[mirror_ubuntu-bionic-kernel.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)
#define FF_LAYOUTSTATS_MAXDEV 4

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_deviceid		devid;
	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;
	unsigned long			flags;
	struct nfs4_ff_layoutstat	read_stat;
	struct nfs4_ff_layoutstat	write_stat;
	ktime_t				start_time;
	u32				report_interval;
};

#define NFS4_FF_MIRROR_STAT_AVAIL	(0)

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 */
	ktime_t			last_report_time; /* Layoutstat report times */
};

struct nfs4_flexfile_layoutreturn_args {
	struct list_head errors;
	struct nfs42_layoutstat_devinfo devinfo[FF_LAYOUTSTATS_MAXDEV];
	unsigned int num_errors;
	unsigned int num_dev;
	struct page *pages[1];
};

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)
{
	/*
	 * Flexfiles should never mark a DS unavailable, but if it does
	 * print a (ratelimited) warning as this can affect performance.
	 */
	if (nfs4_test_deviceid_unavailable(node)) {
		u32 *p = (u32 *)node->deviceid.data;

		pr_warn_ratelimited("NFS: flexfiles layout referencing an "
				"unavailable device [%x%x%x%x]\n",
				p[0], p[1], p[2], p[3]);
		return true;
	}
	return false;
}

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 xdr_stream *xdr, const struct list_head *head);
void ff_layout_free_ds_ioerr(struct list_head *head);
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);
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_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 */
97ba375b PT	23	#define FF_LAYOUTSTATS_REPORT_INTERVAL (60000L)
230bc962	24	#define FF_LAYOUTSTATS_MAXDEV 4
97ba375b	25
d67ae825 TH	26	struct nfs4_ff_ds_version {
	27	u32 version;
	28	u32 minor_version;
	29	u32 rsize;
	30	u32 wsize;
	31	bool tightly_coupled;
	32	};
	33
	34	/* chained in global deviceid hlist */
	35	struct nfs4_ff_layout_ds {
	36	struct nfs4_deviceid_node id_node;
	37	u32 ds_versions_cnt;
	38	struct nfs4_ff_ds_version *ds_versions;
	39	struct nfs4_pnfs_ds *ds;
	40	};
	41
	42	struct nfs4_ff_layout_ds_err {
	43	struct list_head list; /* linked in mirror error_list */
	44	u64 offset;
	45	u64 length;
	46	int status;
	47	enum nfs_opnum4 opnum;
	48	nfs4_stateid stateid;
	49	struct nfs4_deviceid deviceid;
	50	};
	51
abcb7bfc TM	52	struct nfs4_ff_io_stat {
	53	__u64 ops_requested;
	54	__u64 bytes_requested;
	55	__u64 ops_completed;
	56	__u64 bytes_completed;
	57	__u64 bytes_not_delivered;
	58	ktime_t total_busy_time;
	59	ktime_t aggregate_completion_time;
	60	};
	61
	62	struct nfs4_ff_busy_timer {
	63	ktime_t start_time;
	64	atomic_t n_ops;
	65	};
	66
	67	struct nfs4_ff_layoutstat {
	68	struct nfs4_ff_io_stat io_stat;
	69	struct nfs4_ff_busy_timer busy_timer;
	70	};
	71
d67ae825	72	struct nfs4_ff_layout_mirror {
266d12d4 TM	73	struct pnfs_layout_hdr *layout;
266d12d4 TM	74	struct list_head mirrors;
d67ae825 TH	75	u32 ds_count;
d67ae825 TH	76	u32 efficiency;
65990d1a	77	struct nfs4_deviceid devid;
d67ae825 TH	78	struct nfs4_ff_layout_ds *mirror_ds;
	79	u32 fh_versions_cnt;
	80	struct nfs_fh *fh_versions;
	81	nfs4_stateid stateid;
3064b686 JL	82	struct rpc_cred __rcu *ro_cred;
3064b686 JL	83	struct rpc_cred __rcu *rw_cred;
28a0d72c	84	atomic_t ref;
d67ae825	85	spinlock_t lock;
06946c6a	86	unsigned long flags;
abcb7bfc TM	87	struct nfs4_ff_layoutstat read_stat;
abcb7bfc TM	88	struct nfs4_ff_layoutstat write_stat;
d983803d	89	ktime_t start_time;
d0379a5d	90	u32 report_interval;
d67ae825 TH	91	};
d67ae825 TH	92
06946c6a TM	93	#define NFS4_FF_MIRROR_STAT_AVAIL (0)
06946c6a TM	94
d67ae825 TH	95	struct nfs4_ff_layout_segment {
	96	struct pnfs_layout_segment generic_hdr;
	97	u64 stripe_unit;
c0f5f505	98	u32 flags;
d67ae825 TH	99	u32 mirror_array_cnt;
	100	struct nfs4_ff_layout_mirror **mirror_array;
	101	};
	102
	103	struct nfs4_flexfile_layout {
	104	struct pnfs_layout_hdr generic_hdr;
	105	struct pnfs_ds_commit_info commit_info;
266d12d4	106	struct list_head mirrors;
d67ae825	107	struct list_head error_list; /* nfs4_ff_layout_ds_err */
1c8d477a	108	ktime_t last_report_time; /* Layoutstat report times */
d67ae825 TH	109	};
d67ae825 TH	110
5b9b3c85 TM	111	struct nfs4_flexfile_layoutreturn_args {
5b9b3c85 TM	112	struct list_head errors;
230bc962	113	struct nfs42_layoutstat_devinfo devinfo[FF_LAYOUTSTATS_MAXDEV];
5b9b3c85	114	unsigned int num_errors;
230bc962	115	unsigned int num_dev;
d9152114	116	struct page *pages[1];
5b9b3c85 TM	117	};
5b9b3c85 TM	118
d67ae825 TH	119	static inline struct nfs4_flexfile_layout *
	120	FF_LAYOUT_FROM_HDR(struct pnfs_layout_hdr *lo)
	121	{
	122	return container_of(lo, struct nfs4_flexfile_layout, generic_hdr);
	123	}
	124
	125	static inline struct nfs4_ff_layout_segment *
	126	FF_LAYOUT_LSEG(struct pnfs_layout_segment *lseg)
	127	{
	128	return container_of(lseg,
	129	struct nfs4_ff_layout_segment,
	130	generic_hdr);
	131	}
	132
	133	static inline struct nfs4_deviceid_node *
	134	FF_LAYOUT_DEVID_NODE(struct pnfs_layout_segment *lseg, u32 idx)
	135	{
	136	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt \|\|
	137	FF_LAYOUT_LSEG(lseg)->mirror_array[idx] == NULL \|\|
	138	FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds == NULL)
	139	return NULL;
	140	return &FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds->id_node;
	141	}
	142
	143	static inline struct nfs4_ff_layout_ds *
	144	FF_LAYOUT_MIRROR_DS(struct nfs4_deviceid_node *node)
	145	{
	146	return container_of(node, struct nfs4_ff_layout_ds, id_node);
	147	}
	148
	149	static inline struct nfs4_ff_layout_mirror *
	150	FF_LAYOUT_COMP(struct pnfs_layout_segment *lseg, u32 idx)
	151	{
	152	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt)
	153	return NULL;
	154	return FF_LAYOUT_LSEG(lseg)->mirror_array[idx];
	155	}
	156
	157	static inline u32
	158	FF_LAYOUT_MIRROR_COUNT(struct pnfs_layout_segment *lseg)
	159	{
	160	return FF_LAYOUT_LSEG(lseg)->mirror_array_cnt;
	161	}
	162
260074cd TM	163	static inline bool
	164	ff_layout_no_fallback_to_mds(struct pnfs_layout_segment *lseg)
	165	{
	166	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_IO_THRU_MDS;
	167	}
	168
fb1084e3 TH	169	static inline bool
	170	ff_layout_no_read_on_rw(struct pnfs_layout_segment *lseg)
	171	{
	172	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_READ_IO;
	173	}
	174
d67ae825 TH	175	static inline bool
	176	ff_layout_test_devid_unavailable(struct nfs4_deviceid_node *node)
	177	{
da066f3f WAA	178	/*
	179	* Flexfiles should never mark a DS unavailable, but if it does
	180	* print a (ratelimited) warning as this can affect performance.
	181	*/
	182	if (nfs4_test_deviceid_unavailable(node)) {
	183	u32 p = (u32 )node->deviceid.data;
	184
	185	pr_warn_ratelimited("NFS: flexfiles layout referencing an "
	186	"unavailable device [%x%x%x%x]\n",
	187	p[0], p[1], p[2], p[3]);
	188	return true;
	189	}
	190	return false;
d67ae825 TH	191	}
	192
	193	static inline int
	194	nfs4_ff_layout_ds_version(struct pnfs_layout_segment *lseg, u32 ds_idx)
	195	{
	196	return FF_LAYOUT_COMP(lseg, ds_idx)->mirror_ds->ds_versions[0].version;
	197	}
	198
	199	struct nfs4_ff_layout_ds *
	200	nfs4_ff_alloc_deviceid_node(struct nfs_server server, struct pnfs_device pdev,
	201	gfp_t gfp_flags);
	202	void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
	203	void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
	204	int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
	205	struct nfs4_ff_layout_mirror *mirror, u64 offset,
	206	u64 length, int status, enum nfs_opnum4 opnum,
	207	gfp_t gfp_flags);
5b9b3c85 TM	208	int ff_layout_encode_ds_ioerr(struct xdr_stream xdr, const struct list_head head);
	209	void ff_layout_free_ds_ioerr(struct list_head *head);
	210	unsigned int ff_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
	211	const struct pnfs_layout_range *range,
	212	struct list_head *head,
	213	unsigned int maxnum);
d67ae825 TH	214	struct nfs_fh *
	215	nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx);
	216
	217	struct nfs4_pnfs_ds *
	218	nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
	219	bool fail_return);
	220
	221	struct rpc_clnt *
	222	nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg,
	223	u32 ds_idx,
	224	struct nfs_client *ds_clp,
	225	struct inode *inode);
	226	struct rpc_cred ff_layout_get_ds_cred(struct pnfs_layout_segment lseg,
	227	u32 ds_idx, struct rpc_cred *mdscred);
3b13b4b3	228	bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg);
fb1084e3 TH	229	bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg);
fb1084e3 TH	230
d67ae825	231	#endif /* FS_NFS_NFS4FLEXFILELAYOUT_H */