[deliverable/linux.git] / fs / nfs / flexfilelayout / flexfilelayout.h

/*
 * NFSv4 flexfile layout driver data structures.
 *
 * Copyright (c) 2014, Primary Data, Inc. All rights reserved.
 *
 * Tao Peng <bergwolf@primarydata.com>
 */

#ifndef FS_NFS_NFS4FLEXFILELAYOUT_H
#define FS_NFS_NFS4FLEXFILELAYOUT_H

#define FF_FLAGS_NO_LAYOUTCOMMIT 1
#define FF_FLAGS_NO_IO_THRU_MDS  2
#define FF_FLAGS_NO_READ_IO      4

#include "../pnfs.h"

/* XXX: Let's filter out insanely large mirror count for now to avoid oom
 * due to network error etc. */
#define NFS4_FLEXFILE_LAYOUT_MAX_MIRROR_CNT 4096

/* LAYOUTSTATS report interval in ms */
#define FF_LAYOUTSTATS_REPORT_INTERVAL (60000L)

struct nfs4_ff_ds_version {
	u32				version;
	u32				minor_version;
	u32				rsize;
	u32				wsize;
	bool				tightly_coupled;
};

/* chained in global deviceid hlist */
struct nfs4_ff_layout_ds {
	struct nfs4_deviceid_node	id_node;
	u32				ds_versions_cnt;
	struct nfs4_ff_ds_version	*ds_versions;
	struct nfs4_pnfs_ds		*ds;
};

struct nfs4_ff_layout_ds_err {
	struct list_head		list; /* linked in mirror error_list */
	u64				offset;
	u64				length;
	int				status;
	enum nfs_opnum4			opnum;
	nfs4_stateid			stateid;
	struct nfs4_deviceid		deviceid;
};

struct nfs4_ff_io_stat {
	__u64				ops_requested;
	__u64				bytes_requested;
	__u64				ops_completed;
	__u64				bytes_completed;
	__u64				bytes_not_delivered;
	ktime_t				total_busy_time;
	ktime_t				aggregate_completion_time;
};

struct nfs4_ff_busy_timer {
	ktime_t start_time;
	atomic_t n_ops;
};

struct nfs4_ff_layoutstat {
	struct nfs4_ff_io_stat io_stat;
	struct nfs4_ff_busy_timer busy_timer;
};

struct nfs4_ff_layout_mirror {
	struct pnfs_layout_hdr		*layout;
	struct list_head		mirrors;
	u32				ds_count;
	u32				efficiency;
	struct nfs4_ff_layout_ds	*mirror_ds;
	u32				fh_versions_cnt;
	struct nfs_fh			*fh_versions;
	nfs4_stateid			stateid;
	struct rpc_cred	__rcu		*ro_cred;
	struct rpc_cred	__rcu		*rw_cred;
	atomic_t			ref;
	spinlock_t			lock;
	struct nfs4_ff_layoutstat	read_stat;
	struct nfs4_ff_layoutstat	write_stat;
	ktime_t				start_time;
	ktime_t				last_report_time;
	u32				report_interval;
};

struct nfs4_ff_layout_segment {
	struct pnfs_layout_segment	generic_hdr;
	u64				stripe_unit;
	u32				flags;
	u32				mirror_array_cnt;
	struct nfs4_ff_layout_mirror	**mirror_array;
};

struct nfs4_flexfile_layout {
	struct pnfs_layout_hdr generic_hdr;
	struct pnfs_ds_commit_info commit_info;
	struct list_head	mirrors;
	struct list_head	error_list; /* nfs4_ff_layout_ds_err */
};

static inline struct nfs4_flexfile_layout *
FF_LAYOUT_FROM_HDR(struct pnfs_layout_hdr *lo)
{
	return container_of(lo, struct nfs4_flexfile_layout, generic_hdr);
}

static inline struct nfs4_ff_layout_segment *
FF_LAYOUT_LSEG(struct pnfs_layout_segment *lseg)
{
	return container_of(lseg,
			    struct nfs4_ff_layout_segment,
			    generic_hdr);
}

static inline struct nfs4_deviceid_node *
FF_LAYOUT_DEVID_NODE(struct pnfs_layout_segment *lseg, u32 idx)
{
	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt ||
	    FF_LAYOUT_LSEG(lseg)->mirror_array[idx] == NULL ||
	    FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds == NULL)
		return NULL;
	return &FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds->id_node;
}

static inline struct nfs4_ff_layout_ds *
FF_LAYOUT_MIRROR_DS(struct nfs4_deviceid_node *node)
{
	return container_of(node, struct nfs4_ff_layout_ds, id_node);
}

static inline struct nfs4_ff_layout_mirror *
FF_LAYOUT_COMP(struct pnfs_layout_segment *lseg, u32 idx)
{
	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt)
		return NULL;
	return FF_LAYOUT_LSEG(lseg)->mirror_array[idx];
}

static inline u32
FF_LAYOUT_MIRROR_COUNT(struct pnfs_layout_segment *lseg)
{
	return FF_LAYOUT_LSEG(lseg)->mirror_array_cnt;
}

static inline bool
ff_layout_no_fallback_to_mds(struct pnfs_layout_segment *lseg)
{
	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_IO_THRU_MDS;
}

static inline bool
ff_layout_no_read_on_rw(struct pnfs_layout_segment *lseg)
{
	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_READ_IO;
}

static inline bool
ff_layout_test_devid_unavailable(struct nfs4_deviceid_node *node)
{
	return nfs4_test_deviceid_unavailable(node);
}

static inline int
nfs4_ff_layout_ds_version(struct pnfs_layout_segment *lseg, u32 ds_idx)
{
	return FF_LAYOUT_COMP(lseg, ds_idx)->mirror_ds->ds_versions[0].version;
}

struct nfs4_ff_layout_ds *
nfs4_ff_alloc_deviceid_node(struct nfs_server *server, struct pnfs_device *pdev,
			    gfp_t gfp_flags);
void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
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);
int ff_layout_encode_ds_ioerr(struct nfs4_flexfile_layout *flo,
			      struct xdr_stream *xdr, int *count,
			      const struct pnfs_layout_range *range);
struct nfs_fh *
nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx);

struct nfs4_pnfs_ds *
nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
			  bool fail_return);

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 rpc_cred *ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg,
				       u32 ds_idx, struct rpc_cred *mdscred);
bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg);
bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg);
bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg);

#endif /* FS_NFS_NFS4FLEXFILELAYOUT_H */
Commit	Line	Data
d67ae825 TH	1	/*
	2	* NFSv4 flexfile layout driver data structures.
	3	*
	4	* Copyright (c) 2014, Primary Data, Inc. All rights reserved.
	5	*
	6	* Tao Peng <bergwolf@primarydata.com>
	7	*/
	8
	9	#ifndef FS_NFS_NFS4FLEXFILELAYOUT_H
	10	#define FS_NFS_NFS4FLEXFILELAYOUT_H
	11
c0f5f505	12	#define FF_FLAGS_NO_LAYOUTCOMMIT 1
fb1084e3 TH	13	#define FF_FLAGS_NO_IO_THRU_MDS 2
fb1084e3 TH	14	#define FF_FLAGS_NO_READ_IO 4
c0f5f505	15
d67ae825 TH	16	#include "../pnfs.h"
	17
	18	/* XXX: Let's filter out insanely large mirror count for now to avoid oom
	19	* due to network error etc. */
	20	#define NFS4_FLEXFILE_LAYOUT_MAX_MIRROR_CNT 4096
	21
97ba375b PT	22	/* LAYOUTSTATS report interval in ms */
	23	#define FF_LAYOUTSTATS_REPORT_INTERVAL (60000L)
	24
d67ae825 TH	25	struct nfs4_ff_ds_version {
	26	u32 version;
	27	u32 minor_version;
	28	u32 rsize;
	29	u32 wsize;
	30	bool tightly_coupled;
	31	};
	32
	33	/* chained in global deviceid hlist */
	34	struct nfs4_ff_layout_ds {
	35	struct nfs4_deviceid_node id_node;
	36	u32 ds_versions_cnt;
	37	struct nfs4_ff_ds_version *ds_versions;
	38	struct nfs4_pnfs_ds *ds;
	39	};
	40
	41	struct nfs4_ff_layout_ds_err {
	42	struct list_head list; /* linked in mirror error_list */
	43	u64 offset;
	44	u64 length;
	45	int status;
	46	enum nfs_opnum4 opnum;
	47	nfs4_stateid stateid;
	48	struct nfs4_deviceid deviceid;
	49	};
	50
abcb7bfc TM	51	struct nfs4_ff_io_stat {
	52	__u64 ops_requested;
	53	__u64 bytes_requested;
	54	__u64 ops_completed;
	55	__u64 bytes_completed;
	56	__u64 bytes_not_delivered;
	57	ktime_t total_busy_time;
	58	ktime_t aggregate_completion_time;
	59	};
	60
	61	struct nfs4_ff_busy_timer {
	62	ktime_t start_time;
	63	atomic_t n_ops;
	64	};
	65
	66	struct nfs4_ff_layoutstat {
	67	struct nfs4_ff_io_stat io_stat;
	68	struct nfs4_ff_busy_timer busy_timer;
	69	};
	70
d67ae825	71	struct nfs4_ff_layout_mirror {
266d12d4 TM	72	struct pnfs_layout_hdr *layout;
266d12d4 TM	73	struct list_head mirrors;
d67ae825 TH	74	u32 ds_count;
	75	u32 efficiency;
	76	struct nfs4_ff_layout_ds *mirror_ds;
	77	u32 fh_versions_cnt;
	78	struct nfs_fh *fh_versions;
	79	nfs4_stateid stateid;
3064b686 JL	80	struct rpc_cred __rcu *ro_cred;
3064b686 JL	81	struct rpc_cred __rcu *rw_cred;
28a0d72c	82	atomic_t ref;
d67ae825	83	spinlock_t lock;
abcb7bfc TM	84	struct nfs4_ff_layoutstat read_stat;
abcb7bfc TM	85	struct nfs4_ff_layoutstat write_stat;
d983803d	86	ktime_t start_time;
97ba375b	87	ktime_t last_report_time;
d0379a5d	88	u32 report_interval;
d67ae825 TH	89	};
	90
	91	struct nfs4_ff_layout_segment {
	92	struct pnfs_layout_segment generic_hdr;
	93	u64 stripe_unit;
c0f5f505	94	u32 flags;
d67ae825 TH	95	u32 mirror_array_cnt;
	96	struct nfs4_ff_layout_mirror **mirror_array;
	97	};
	98
	99	struct nfs4_flexfile_layout {
	100	struct pnfs_layout_hdr generic_hdr;
	101	struct pnfs_ds_commit_info commit_info;
266d12d4	102	struct list_head mirrors;
d67ae825 TH	103	struct list_head error_list; /* nfs4_ff_layout_ds_err */
	104	};
	105
	106	static inline struct nfs4_flexfile_layout *
	107	FF_LAYOUT_FROM_HDR(struct pnfs_layout_hdr *lo)
	108	{
	109	return container_of(lo, struct nfs4_flexfile_layout, generic_hdr);
	110	}
	111
	112	static inline struct nfs4_ff_layout_segment *
	113	FF_LAYOUT_LSEG(struct pnfs_layout_segment *lseg)
	114	{
	115	return container_of(lseg,
	116	struct nfs4_ff_layout_segment,
	117	generic_hdr);
	118	}
	119
	120	static inline struct nfs4_deviceid_node *
	121	FF_LAYOUT_DEVID_NODE(struct pnfs_layout_segment *lseg, u32 idx)
	122	{
	123	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt \|\|
	124	FF_LAYOUT_LSEG(lseg)->mirror_array[idx] == NULL \|\|
	125	FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds == NULL)
	126	return NULL;
	127	return &FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds->id_node;
	128	}
	129
	130	static inline struct nfs4_ff_layout_ds *
	131	FF_LAYOUT_MIRROR_DS(struct nfs4_deviceid_node *node)
	132	{
	133	return container_of(node, struct nfs4_ff_layout_ds, id_node);
	134	}
	135
	136	static inline struct nfs4_ff_layout_mirror *
	137	FF_LAYOUT_COMP(struct pnfs_layout_segment *lseg, u32 idx)
	138	{
	139	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt)
	140	return NULL;
	141	return FF_LAYOUT_LSEG(lseg)->mirror_array[idx];
	142	}
	143
	144	static inline u32
	145	FF_LAYOUT_MIRROR_COUNT(struct pnfs_layout_segment *lseg)
	146	{
	147	return FF_LAYOUT_LSEG(lseg)->mirror_array_cnt;
	148	}
	149
260074cd TM	150	static inline bool
	151	ff_layout_no_fallback_to_mds(struct pnfs_layout_segment *lseg)
	152	{
	153	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_IO_THRU_MDS;
	154	}
	155
fb1084e3 TH	156	static inline bool
	157	ff_layout_no_read_on_rw(struct pnfs_layout_segment *lseg)
	158	{
	159	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_READ_IO;
	160	}
	161
d67ae825 TH	162	static inline bool
	163	ff_layout_test_devid_unavailable(struct nfs4_deviceid_node *node)
	164	{
	165	return nfs4_test_deviceid_unavailable(node);
	166	}
	167
	168	static inline int
	169	nfs4_ff_layout_ds_version(struct pnfs_layout_segment *lseg, u32 ds_idx)
	170	{
	171	return FF_LAYOUT_COMP(lseg, ds_idx)->mirror_ds->ds_versions[0].version;
	172	}
	173
	174	struct nfs4_ff_layout_ds *
	175	nfs4_ff_alloc_deviceid_node(struct nfs_server server, struct pnfs_device pdev,
	176	gfp_t gfp_flags);
	177	void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
	178	void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
	179	int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
	180	struct nfs4_ff_layout_mirror *mirror, u64 offset,
	181	u64 length, int status, enum nfs_opnum4 opnum,
	182	gfp_t gfp_flags);
	183	int ff_layout_encode_ds_ioerr(struct nfs4_flexfile_layout *flo,
	184	struct xdr_stream xdr, int count,
	185	const struct pnfs_layout_range *range);
	186	struct nfs_fh *
	187	nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx);
	188
	189	struct nfs4_pnfs_ds *
	190	nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
	191	bool fail_return);
	192
	193	struct rpc_clnt *
	194	nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg,
	195	u32 ds_idx,
	196	struct nfs_client *ds_clp,
	197	struct inode *inode);
	198	struct rpc_cred ff_layout_get_ds_cred(struct pnfs_layout_segment lseg,
	199	u32 ds_idx, struct rpc_cred *mdscred);
	200	bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg);
3b13b4b3	201	bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg);
fb1084e3 TH	202	bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg);
fb1084e3 TH	203
d67ae825	204	#endif /* FS_NFS_NFS4FLEXFILELAYOUT_H */