[linux-2.6-block.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 = NFS4_DEF_DS_TIMEO;
static unsigned int dataserver_retrans = NFS4_DEF_DS_RETRANS;

void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
{
	if (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_mark_deviceid_unavailable(devid);
	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)
{
	if (mirror == NULL || mirror->mirror_ds == NULL) {
		pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
					lseg);
		return false;
	}
	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;
}

static u64
end_offset(u64 start, u64 len)
{
	u64 end;

	end = start + len;
	return end >= start ? end : NFS4_MAX_UINT64;
}

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

	end = max_t(u64, end_offset(err->offset, err->length),
		    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, &e2->stateid, sizeof(e1->stateid));
	if (ret != 0)
		return ret;
	ret = memcmp(&e1->deviceid, &e2->deviceid, sizeof(e1->deviceid));
	if (ret != 0)
		return ret;
	if (end_offset(e1->offset, e1->length) < e2->offset)
		return -1;
	if (e1->offset > 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_del(&err->list);
		kfree(err);
	}

	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;
}

/* currently we only support AUTH_NONE and AUTH_SYS */
static rpc_authflavor_t
nfs4_ff_layout_choose_authflavor(struct nfs4_ff_layout_mirror *mirror)
{
	if (mirror->uid == (u32)-1)
		return RPC_AUTH_NULL;
	return RPC_AUTH_UNIX;
}

/* fetch cred for NFSv3 DS */
static int ff_layout_update_mirror_cred(struct nfs4_ff_layout_mirror *mirror,
				      struct nfs4_pnfs_ds *ds)
{
	if (ds->ds_clp && !mirror->cred &&
	    mirror->mirror_ds->ds_versions[0].version == 3) {
		struct rpc_auth *auth = ds->ds_clp->cl_rpcclient->cl_auth;
		struct rpc_cred *cred;
		struct auth_cred acred = {
			.uid = make_kuid(&init_user_ns, mirror->uid),
			.gid = make_kgid(&init_user_ns, mirror->gid),
		};

		/* AUTH_NULL ignores acred */
		cred = auth->au_ops->lookup_cred(auth, &acred, 0);
		if (IS_ERR(cred)) {
			dprintk("%s: lookup_cred failed with %ld\n",
				__func__, PTR_ERR(cred));
			return PTR_ERR(cred);
		} else {
			if (cmpxchg(&mirror->cred, NULL, cred))
				put_rpccred(cred);
		}
	}
	return 0;
}

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)) {
		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;
}

/* Upon return, either ds is connected, or ds is NULL */
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;
	rpc_authflavor_t flavor;

	if (!ff_layout_mirror_valid(lseg, mirror)) {
		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;

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

	flavor = nfs4_ff_layout_choose_authflavor(mirror);

	/* 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,
			     flavor);

	/* 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;
	} else {
		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) {
			if (ff_layout_has_available_ds(lseg))
				set_bit(NFS_LAYOUT_RETURN_REQUESTED,
					&lseg->pls_layout->plh_flags);
			else
				pnfs_error_mark_layout_for_return(ino, lseg);
		} else
			pnfs_error_mark_layout_for_return(ino, lseg);
		ds = NULL;
		goto out;
	}
out_update_creds:
	if (ff_layout_update_mirror_cred(mirror, ds))
		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 = ERR_PTR(-EINVAL);

	if (!nfs4_ff_layout_prepare_ds(lseg, ds_idx, true))
		goto out;

	if (mirror && mirror->cred)
		cred = mirror->cred;
	else
		cred = mdscred;
out:
	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();
	}
}

static bool is_range_intersecting(u64 offset1, u64 length1,
				  u64 offset2, u64 length2)
{
	u64 end1 = end_offset(offset1, length1);
	u64 end2 = end_offset(offset2, length2);

	return (end1 == NFS4_MAX_UINT64 || end1 > offset2) &&
	       (end2 == NFS4_MAX_UINT64 || end2 > offset1);
}

/* called with inode i_lock held */
int ff_layout_encode_ds_ioerr(struct nfs4_flexfile_layout *flo,
			      struct xdr_stream *xdr, int *count,
			      const struct pnfs_layout_range *range)
{
	struct nfs4_ff_layout_ds_err *err, *n;
	__be32 *p;

	list_for_each_entry_safe(err, n, &flo->error_list, list) {
		if (!is_range_intersecting(err->offset, err->length,
					   range->offset, range->length))
			continue;
		/* 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);
		*count += 1;
		list_del(&err->list);
		dprintk("%s: offset %llu length %llu status %d op %d count %d\n",
			__func__, err->offset, err->length, err->status,
			err->opnum, *count);
		kfree(err);
	}

	return 0;
}

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 && mirror->mirror_ds) {
			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 || !mirror->mirror_ds)
			return false;
		devid = &mirror->mirror_ds->id_node;
		if (ff_layout_test_devid_unavailable(devid))
			return false;
	}

	return FF_LAYOUT_MIRROR_COUNT(lseg) != 0;
}

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);
}

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
	20	static unsigned int dataserver_timeo = NFS4_DEF_DS_TIMEO;
	21	static unsigned int dataserver_retrans = NFS4_DEF_DS_RETRANS;
	22
	23	void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
	24	{
	25	if (mirror_ds)
	26	nfs4_put_deviceid_node(&mirror_ds->id_node);
	27	}
	28
	29	void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
	30	{
	31	nfs4_print_deviceid(&mirror_ds->id_node.deviceid);
	32	nfs4_pnfs_ds_put(mirror_ds->ds);
84a80f62	33	kfree_rcu(mirror_ds, id_node.rcu);
d67ae825 TH	34	}
	35
	36	/* Decode opaque device data and construct new_ds using it */
	37	struct nfs4_ff_layout_ds *
	38	nfs4_ff_alloc_deviceid_node(struct nfs_server server, struct pnfs_device pdev,
	39	gfp_t gfp_flags)
	40	{
	41	struct xdr_stream stream;
	42	struct xdr_buf buf;
	43	struct page *scratch;
	44	struct list_head dsaddrs;
	45	struct nfs4_pnfs_ds_addr *da;
	46	struct nfs4_ff_layout_ds *new_ds = NULL;
	47	struct nfs4_ff_ds_version *ds_versions = NULL;
	48	u32 mp_count;
	49	u32 version_count;
	50	__be32 *p;
	51	int i, ret = -ENOMEM;
	52
	53	/* set up xdr stream */
	54	scratch = alloc_page(gfp_flags);
	55	if (!scratch)
	56	goto out_err;
	57
	58	new_ds = kzalloc(sizeof(struct nfs4_ff_layout_ds), gfp_flags);
	59	if (!new_ds)
	60	goto out_scratch;
	61
	62	nfs4_init_deviceid_node(&new_ds->id_node,
	63	server,
	64	&pdev->dev_id);
	65	INIT_LIST_HEAD(&dsaddrs);
	66
	67	xdr_init_decode_pages(&stream, &buf, pdev->pages, pdev->pglen);
	68	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
	69
	70	/* multipath count */
	71	p = xdr_inline_decode(&stream, 4);
	72	if (unlikely(!p))
	73	goto out_err_drain_dsaddrs;
	74	mp_count = be32_to_cpup(p);
	75	dprintk("%s: multipath ds count %d\n", __func__, mp_count);
	76
	77	for (i = 0; i < mp_count; i++) {
	78	/* multipath ds */
	79	da = nfs4_decode_mp_ds_addr(server->nfs_client->cl_net,
	80	&stream, gfp_flags);
	81	if (da)
	82	list_add_tail(&da->da_node, &dsaddrs);
	83	}
	84	if (list_empty(&dsaddrs)) {
	85	dprintk("%s: no suitable DS addresses found\n",
	86	__func__);
	87	ret = -ENOMEDIUM;
	88	goto out_err_drain_dsaddrs;
	89	}
	90
	91	/* version count */
	92	p = xdr_inline_decode(&stream, 4);
	93	if (unlikely(!p))
	94	goto out_err_drain_dsaddrs;
	95	version_count = be32_to_cpup(p);
	96	dprintk("%s: version count %d\n", __func__, version_count);
	97
98	ds_versions = kzalloc(version_count * sizeof(struct nfs4_ff_ds_version),
99	gfp_flags);
100	if (!ds_versions)
101	goto out_scratch;
102
103	for (i = 0; i < version_count; i++) {
104	/* 20 = version(4) + minor_version(4) + rsize(4) + wsize(4) +
105	* tightly_coupled(4) */
106	p = xdr_inline_decode(&stream, 20);
107	if (unlikely(!p))
108	goto out_err_drain_dsaddrs;
109	ds_versions[i].version = be32_to_cpup(p++);
110	ds_versions[i].minor_version = be32_to_cpup(p++);
111	ds_versions[i].rsize = nfs_block_size(be32_to_cpup(p++), NULL);
112	ds_versions[i].wsize = nfs_block_size(be32_to_cpup(p++), NULL);
113	ds_versions[i].tightly_coupled = be32_to_cpup(p);
114
115	if (ds_versions[i].rsize > NFS_MAX_FILE_IO_SIZE)
116	ds_versions[i].rsize = NFS_MAX_FILE_IO_SIZE;
117	if (ds_versions[i].wsize > NFS_MAX_FILE_IO_SIZE)
118	ds_versions[i].wsize = NFS_MAX_FILE_IO_SIZE;
119
120	if (ds_versions[i].version != 3 \|\| ds_versions[i].minor_version != 0) {
121	dprintk("%s: [%d] unsupported ds version %d-%d\n", __func__,
122	i, ds_versions[i].version,
123	ds_versions[i].minor_version);
124	ret = -EPROTONOSUPPORT;
125	goto out_err_drain_dsaddrs;
126	}
127
128	dprintk("%s: [%d] vers %u minor_ver %u rsize %u wsize %u coupled %d\n",
129	__func__, i, ds_versions[i].version,
130	ds_versions[i].minor_version,
131	ds_versions[i].rsize,
132	ds_versions[i].wsize,
133	ds_versions[i].tightly_coupled);
134	}
135
136	new_ds->ds_versions = ds_versions;
137	new_ds->ds_versions_cnt = version_count;
138
139	new_ds->ds = nfs4_pnfs_ds_add(&dsaddrs, gfp_flags);
140	if (!new_ds->ds)
141	goto out_err_drain_dsaddrs;
142
143	/* If DS was already in cache, free ds addrs */
144	while (!list_empty(&dsaddrs)) {
145	da = list_first_entry(&dsaddrs,
146	struct nfs4_pnfs_ds_addr,
147	da_node);
148	list_del_init(&da->da_node);
149	kfree(da->da_remotestr);
150	kfree(da);
151	}
152
153	__free_page(scratch);
154	return new_ds;
155
156	out_err_drain_dsaddrs:
157	while (!list_empty(&dsaddrs)) {
158	da = list_first_entry(&dsaddrs, struct nfs4_pnfs_ds_addr,
159	da_node);
160	list_del_init(&da->da_node);
161	kfree(da->da_remotestr);
162	kfree(da);
163	}
164
165	kfree(ds_versions);
166	out_scratch:
167	__free_page(scratch);
168	out_err:
169	kfree(new_ds);
170
171	dprintk("%s ERROR: returning %d\n", __func__, ret);
172	return NULL;
173	}
174
889d94d4 TM	175	static void ff_layout_mark_devid_invalid(struct pnfs_layout_segment *lseg,
	176	struct nfs4_deviceid_node *devid)
	177	{
	178	nfs4_mark_deviceid_unavailable(devid);
	179	if (!ff_layout_has_available_ds(lseg))
	180	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
	181	lseg);
	182	}
	183
	184	static bool ff_layout_mirror_valid(struct pnfs_layout_segment *lseg,
	185	struct nfs4_ff_layout_mirror *mirror)
	186	{
	187	if (mirror == NULL \|\| mirror->mirror_ds == NULL) {
	188	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
	189	lseg);
	190	return false;
	191	}
	192	if (mirror->mirror_ds->ds == NULL) {
	193	struct nfs4_deviceid_node *devid;
	194	devid = &mirror->mirror_ds->id_node;
	195	ff_layout_mark_devid_invalid(lseg, devid);
	196	return false;
	197	}
	198	return true;
	199	}
	200
d67ae825 TH	201	static u64
	202	end_offset(u64 start, u64 len)
	203	{
	204	u64 end;
	205
	206	end = start + len;
	207	return end >= start ? end : NFS4_MAX_UINT64;
	208	}
	209
	210	static void extend_ds_error(struct nfs4_ff_layout_ds_err *err,
	211	u64 offset, u64 length)
	212	{
	213	u64 end;
	214
	215	end = max_t(u64, end_offset(err->offset, err->length),
	216	end_offset(offset, length));
	217	err->offset = min_t(u64, err->offset, offset);
	218	err->length = end - err->offset;
	219	}
	220
b819ed4b TM	221	static int
	222	ff_ds_error_match(const struct nfs4_ff_layout_ds_err *e1,
	223	const struct nfs4_ff_layout_ds_err *e2)
d67ae825	224	{
b819ed4b TM	225	int ret;
	226
	227	if (e1->opnum != e2->opnum)
	228	return e1->opnum < e2->opnum ? -1 : 1;
	229	if (e1->status != e2->status)
	230	return e1->status < e2->status ? -1 : 1;
	231	ret = memcmp(&e1->stateid, &e2->stateid, sizeof(e1->stateid));
	232	if (ret != 0)
	233	return ret;
	234	ret = memcmp(&e1->deviceid, &e2->deviceid, sizeof(e1->deviceid));
	235	if (ret != 0)
	236	return ret;
	237	if (end_offset(e1->offset, e1->length) < e2->offset)
	238	return -1;
	239	if (e1->offset > end_offset(e2->offset, e2->length))
	240	return 1;
	241	/* If ranges overlap or are contiguous, they are the same */
	242	return 0;
d67ae825 TH	243	}
d67ae825 TH	244
b819ed4b	245	static void
d67ae825 TH	246	ff_layout_add_ds_error_locked(struct nfs4_flexfile_layout *flo,
	247	struct nfs4_ff_layout_ds_err *dserr)
	248	{
b819ed4b TM	249	struct nfs4_ff_layout_ds_err err, tmp;
	250	struct list_head *head = &flo->error_list;
	251	int match;
	252
	253	/* Do insertion sort w/ merges */
	254	list_for_each_entry_safe(err, tmp, &flo->error_list, list) {
	255	match = ff_ds_error_match(err, dserr);
	256	if (match < 0)
	257	continue;
	258	if (match > 0) {
	259	/* Add entry "dserr" _before_ entry "err" */
	260	head = &err->list;
d67ae825 TH	261	break;
d67ae825 TH	262	}
b819ed4b TM	263	/* Entries match, so merge "err" into "dserr" */
	264	extend_ds_error(dserr, err->offset, err->length);
	265	list_del(&err->list);
	266	kfree(err);
d67ae825 TH	267	}
d67ae825 TH	268
b819ed4b	269	list_add_tail(&dserr->list, head);
d67ae825 TH	270	}
	271
	272	int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
	273	struct nfs4_ff_layout_mirror *mirror, u64 offset,
	274	u64 length, int status, enum nfs_opnum4 opnum,
	275	gfp_t gfp_flags)
	276	{
	277	struct nfs4_ff_layout_ds_err *dserr;
d67ae825 TH	278
	279	if (status == 0)
	280	return 0;
	281
	282	if (mirror->mirror_ds == NULL)
	283	return -EINVAL;
	284
d67ae825 TH	285	dserr = kmalloc(sizeof(*dserr), gfp_flags);
	286	if (!dserr)
	287	return -ENOMEM;
	288
	289	INIT_LIST_HEAD(&dserr->list);
	290	dserr->offset = offset;
	291	dserr->length = length;
	292	dserr->status = status;
	293	dserr->opnum = opnum;
	294	nfs4_stateid_copy(&dserr->stateid, &mirror->stateid);
	295	memcpy(&dserr->deviceid, &mirror->mirror_ds->id_node.deviceid,
	296	NFS4_DEVICEID4_SIZE);
	297
	298	spin_lock(&flo->generic_hdr.plh_inode->i_lock);
b819ed4b	299	ff_layout_add_ds_error_locked(flo, dserr);
d67ae825	300	spin_unlock(&flo->generic_hdr.plh_inode->i_lock);
d67ae825 TH	301
	302	return 0;
	303	}
	304
	305	/* currently we only support AUTH_NONE and AUTH_SYS */
	306	static rpc_authflavor_t
	307	nfs4_ff_layout_choose_authflavor(struct nfs4_ff_layout_mirror *mirror)
	308	{
	309	if (mirror->uid == (u32)-1)
	310	return RPC_AUTH_NULL;
	311	return RPC_AUTH_UNIX;
	312	}
	313
	314	/* fetch cred for NFSv3 DS */
	315	static int ff_layout_update_mirror_cred(struct nfs4_ff_layout_mirror *mirror,
	316	struct nfs4_pnfs_ds *ds)
	317	{
	318	if (ds->ds_clp && !mirror->cred &&
	319	mirror->mirror_ds->ds_versions[0].version == 3) {
	320	struct rpc_auth *auth = ds->ds_clp->cl_rpcclient->cl_auth;
	321	struct rpc_cred *cred;
	322	struct auth_cred acred = {
	323	.uid = make_kuid(&init_user_ns, mirror->uid),
	324	.gid = make_kgid(&init_user_ns, mirror->gid),
	325	};
	326
	327	/* AUTH_NULL ignores acred */
	328	cred = auth->au_ops->lookup_cred(auth, &acred, 0);
	329	if (IS_ERR(cred)) {
	330	dprintk("%s: lookup_cred failed with %ld\n",
	331	__func__, PTR_ERR(cred));
	332	return PTR_ERR(cred);
	333	} else {
a24221dc JL	334	if (cmpxchg(&mirror->cred, NULL, cred))
a24221dc JL	335	put_rpccred(cred);
d67ae825 TH	336	}
	337	}
	338	return 0;
	339	}
	340
	341	struct nfs_fh *
	342	nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx)
	343	{
	344	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, mirror_idx);
	345	struct nfs_fh *fh = NULL;
d67ae825	346
889d94d4 TM	347	if (!ff_layout_mirror_valid(lseg, mirror)) {
889d94d4 TM	348	pr_err_ratelimited("NFS: %s: No data server for mirror offset index %d\n",
d67ae825	349	__func__, mirror_idx);
d67ae825 TH	350	goto out;
	351	}
	352
	353	/* FIXME: For now assume there is only 1 version available for the DS */
	354	fh = &mirror->fh_versions[0];
	355	out:
	356	return fh;
	357	}
	358
	359	/* Upon return, either ds is connected, or ds is NULL */
	360	struct nfs4_pnfs_ds *
	361	nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
	362	bool fail_return)
	363	{
	364	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	365	struct nfs4_pnfs_ds *ds = NULL;
	366	struct nfs4_deviceid_node *devid;
	367	struct inode *ino = lseg->pls_layout->plh_inode;
	368	struct nfs_server *s = NFS_SERVER(ino);
	369	unsigned int max_payload;
	370	rpc_authflavor_t flavor;
	371
889d94d4 TM	372	if (!ff_layout_mirror_valid(lseg, mirror)) {
889d94d4 TM	373	pr_err_ratelimited("NFS: %s: No data server for offset index %d\n",
d67ae825	374	__func__, ds_idx);
d67ae825 TH	375	goto out;
	376	}
	377
	378	devid = &mirror->mirror_ds->id_node;
	379	if (ff_layout_test_devid_unavailable(devid))
	380	goto out;
	381
	382	ds = mirror->mirror_ds->ds;
	383	/* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */
	384	smp_rmb();
	385	if (ds->ds_clp)
0c8315dd	386	goto out_update_creds;
d67ae825 TH	387
	388	flavor = nfs4_ff_layout_choose_authflavor(mirror);
	389
	390	/* FIXME: For now we assume the server sent only one version of NFS
	391	* to use for the DS.
	392	*/
	393	nfs4_pnfs_ds_connect(s, ds, devid, dataserver_timeo,
	394	dataserver_retrans,
	395	mirror->mirror_ds->ds_versions[0].version,
	396	mirror->mirror_ds->ds_versions[0].minor_version,
	397	flavor);
	398
	399	/* connect success, check rsize/wsize limit */
	400	if (ds->ds_clp) {
	401	max_payload =
	402	nfs_block_size(rpc_max_payload(ds->ds_clp->cl_rpcclient),
	403	NULL);
	404	if (mirror->mirror_ds->ds_versions[0].rsize > max_payload)
	405	mirror->mirror_ds->ds_versions[0].rsize = max_payload;
	406	if (mirror->mirror_ds->ds_versions[0].wsize > max_payload)
	407	mirror->mirror_ds->ds_versions[0].wsize = max_payload;
	408	} else {
	409	ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
	410	mirror, lseg->pls_range.offset,
	411	lseg->pls_range.length, NFS4ERR_NXIO,
	412	OP_ILLEGAL, GFP_NOIO);
2e5b29f0	413	if (!fail_return) {
d67ae825	414	if (ff_layout_has_available_ds(lseg))
2370abda	415	set_bit(NFS_LAYOUT_RETURN_REQUESTED,
d67ae825	416	&lseg->pls_layout->plh_flags);
2e5b29f0	417	else
d67ae825	418	pnfs_error_mark_layout_for_return(ino, lseg);
2e5b29f0 TM	419	} else
2e5b29f0 TM	420	pnfs_error_mark_layout_for_return(ino, lseg);
849dc324 JL	421	ds = NULL;
849dc324 JL	422	goto out;
d67ae825	423	}
0c8315dd	424	out_update_creds:
d67ae825 TH	425	if (ff_layout_update_mirror_cred(mirror, ds))
	426	ds = NULL;
	427	out:
	428	return ds;
	429	}
	430
	431	struct rpc_cred *
	432	ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg, u32 ds_idx,
	433	struct rpc_cred *mdscred)
	434	{
	435	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	436	struct rpc_cred *cred = ERR_PTR(-EINVAL);
	437
	438	if (!nfs4_ff_layout_prepare_ds(lseg, ds_idx, true))
	439	goto out;
	440
	441	if (mirror && mirror->cred)
	442	cred = mirror->cred;
	443	else
	444	cred = mdscred;
	445	out:
	446	return cred;
	447	}
	448
	449	/**
	450	* Find or create a DS rpc client with th MDS server rpc client auth flavor
	451	* in the nfs_client cl_ds_clients list.
	452	*/
	453	struct rpc_clnt *
	454	nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg, u32 ds_idx,
	455	struct nfs_client ds_clp, struct inode inode)
	456	{
	457	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
	458
	459	switch (mirror->mirror_ds->ds_versions[0].version) {
	460	case 3:
	461	/* For NFSv3 DS, flavor is set when creating DS connections */
	462	return ds_clp->cl_rpcclient;
	463	case 4:
	464	return nfs4_find_or_create_ds_client(ds_clp, inode);
	465	default:
	466	BUG();
	467	}
	468	}
	469
	470	static bool is_range_intersecting(u64 offset1, u64 length1,
	471	u64 offset2, u64 length2)
	472	{
	473	u64 end1 = end_offset(offset1, length1);
	474	u64 end2 = end_offset(offset2, length2);
	475
	476	return (end1 == NFS4_MAX_UINT64 \|\| end1 > offset2) &&
	477	(end2 == NFS4_MAX_UINT64 \|\| end2 > offset1);
	478	}
	479
	480	/* called with inode i_lock held */
	481	int ff_layout_encode_ds_ioerr(struct nfs4_flexfile_layout *flo,
	482	struct xdr_stream xdr, int count,
	483	const struct pnfs_layout_range *range)
	484	{
	485	struct nfs4_ff_layout_ds_err err, n;
	486	__be32 *p;
	487
	488	list_for_each_entry_safe(err, n, &flo->error_list, list) {
489	if (!is_range_intersecting(err->offset, err->length,
490	range->offset, range->length))
491	continue;
492	/* offset(8) + length(8) + stateid(NFS4_STATEID_SIZE)
d1354907 TM	493	* + array length + deviceid(NFS4_DEVICEID4_SIZE)
d1354907 TM	494	* + status(4) + opnum(4)
d67ae825 TH	495	*/
d67ae825 TH	496	p = xdr_reserve_space(xdr,
d1354907	497	28 + NFS4_STATEID_SIZE + NFS4_DEVICEID4_SIZE);
d67ae825 TH	498	if (unlikely(!p))
	499	return -ENOBUFS;
	500	p = xdr_encode_hyper(p, err->offset);
	501	p = xdr_encode_hyper(p, err->length);
	502	p = xdr_encode_opaque_fixed(p, &err->stateid,
	503	NFS4_STATEID_SIZE);
d1354907 TM	504	/* Encode 1 error */
d1354907 TM	505	*p++ = cpu_to_be32(1);
d67ae825 TH	506	p = xdr_encode_opaque_fixed(p, &err->deviceid,
	507	NFS4_DEVICEID4_SIZE);
	508	*p++ = cpu_to_be32(err->status);
	509	*p++ = cpu_to_be32(err->opnum);
	510	*count += 1;
	511	list_del(&err->list);
d67ae825 TH	512	dprintk("%s: offset %llu length %llu status %d op %d count %d\n",
	513	__func__, err->offset, err->length, err->status,
	514	err->opnum, *count);
480486b4	515	kfree(err);
d67ae825 TH	516	}
	517
	518	return 0;
	519	}
	520
81d6dc8b	521	static bool ff_read_layout_has_available_ds(struct pnfs_layout_segment *lseg)
d67ae825 TH	522	{
	523	struct nfs4_ff_layout_mirror *mirror;
	524	struct nfs4_deviceid_node *devid;
81d6dc8b	525	u32 idx;
d67ae825 TH	526
	527	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
	528	mirror = FF_LAYOUT_COMP(lseg, idx);
	529	if (mirror && mirror->mirror_ds) {
	530	devid = &mirror->mirror_ds->id_node;
	531	if (!ff_layout_test_devid_unavailable(devid))
	532	return true;
	533	}
	534	}
	535
	536	return false;
	537	}
	538
81d6dc8b TM	539	static bool ff_rw_layout_has_available_ds(struct pnfs_layout_segment *lseg)
	540	{
	541	struct nfs4_ff_layout_mirror *mirror;
	542	struct nfs4_deviceid_node *devid;
	543	u32 idx;
	544
	545	for (idx = 0; idx < FF_LAYOUT_MIRROR_COUNT(lseg); idx++) {
	546	mirror = FF_LAYOUT_COMP(lseg, idx);
	547	if (!mirror \|\| !mirror->mirror_ds)
	548	return false;
	549	devid = &mirror->mirror_ds->id_node;
	550	if (ff_layout_test_devid_unavailable(devid))
	551	return false;
	552	}
	553
	554	return FF_LAYOUT_MIRROR_COUNT(lseg) != 0;
	555	}
	556
	557	bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg)
	558	{
	559	if (lseg->pls_range.iomode == IOMODE_READ)
	560	return ff_read_layout_has_available_ds(lseg);
	561	/* Note: RW layout needs all mirrors available */
	562	return ff_rw_layout_has_available_ds(lseg);
	563	}
	564
d67ae825 TH	565	module_param(dataserver_retrans, uint, 0644);
	566	MODULE_PARM_DESC(dataserver_retrans, "The number of times the NFSv4.1 client "
	567	"retries a request before it attempts further "
	568	" recovery action.");
	569	module_param(dataserver_timeo, uint, 0644);
	570	MODULE_PARM_DESC(dataserver_timeo, "The time (in tenths of a second) the "
	571	"NFSv4.1 client waits for a response from a "
	572	" data server before it retries an NFS request.");