	if (mirror) {
	443	cred = ff_layout_get_mirror_cred(mirror, lseg->pls_range.iomode);
	444	if (!cred)
	445	cred = get_rpccred(mdscred);
	446	} else {
	447	cred = get_rpccred(mdscred);
	448	}
d67ae825 TH	449	return cred;
	450	}
	451
	452	/**
	453	* Find or create a DS rpc client with th MDS server rpc client auth flavor
	454	* in the nfs_client cl_ds_clients list.
	455	*/
	456	struct rpc_clnt *
	457	nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg, u32 ds_idx,
	458	struct nfs_client ds_clp, struct inode inode)
	459	{
	460	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	461
	462	switch (mirror->mirror_ds->ds_versions[0].version) {
	463	case 3:
	464	/* For NFSv3 DS, flavor is set when creating DS connections */
	465	return ds_clp->cl_rpcclient;
	466	case 4:
	467	return nfs4_find_or_create_ds_client(ds_clp, inode);
	468	default:
	469	BUG();
	470	}
	471	}
	472
5b9b3c85 TM	473	void ff_layout_free_ds_ioerr(struct list_head *head)
	474	{
	475	struct nfs4_ff_layout_ds_err *err;
	476
	477	while (!list_empty(head)) {
	478	err = list_first_entry(head,
	479	struct nfs4_ff_layout_ds_err,
	480	list);
	481	list_del(&err->list);
	482	kfree(err);
	483	}
	484	}
	485
d67ae825	486	/* called with inode i_lock held */
5b9b3c85	487	int ff_layout_encode_ds_ioerr(struct xdr_stream xdr, const struct list_head head)
d67ae825	488	{
5b9b3c85	489	struct nfs4_ff_layout_ds_err *err;
d67ae825 TH	490	__be32 *p;
d67ae825 TH	491
5b9b3c85	492	list_for_each_entry(err, head, list) {
d67ae825	493	/* offset(8) + length(8) + stateid(NFS4_STATEID_SIZE)
d1354907 TM	494	* + array length + deviceid(NFS4_DEVICEID4_SIZE)
d1354907 TM	495	* + status(4) + opnum(4)
d67ae825 TH	496	*/
d67ae825 TH	497	p = xdr_reserve_space(xdr,
d1354907	498	28 + NFS4_STATEID_SIZE + NFS4_DEVICEID4_SIZE);
d67ae825 TH	499	if (unlikely(!p))
	500	return -ENOBUFS;
	501	p = xdr_encode_hyper(p, err->offset);
	502	p = xdr_encode_hyper(p, err->length);
	503	p = xdr_encode_opaque_fixed(p, &err->stateid,
	504	NFS4_STATEID_SIZE);
d1354907 TM	505	/* Encode 1 error */
d1354907 TM	506	*p++ = cpu_to_be32(1);
d67ae825 TH	507	p = xdr_encode_opaque_fixed(p, &err->deviceid,
	508	NFS4_DEVICEID4_SIZE);
	509	*p++ = cpu_to_be32(err->status);
	510	*p++ = cpu_to_be32(err->opnum);
5b9b3c85	511	dprintk("%s: offset %llu length %llu status %d op %d\n",
d67ae825	512	__func__, err->offset, err->length, err->status,
5b9b3c85	513	err->opnum);
d67ae825 TH	514	}
	515
	516	return 0;
	517	}
	518
5b9b3c85 TM	519	static
	520	unsigned int do_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
	521	const struct pnfs_layout_range *range,
	522	struct list_head *head,
	523	unsigned int maxnum)
	524	{
	525	struct nfs4_flexfile_layout *flo = FF_LAYOUT_FROM_HDR(lo);
	526	struct inode *inode = lo->plh_inode;
	527	struct nfs4_ff_layout_ds_err err, n;
	528	unsigned int ret = 0;
	529
	530	spin_lock(&inode->i_lock);
	531	list_for_each_entry_safe(err, n, &flo->error_list, list) {
	532	if (!pnfs_is_range_intersecting(err->offset,
	533	pnfs_end_offset(err->offset, err->length),
	534	range->offset,
	535	pnfs_end_offset(range->offset, range->length)))
	536	continue;
	537	if (!maxnum)
	538	break;
	539	list_move(&err->list, head);
	540	maxnum--;
	541	ret++;
	542	}
	543	spin_unlock(&inode->i_lock);
	544	return ret;
	545	}
	546
	547	unsigned int ff_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
	548	const struct pnfs_layout_range *range,
	549	struct list_head *head,
	550	unsigned int maxnum)
	551	{
	552	unsigned int ret;
	553
	554	ret = do_layout_fetch_ds_ioerr(lo, range, head, maxnum);
	555	/* If we're over the max, discard all remaining entries */
	556	if (ret == maxnum) {
	557	LIST_HEAD(discard);
	558	do_layout_fetch_ds_ioerr(lo, range, &discard, -1);
	559	ff_layout_free_ds_ioerr(&discard);
	560	}
	561	return ret;
	562	}
	563
81d6dc8b	564	static bool ff_read_layout_has_available_ds(struct pnfs_layout_segment *lseg)
d67ae825 TH	565	{
	566	struct nfs4_ff_layout_mirror *mirror;
	567	struct nfs4_deviceid_node *devid;
81d6dc8b	568	u32 idx;
d67ae825 TH	569
	570	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
	571	mirror = FF_LAYOUT_COMP(lseg, idx);
65990d1a FI	572	if (mirror) {
	573	if (!mirror->mirror_ds)
	574	return true;
	575	if (IS_ERR(mirror->mirror_ds))
	576	continue;
d67ae825 TH	577	devid = &mirror->mirror_ds->id_node;
	578	if (!ff_layout_test_devid_unavailable(devid))
	579	return true;
	580	}
	581	}
	582
	583	return false;
	584	}
	585
81d6dc8b TM	586	static bool ff_rw_layout_has_available_ds(struct pnfs_layout_segment *lseg)
	587	{
	588	struct nfs4_ff_layout_mirror *mirror;
	589	struct nfs4_deviceid_node *devid;
	590	u32 idx;
	591
	592	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
	593	mirror = FF_LAYOUT_COMP(lseg, idx);
65990d1a	594	if (!mirror \|\| IS_ERR(mirror->mirror_ds))
81d6dc8b	595	return false;
65990d1a FI	596	if (!mirror->mirror_ds)
65990d1a FI	597	continue;
81d6dc8b TM	598	devid = &mirror->mirror_ds->id_node;
	599	if (ff_layout_test_devid_unavailable(devid))
	600	return false;
	601	}
	602
	603	return FF_LAYOUT_MIRROR_COUNT(lseg) != 0;
	604	}
	605
65990d1a	606	static bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg)
81d6dc8b TM	607	{
	608	if (lseg->pls_range.iomode == IOMODE_READ)
	609	return ff_read_layout_has_available_ds(lseg);
	610	/* Note: RW layout needs all mirrors available */
	611	return ff_rw_layout_has_available_ds(lseg);
	612	}
	613
3b13b4b3 TH	614	bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg)
	615	{
	616	return ff_layout_no_fallback_to_mds(lseg) \|\|
	617	ff_layout_has_available_ds(lseg);
	618	}
	619
fb1084e3 TH	620	bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg)
	621	{
	622	return lseg->pls_range.iomode == IOMODE_RW &&
	623	ff_layout_no_read_on_rw(lseg);
	624	}
	625
d67ae825 TH	626	module_param(dataserver_retrans, uint, 0644);
	627	MODULE_PARM_DESC(dataserver_retrans, "The number of times the NFSv4.1 client "
	628	"retries a request before it attempts further "
	629	" recovery action.");
	630	module_param(dataserver_timeo, uint, 0644);
	631	MODULE_PARM_DESC(dataserver_timeo, "The time (in tenths of a second) the "
	632	"NFSv4.1 client waits for a response from a "
	633	" data server before it retries an NFS request.");