[linux-2.6-block.git] / fs / nfs / flexfilelayout / flexfilelayoutdev.c

// SPDX-License-Identifier: GPL-2.0
/*
 * 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(mirror_ds->ds_versions);
	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 = kcalloc(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;

		/*
		 * check for valid major/minor combination.
		 * currently we support dataserver which talk:
		 *   v3, v4.0, v4.1, v4.2
		 */
		if (!((ds_versions[i].version == 3 && ds_versions[i].minor_version == 0) ||
			(ds_versions[i].version == 4 && ds_versions[i].minor_version < 3))) {
			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 (IS_ERR(mirror->mirror_ds))
		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;
	int status;

	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.
	 */
	status = 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 (!status) {
		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;
	}
out_fail:
	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);
	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 && !mirror->mirror_ds->ds_versions[0].tightly_coupled) {
		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
b2441318	1	// SPDX-License-Identifier: GPL-2.0
d67ae825 TH	2	/*
	3	* Device operations for the pnfs nfs4 file layout driver.
	4	*
	5	* Copyright (c) 2014, Primary Data, Inc. All rights reserved.
	6	*
	7	* Tao Peng <bergwolf@primarydata.com>
	8	*/
	9
	10	#include <linux/nfs_fs.h>
	11	#include <linux/vmalloc.h>
	12	#include <linux/module.h>
	13	#include <linux/sunrpc/addr.h>
	14
	15	#include "../internal.h"
	16	#include "../nfs4session.h"
	17	#include "flexfilelayout.h"
	18
	19	#define NFSDBG_FACILITY NFSDBG_PNFS_LD
	20
15d03055 TM	21	static unsigned int dataserver_timeo = NFS_DEF_TCP_RETRANS;
15d03055 TM	22	static unsigned int dataserver_retrans;
d67ae825	23
65990d1a FI	24	static bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg);
65990d1a FI	25
d67ae825 TH	26	void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
d67ae825 TH	27	{
65990d1a	28	if (!IS_ERR_OR_NULL(mirror_ds))
d67ae825 TH	29	nfs4_put_deviceid_node(&mirror_ds->id_node);
	30	}
	31
	32	void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
	33	{
	34	nfs4_print_deviceid(&mirror_ds->id_node.deviceid);
	35	nfs4_pnfs_ds_put(mirror_ds->ds);
1feb2616	36	kfree(mirror_ds->ds_versions);
84a80f62	37	kfree_rcu(mirror_ds, id_node.rcu);
d67ae825 TH	38	}
	39
	40	/* Decode opaque device data and construct new_ds using it */
	41	struct nfs4_ff_layout_ds *
	42	nfs4_ff_alloc_deviceid_node(struct nfs_server server, struct pnfs_device pdev,
	43	gfp_t gfp_flags)
	44	{
	45	struct xdr_stream stream;
	46	struct xdr_buf buf;
	47	struct page *scratch;
	48	struct list_head dsaddrs;
	49	struct nfs4_pnfs_ds_addr *da;
	50	struct nfs4_ff_layout_ds *new_ds = NULL;
	51	struct nfs4_ff_ds_version *ds_versions = NULL;
	52	u32 mp_count;
	53	u32 version_count;
	54	__be32 *p;
	55	int i, ret = -ENOMEM;
	56
	57	/* set up xdr stream */
	58	scratch = alloc_page(gfp_flags);
	59	if (!scratch)
	60	goto out_err;
	61
	62	new_ds = kzalloc(sizeof(struct nfs4_ff_layout_ds), gfp_flags);
	63	if (!new_ds)
	64	goto out_scratch;
	65
	66	nfs4_init_deviceid_node(&new_ds->id_node,
	67	server,
	68	&pdev->dev_id);
	69	INIT_LIST_HEAD(&dsaddrs);
	70
	71	xdr_init_decode_pages(&stream, &buf, pdev->pages, pdev->pglen);
	72	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
	73
	74	/* multipath count */
	75	p = xdr_inline_decode(&stream, 4);
	76	if (unlikely(!p))
	77	goto out_err_drain_dsaddrs;
	78	mp_count = be32_to_cpup(p);
	79	dprintk("%s: multipath ds count %d\n", __func__, mp_count);
	80
	81	for (i = 0; i < mp_count; i++) {
	82	/* multipath ds */
	83	da = nfs4_decode_mp_ds_addr(server->nfs_client->cl_net,
	84	&stream, gfp_flags);
	85	if (da)
	86	list_add_tail(&da->da_node, &dsaddrs);
	87	}
	88	if (list_empty(&dsaddrs)) {
	89	dprintk("%s: no suitable DS addresses found\n",
	90	__func__);
	91	ret = -ENOMEDIUM;
	92	goto out_err_drain_dsaddrs;
	93	}
	94
	95	/* version count */
	96	p = xdr_inline_decode(&stream, 4);
	97	if (unlikely(!p))
	98	goto out_err_drain_dsaddrs;
	99	version_count = be32_to_cpup(p);
	100	dprintk("%s: version count %d\n", __func__, version_count);
	101
6396bb22 KC	102	ds_versions = kcalloc(version_count,
6396bb22 KC	103	sizeof(struct nfs4_ff_ds_version),
d67ae825 TH	104	gfp_flags);
	105	if (!ds_versions)
	106	goto out_scratch;
	107
	108	for (i = 0; i < version_count; i++) {
	109	/* 20 = version(4) + minor_version(4) + rsize(4) + wsize(4) +
	110	* tightly_coupled(4) */
	111	p = xdr_inline_decode(&stream, 20);
	112	if (unlikely(!p))
	113	goto out_err_drain_dsaddrs;
	114	ds_versions[i].version = be32_to_cpup(p++);
	115	ds_versions[i].minor_version = be32_to_cpup(p++);
	116	ds_versions[i].rsize = nfs_block_size(be32_to_cpup(p++), NULL);
	117	ds_versions[i].wsize = nfs_block_size(be32_to_cpup(p++), NULL);
	118	ds_versions[i].tightly_coupled = be32_to_cpup(p);
	119
	120	if (ds_versions[i].rsize > NFS_MAX_FILE_IO_SIZE)
	121	ds_versions[i].rsize = NFS_MAX_FILE_IO_SIZE;
	122	if (ds_versions[i].wsize > NFS_MAX_FILE_IO_SIZE)
	123	ds_versions[i].wsize = NFS_MAX_FILE_IO_SIZE;
	124
a7878ca1 TM	125	/*
	126	* check for valid major/minor combination.
	127	* currently we support dataserver which talk:
	128	* v3, v4.0, v4.1, v4.2
	129	*/
	130	if (!((ds_versions[i].version == 3 && ds_versions[i].minor_version == 0) \|\|
	131	(ds_versions[i].version == 4 && ds_versions[i].minor_version < 3))) {
d67ae825 TH	132	dprintk("%s: [%d] unsupported ds version %d-%d\n", __func__,
	133	i, ds_versions[i].version,
	134	ds_versions[i].minor_version);
	135	ret = -EPROTONOSUPPORT;
	136	goto out_err_drain_dsaddrs;
	137	}
	138
	139	dprintk("%s: [%d] vers %u minor_ver %u rsize %u wsize %u coupled %d\n",
	140	__func__, i, ds_versions[i].version,
	141	ds_versions[i].minor_version,
	142	ds_versions[i].rsize,
	143	ds_versions[i].wsize,
	144	ds_versions[i].tightly_coupled);
	145	}
	146
	147	new_ds->ds_versions = ds_versions;
	148	new_ds->ds_versions_cnt = version_count;
	149
	150	new_ds->ds = nfs4_pnfs_ds_add(&dsaddrs, gfp_flags);
	151	if (!new_ds->ds)
	152	goto out_err_drain_dsaddrs;
	153
	154	/* If DS was already in cache, free ds addrs */
	155	while (!list_empty(&dsaddrs)) {
	156	da = list_first_entry(&dsaddrs,
	157	struct nfs4_pnfs_ds_addr,
	158	da_node);
	159	list_del_init(&da->da_node);
	160	kfree(da->da_remotestr);
	161	kfree(da);
	162	}
	163
	164	__free_page(scratch);
	165	return new_ds;
	166
	167	out_err_drain_dsaddrs:
	168	while (!list_empty(&dsaddrs)) {
	169	da = list_first_entry(&dsaddrs, struct nfs4_pnfs_ds_addr,
	170	da_node);
	171	list_del_init(&da->da_node);
	172	kfree(da->da_remotestr);
	173	kfree(da);
	174	}
	175
	176	kfree(ds_versions);
	177	out_scratch:
	178	__free_page(scratch);
	179	out_err:
	180	kfree(new_ds);
	181
	182	dprintk("%s ERROR: returning %d\n", __func__, ret);
	183	return NULL;
	184	}
	185
889d94d4 TM	186	static void ff_layout_mark_devid_invalid(struct pnfs_layout_segment *lseg,
	187	struct nfs4_deviceid_node *devid)
	188	{
1c48cee8	189	nfs4_delete_deviceid(devid->ld, devid->nfs_client, &devid->deviceid);
889d94d4 TM	190	if (!ff_layout_has_available_ds(lseg))
	191	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
	192	lseg);
	193	}
	194
	195	static bool ff_layout_mirror_valid(struct pnfs_layout_segment *lseg,
65990d1a FI	196	struct nfs4_ff_layout_mirror *mirror,
65990d1a FI	197	bool create)
889d94d4	198	{
65990d1a FI	199	if (mirror == NULL \|\| IS_ERR(mirror->mirror_ds))
	200	goto outerr;
	201	if (mirror->mirror_ds == NULL) {
	202	if (create) {
	203	struct nfs4_deviceid_node *node;
	204	struct pnfs_layout_hdr *lh = lseg->pls_layout;
	205	struct nfs4_ff_layout_ds *mirror_ds = ERR_PTR(-ENODEV);
	206
	207	node = nfs4_find_get_deviceid(NFS_SERVER(lh->plh_inode),
	208	&mirror->devid, lh->plh_lc_cred,
	209	GFP_KERNEL);
	210	if (node)
	211	mirror_ds = FF_LAYOUT_MIRROR_DS(node);
	212
	213	/* check for race with another call to this function */
	214	if (cmpxchg(&mirror->mirror_ds, NULL, mirror_ds) &&
	215	mirror_ds != ERR_PTR(-ENODEV))
	216	nfs4_put_deviceid_node(node);
	217	} else
	218	goto outerr;
889d94d4	219	}
f17f8a14 TM	220
	221	if (IS_ERR(mirror->mirror_ds))
	222	goto outerr;
	223
889d94d4 TM	224	if (mirror->mirror_ds->ds == NULL) {
	225	struct nfs4_deviceid_node *devid;
	226	devid = &mirror->mirror_ds->id_node;
	227	ff_layout_mark_devid_invalid(lseg, devid);
	228	return false;
	229	}
	230	return true;
65990d1a FI	231	outerr:
	232	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode, lseg);
	233	return false;
889d94d4 TM	234	}
889d94d4 TM	235
d67ae825 TH	236	static void extend_ds_error(struct nfs4_ff_layout_ds_err *err,
	237	u64 offset, u64 length)
	238	{
	239	u64 end;
	240
17822b20 TM	241	end = max_t(u64, pnfs_end_offset(err->offset, err->length),
17822b20 TM	242	pnfs_end_offset(offset, length));
d67ae825 TH	243	err->offset = min_t(u64, err->offset, offset);
	244	err->length = end - err->offset;
	245	}
	246
b819ed4b TM	247	static int
	248	ff_ds_error_match(const struct nfs4_ff_layout_ds_err *e1,
	249	const struct nfs4_ff_layout_ds_err *e2)
d67ae825	250	{
b819ed4b TM	251	int ret;
	252
	253	if (e1->opnum != e2->opnum)
	254	return e1->opnum < e2->opnum ? -1 : 1;
	255	if (e1->status != e2->status)
	256	return e1->status < e2->status ? -1 : 1;
93b717fd TM	257	ret = memcmp(e1->stateid.data, e2->stateid.data,
93b717fd TM	258	sizeof(e1->stateid.data));
b819ed4b TM	259	if (ret != 0)
	260	return ret;
	261	ret = memcmp(&e1->deviceid, &e2->deviceid, sizeof(e1->deviceid));
	262	if (ret != 0)
	263	return ret;
17822b20	264	if (pnfs_end_offset(e1->offset, e1->length) < e2->offset)
b819ed4b	265	return -1;
17822b20	266	if (e1->offset > pnfs_end_offset(e2->offset, e2->length))
b819ed4b TM	267	return 1;
	268	/* If ranges overlap or are contiguous, they are the same */
	269	return 0;
d67ae825 TH	270	}
d67ae825 TH	271
b819ed4b	272	static void
d67ae825 TH	273	ff_layout_add_ds_error_locked(struct nfs4_flexfile_layout *flo,
	274	struct nfs4_ff_layout_ds_err *dserr)
	275	{
b819ed4b TM	276	struct nfs4_ff_layout_ds_err err, tmp;
	277	struct list_head *head = &flo->error_list;
	278	int match;
	279
	280	/* Do insertion sort w/ merges */
	281	list_for_each_entry_safe(err, tmp, &flo->error_list, list) {
	282	match = ff_ds_error_match(err, dserr);
	283	if (match < 0)
	284	continue;
	285	if (match > 0) {
	286	/* Add entry "dserr" _before_ entry "err" */
	287	head = &err->list;
d67ae825 TH	288	break;
d67ae825 TH	289	}
b819ed4b TM	290	/* Entries match, so merge "err" into "dserr" */
b819ed4b TM	291	extend_ds_error(dserr, err->offset, err->length);
cb067935	292	list_replace(&err->list, &dserr->list);
b819ed4b	293	kfree(err);
cb067935	294	return;
d67ae825 TH	295	}
d67ae825 TH	296
b819ed4b	297	list_add_tail(&dserr->list, head);
d67ae825 TH	298	}
	299
	300	int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
	301	struct nfs4_ff_layout_mirror *mirror, u64 offset,
	302	u64 length, int status, enum nfs_opnum4 opnum,
	303	gfp_t gfp_flags)
	304	{
	305	struct nfs4_ff_layout_ds_err *dserr;
d67ae825 TH	306
	307	if (status == 0)
	308	return 0;
	309
	310	if (mirror->mirror_ds == NULL)
	311	return -EINVAL;
	312
d67ae825 TH	313	dserr = kmalloc(sizeof(*dserr), gfp_flags);
	314	if (!dserr)
	315	return -ENOMEM;
	316
	317	INIT_LIST_HEAD(&dserr->list);
	318	dserr->offset = offset;
	319	dserr->length = length;
	320	dserr->status = status;
	321	dserr->opnum = opnum;
	322	nfs4_stateid_copy(&dserr->stateid, &mirror->stateid);
	323	memcpy(&dserr->deviceid, &mirror->mirror_ds->id_node.deviceid,
	324	NFS4_DEVICEID4_SIZE);
	325
	326	spin_lock(&flo->generic_hdr.plh_inode->i_lock);
b819ed4b	327	ff_layout_add_ds_error_locked(flo, dserr);
d67ae825	328	spin_unlock(&flo->generic_hdr.plh_inode->i_lock);
d67ae825 TH	329
	330	return 0;
	331	}
	332
57f3f4c0 JL	333	static struct rpc_cred *
	334	ff_layout_get_mirror_cred(struct nfs4_ff_layout_mirror *mirror, u32 iomode)
	335	{
3064b686	336	struct rpc_cred cred, __rcu *pcred;
57f3f4c0	337
3064b686 JL	338	if (iomode == IOMODE_READ)
	339	pcred = &mirror->ro_cred;
	340	else
	341	pcred = &mirror->rw_cred;
57f3f4c0 JL	342
	343	rcu_read_lock();
	344	do {
	345	cred = rcu_dereference(*pcred);
	346	if (!cred)
	347	break;
	348
	349	cred = get_rpccred_rcu(cred);
	350	} while(!cred);
	351	rcu_read_unlock();
	352	return cred;
	353	}
	354
d67ae825 TH	355	struct nfs_fh *
	356	nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx)
	357	{
	358	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, mirror_idx);
	359	struct nfs_fh *fh = NULL;
d67ae825	360
65990d1a	361	if (!ff_layout_mirror_valid(lseg, mirror, false)) {
889d94d4	362	pr_err_ratelimited("NFS: %s: No data server for mirror offset index %d\n",
d67ae825	363	__func__, mirror_idx);
d67ae825 TH	364	goto out;
	365	}
	366
	367	/* FIXME: For now assume there is only 1 version available for the DS */
	368	fh = &mirror->fh_versions[0];
	369	out:
	370	return fh;
	371	}
	372
95e2b7e9 JL	373	/**
	374	* nfs4_ff_layout_prepare_ds - prepare a DS connection for an RPC call
	375	* @lseg: the layout segment we're operating on
	376	* @ds_idx: index of the DS to use
	377	* @fail_return: return layout on connect failure?
	378	*
	379	* Try to prepare a DS connection to accept an RPC call. This involves
	380	* selecting a mirror to use and connecting the client to it if it's not
	381	* already connected.
	382	*
	383	* Since we only need a single functioning mirror to satisfy a read, we don't
	384	* want to return the layout if there is one. For writes though, any down
	385	* mirror should result in a LAYOUTRETURN. @fail_return is how we distinguish
	386	* between the two cases.
	387	*
	388	* Returns a pointer to a connected DS object on success or NULL on failure.
	389	*/
d67ae825 TH	390	struct nfs4_pnfs_ds *
	391	nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
	392	bool fail_return)
	393	{
	394	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	395	struct nfs4_pnfs_ds *ds = NULL;
	396	struct nfs4_deviceid_node *devid;
	397	struct inode *ino = lseg->pls_layout->plh_inode;
	398	struct nfs_server *s = NFS_SERVER(ino);
	399	unsigned int max_payload;
a33e4b03	400	int status;
d67ae825	401
65990d1a	402	if (!ff_layout_mirror_valid(lseg, mirror, true)) {
889d94d4	403	pr_err_ratelimited("NFS: %s: No data server for offset index %d\n",
d67ae825	404	__func__, ds_idx);
d67ae825 TH	405	goto out;
	406	}
	407
	408	devid = &mirror->mirror_ds->id_node;
	409	if (ff_layout_test_devid_unavailable(devid))
3dc14735	410	goto out_fail;
d67ae825 TH	411
	412	ds = mirror->mirror_ds->ds;
	413	/* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */
	414	smp_rmb();
	415	if (ds->ds_clp)
90a0be00	416	goto out;
d67ae825 TH	417
	418	/* FIXME: For now we assume the server sent only one version of NFS
	419	* to use for the DS.
	420	*/
a33e4b03	421	status = nfs4_pnfs_ds_connect(s, ds, devid, dataserver_timeo,
d67ae825 TH	422	dataserver_retrans,
d67ae825 TH	423	mirror->mirror_ds->ds_versions[0].version,
7d38de3f	424	mirror->mirror_ds->ds_versions[0].minor_version);
d67ae825 TH	425
d67ae825 TH	426	/* connect success, check rsize/wsize limit */
260f32ad	427	if (!status) {
d67ae825 TH	428	max_payload =
	429	nfs_block_size(rpc_max_payload(ds->ds_clp->cl_rpcclient),
	430	NULL);
	431	if (mirror->mirror_ds->ds_versions[0].rsize > max_payload)
	432	mirror->mirror_ds->ds_versions[0].rsize = max_payload;
	433	if (mirror->mirror_ds->ds_versions[0].wsize > max_payload)
	434	mirror->mirror_ds->ds_versions[0].wsize = max_payload;
3dc14735	435	goto out;
d67ae825	436	}
da066f3f	437	out_fail:
3dc14735 TM	438	ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
	439	mirror, lseg->pls_range.offset,
	440	lseg->pls_range.length, NFS4ERR_NXIO,
	441	OP_ILLEGAL, GFP_NOIO);
3dc14735 TM	442	if (fail_return \|\| !ff_layout_has_available_ds(lseg))
	443	pnfs_error_mark_layout_for_return(ino, lseg);
	444	ds = NULL;
d67ae825 TH	445	out:
	446	return ds;
	447	}
	448
	449	struct rpc_cred *
	450	ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg, u32 ds_idx,
	451	struct rpc_cred *mdscred)
	452	{
	453	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
547a6376	454	struct rpc_cred *cred;
d67ae825	455
10ec57e4	456	if (mirror && !mirror->mirror_ds->ds_versions[0].tightly_coupled) {
57f3f4c0 JL	457	cred = ff_layout_get_mirror_cred(mirror, lseg->pls_range.iomode);
	458	if (!cred)
	459	cred = get_rpccred(mdscred);
	460	} else {
	461	cred = get_rpccred(mdscred);
	462	}
d67ae825 TH	463	return cred;
	464	}
	465
	466	/**
	467	* Find or create a DS rpc client with th MDS server rpc client auth flavor
	468	* in the nfs_client cl_ds_clients list.
	469	*/
	470	struct rpc_clnt *
	471	nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg, u32 ds_idx,
	472	struct nfs_client ds_clp, struct inode inode)
	473	{
	474	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	475
	476	switch (mirror->mirror_ds->ds_versions[0].version) {
	477	case 3:
	478	/* For NFSv3 DS, flavor is set when creating DS connections */
	479	return ds_clp->cl_rpcclient;
	480	case 4:
	481	return nfs4_find_or_create_ds_client(ds_clp, inode);
	482	default:
	483	BUG();
	484	}
	485	}
	486
5b9b3c85 TM	487	void ff_layout_free_ds_ioerr(struct list_head *head)
	488	{
	489	struct nfs4_ff_layout_ds_err *err;
	490
	491	while (!list_empty(head)) {
	492	err = list_first_entry(head,
	493	struct nfs4_ff_layout_ds_err,
	494	list);
	495	list_del(&err->list);
	496	kfree(err);
	497	}
	498	}
	499
d67ae825	500	/* called with inode i_lock held */
5b9b3c85	501	int ff_layout_encode_ds_ioerr(struct xdr_stream xdr, const struct list_head head)
d67ae825	502	{
5b9b3c85	503	struct nfs4_ff_layout_ds_err *err;
d67ae825 TH	504	__be32 *p;
d67ae825 TH	505
5b9b3c85	506	list_for_each_entry(err, head, list) {
d67ae825	507	/* offset(8) + length(8) + stateid(NFS4_STATEID_SIZE)
d1354907 TM	508	* + array length + deviceid(NFS4_DEVICEID4_SIZE)
d1354907 TM	509	* + status(4) + opnum(4)
d67ae825 TH	510	*/
d67ae825 TH	511	p = xdr_reserve_space(xdr,
d1354907	512	28 + NFS4_STATEID_SIZE + NFS4_DEVICEID4_SIZE);
d67ae825 TH	513	if (unlikely(!p))
	514	return -ENOBUFS;
	515	p = xdr_encode_hyper(p, err->offset);
	516	p = xdr_encode_hyper(p, err->length);
	517	p = xdr_encode_opaque_fixed(p, &err->stateid,
	518	NFS4_STATEID_SIZE);
d1354907 TM	519	/* Encode 1 error */
d1354907 TM	520	*p++ = cpu_to_be32(1);
d67ae825 TH	521	p = xdr_encode_opaque_fixed(p, &err->deviceid,
	522	NFS4_DEVICEID4_SIZE);
	523	*p++ = cpu_to_be32(err->status);
	524	*p++ = cpu_to_be32(err->opnum);
5b9b3c85	525	dprintk("%s: offset %llu length %llu status %d op %d\n",
d67ae825	526	__func__, err->offset, err->length, err->status,
5b9b3c85	527	err->opnum);
d67ae825 TH	528	}
	529
	530	return 0;
	531	}
	532
5b9b3c85 TM	533	static
	534	unsigned int do_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
	535	const struct pnfs_layout_range *range,
	536	struct list_head *head,
	537	unsigned int maxnum)
	538	{
	539	struct nfs4_flexfile_layout *flo = FF_LAYOUT_FROM_HDR(lo);
	540	struct inode *inode = lo->plh_inode;
	541	struct nfs4_ff_layout_ds_err err, n;
	542	unsigned int ret = 0;
	543
	544	spin_lock(&inode->i_lock);
	545	list_for_each_entry_safe(err, n, &flo->error_list, list) {
	546	if (!pnfs_is_range_intersecting(err->offset,
	547	pnfs_end_offset(err->offset, err->length),
	548	range->offset,
	549	pnfs_end_offset(range->offset, range->length)))
	550	continue;
	551	if (!maxnum)
	552	break;
	553	list_move(&err->list, head);
	554	maxnum--;
	555	ret++;
	556	}
	557	spin_unlock(&inode->i_lock);
	558	return ret;
	559	}
	560
	561	unsigned int ff_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
	562	const struct pnfs_layout_range *range,
	563	struct list_head *head,
	564	unsigned int maxnum)
	565	{
	566	unsigned int ret;
	567
	568	ret = do_layout_fetch_ds_ioerr(lo, range, head, maxnum);
	569	/* If we're over the max, discard all remaining entries */
	570	if (ret == maxnum) {
	571	LIST_HEAD(discard);
	572	do_layout_fetch_ds_ioerr(lo, range, &discard, -1);
	573	ff_layout_free_ds_ioerr(&discard);
	574	}
	575	return ret;
	576	}
	577
81d6dc8b	578	static bool ff_read_layout_has_available_ds(struct pnfs_layout_segment *lseg)
d67ae825 TH	579	{
	580	struct nfs4_ff_layout_mirror *mirror;
	581	struct nfs4_deviceid_node *devid;
81d6dc8b	582	u32 idx;
d67ae825 TH	583
	584	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
	585	mirror = FF_LAYOUT_COMP(lseg, idx);
65990d1a FI	586	if (mirror) {
	587	if (!mirror->mirror_ds)
	588	return true;
	589	if (IS_ERR(mirror->mirror_ds))
	590	continue;
d67ae825 TH	591	devid = &mirror->mirror_ds->id_node;
	592	if (!ff_layout_test_devid_unavailable(devid))
	593	return true;
	594	}
	595	}
	596
	597	return false;
	598	}
	599
81d6dc8b TM	600	static bool ff_rw_layout_has_available_ds(struct pnfs_layout_segment *lseg)
	601	{
	602	struct nfs4_ff_layout_mirror *mirror;
	603	struct nfs4_deviceid_node *devid;
	604	u32 idx;
	605
	606	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
	607	mirror = FF_LAYOUT_COMP(lseg, idx);
65990d1a	608	if (!mirror \|\| IS_ERR(mirror->mirror_ds))
81d6dc8b	609	return false;
65990d1a FI	610	if (!mirror->mirror_ds)
65990d1a FI	611	continue;
81d6dc8b TM	612	devid = &mirror->mirror_ds->id_node;
	613	if (ff_layout_test_devid_unavailable(devid))
	614	return false;
	615	}
	616
	617	return FF_LAYOUT_MIRROR_COUNT(lseg) != 0;
	618	}
	619
65990d1a	620	static bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg)
81d6dc8b TM	621	{
	622	if (lseg->pls_range.iomode == IOMODE_READ)
	623	return ff_read_layout_has_available_ds(lseg);
	624	/* Note: RW layout needs all mirrors available */
	625	return ff_rw_layout_has_available_ds(lseg);
	626	}
	627
3b13b4b3 TH	628	bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg)
	629	{
	630	return ff_layout_no_fallback_to_mds(lseg) \|\|
	631	ff_layout_has_available_ds(lseg);
	632	}
	633
fb1084e3 TH	634	bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg)
	635	{
	636	return lseg->pls_range.iomode == IOMODE_RW &&
	637	ff_layout_no_read_on_rw(lseg);
	638	}
	639
d67ae825 TH	640	module_param(dataserver_retrans, uint, 0644);
	641	MODULE_PARM_DESC(dataserver_retrans, "The number of times the NFSv4.1 client "
	642	"retries a request before it attempts further "
	643	" recovery action.");
	644	module_param(dataserver_timeo, uint, 0644);
	645	MODULE_PARM_DESC(dataserver_timeo, "The time (in tenths of a second) the "
	646	"NFSv4.1 client waits for a response from a "
	647	" data server before it retries an NFS request.");