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

/* SPDX-License-Identifier: GPL-2.0 */
/*
 * NFSv4 flexfile layout driver data structures.
 *
 * Copyright (c) 2014, Primary Data, Inc. All rights reserved.
 *
 * Tao Peng <[email protected]>
 */

#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 <linux/refcount.h>
#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;
	refcount_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);
int
nfs4_ff_layout_select_ds_stateid(struct pnfs_layout_segment *lseg,
				u32 mirror_idx,
				nfs4_stateid *stateid);

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
b2441318	1	/* SPDX-License-Identifier: GPL-2.0 */
d67ae825 TH	2	/*
	3	* NFSv4 flexfile layout driver data structures.
	4	*
	5	* Copyright (c) 2014, Primary Data, Inc. All rights reserved.
	6	*
	7	* Tao Peng <[email protected]>
	8	*/
	9
	10	#ifndef FS_NFS_NFS4FLEXFILELAYOUT_H
	11	#define FS_NFS_NFS4FLEXFILELAYOUT_H
	12
c0f5f505	13	#define FF_FLAGS_NO_LAYOUTCOMMIT 1
fb1084e3 TH	14	#define FF_FLAGS_NO_IO_THRU_MDS 2
fb1084e3 TH	15	#define FF_FLAGS_NO_READ_IO 4
c0f5f505	16
81a090b9	17	#include <linux/refcount.h>
d67ae825 TH	18	#include "../pnfs.h"
	19
	20	/* XXX: Let's filter out insanely large mirror count for now to avoid oom
	21	* due to network error etc. */
	22	#define NFS4_FLEXFILE_LAYOUT_MAX_MIRROR_CNT 4096
	23
97ba375b PT	24	/* LAYOUTSTATS report interval in ms */
97ba375b PT	25	#define FF_LAYOUTSTATS_REPORT_INTERVAL (60000L)
230bc962	26	#define FF_LAYOUTSTATS_MAXDEV 4
97ba375b	27
d67ae825 TH	28	struct nfs4_ff_ds_version {
	29	u32 version;
	30	u32 minor_version;
	31	u32 rsize;
	32	u32 wsize;
	33	bool tightly_coupled;
	34	};
	35
	36	/* chained in global deviceid hlist */
	37	struct nfs4_ff_layout_ds {
	38	struct nfs4_deviceid_node id_node;
	39	u32 ds_versions_cnt;
	40	struct nfs4_ff_ds_version *ds_versions;
	41	struct nfs4_pnfs_ds *ds;
	42	};
	43
	44	struct nfs4_ff_layout_ds_err {
	45	struct list_head list; /* linked in mirror error_list */
	46	u64 offset;
	47	u64 length;
	48	int status;
	49	enum nfs_opnum4 opnum;
	50	nfs4_stateid stateid;
	51	struct nfs4_deviceid deviceid;
	52	};
	53
abcb7bfc TM	54	struct nfs4_ff_io_stat {
	55	__u64 ops_requested;
	56	__u64 bytes_requested;
	57	__u64 ops_completed;
	58	__u64 bytes_completed;
	59	__u64 bytes_not_delivered;
	60	ktime_t total_busy_time;
	61	ktime_t aggregate_completion_time;
	62	};
	63
	64	struct nfs4_ff_busy_timer {
	65	ktime_t start_time;
	66	atomic_t n_ops;
	67	};
	68
	69	struct nfs4_ff_layoutstat {
	70	struct nfs4_ff_io_stat io_stat;
	71	struct nfs4_ff_busy_timer busy_timer;
	72	};
	73
d67ae825	74	struct nfs4_ff_layout_mirror {
266d12d4 TM	75	struct pnfs_layout_hdr *layout;
266d12d4 TM	76	struct list_head mirrors;
d67ae825 TH	77	u32 ds_count;
d67ae825 TH	78	u32 efficiency;
65990d1a	79	struct nfs4_deviceid devid;
d67ae825 TH	80	struct nfs4_ff_layout_ds *mirror_ds;
	81	u32 fh_versions_cnt;
	82	struct nfs_fh *fh_versions;
	83	nfs4_stateid stateid;
3064b686 JL	84	struct rpc_cred __rcu *ro_cred;
3064b686 JL	85	struct rpc_cred __rcu *rw_cred;
81a090b9	86	refcount_t ref;
d67ae825	87	spinlock_t lock;
06946c6a	88	unsigned long flags;
abcb7bfc TM	89	struct nfs4_ff_layoutstat read_stat;
abcb7bfc TM	90	struct nfs4_ff_layoutstat write_stat;
d983803d	91	ktime_t start_time;
d0379a5d	92	u32 report_interval;
d67ae825 TH	93	};
d67ae825 TH	94
06946c6a TM	95	#define NFS4_FF_MIRROR_STAT_AVAIL (0)
06946c6a TM	96
d67ae825 TH	97	struct nfs4_ff_layout_segment {
	98	struct pnfs_layout_segment generic_hdr;
	99	u64 stripe_unit;
c0f5f505	100	u32 flags;
d67ae825 TH	101	u32 mirror_array_cnt;
	102	struct nfs4_ff_layout_mirror **mirror_array;
	103	};
	104
	105	struct nfs4_flexfile_layout {
	106	struct pnfs_layout_hdr generic_hdr;
	107	struct pnfs_ds_commit_info commit_info;
266d12d4	108	struct list_head mirrors;
d67ae825	109	struct list_head error_list; /* nfs4_ff_layout_ds_err */
1c8d477a	110	ktime_t last_report_time; /* Layoutstat report times */
d67ae825 TH	111	};
d67ae825 TH	112
5b9b3c85 TM	113	struct nfs4_flexfile_layoutreturn_args {
5b9b3c85 TM	114	struct list_head errors;
230bc962	115	struct nfs42_layoutstat_devinfo devinfo[FF_LAYOUTSTATS_MAXDEV];
5b9b3c85	116	unsigned int num_errors;
230bc962	117	unsigned int num_dev;
d9152114	118	struct page *pages[1];
5b9b3c85 TM	119	};
5b9b3c85 TM	120
d67ae825 TH	121	static inline struct nfs4_flexfile_layout *
	122	FF_LAYOUT_FROM_HDR(struct pnfs_layout_hdr *lo)
	123	{
	124	return container_of(lo, struct nfs4_flexfile_layout, generic_hdr);
	125	}
	126
	127	static inline struct nfs4_ff_layout_segment *
	128	FF_LAYOUT_LSEG(struct pnfs_layout_segment *lseg)
	129	{
	130	return container_of(lseg,
	131	struct nfs4_ff_layout_segment,
	132	generic_hdr);
	133	}
	134
	135	static inline struct nfs4_deviceid_node *
	136	FF_LAYOUT_DEVID_NODE(struct pnfs_layout_segment *lseg, u32 idx)
	137	{
	138	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt \|\|
	139	FF_LAYOUT_LSEG(lseg)->mirror_array[idx] == NULL \|\|
	140	FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds == NULL)
	141	return NULL;
	142	return &FF_LAYOUT_LSEG(lseg)->mirror_array[idx]->mirror_ds->id_node;
	143	}
	144
	145	static inline struct nfs4_ff_layout_ds *
	146	FF_LAYOUT_MIRROR_DS(struct nfs4_deviceid_node *node)
	147	{
	148	return container_of(node, struct nfs4_ff_layout_ds, id_node);
	149	}
	150
	151	static inline struct nfs4_ff_layout_mirror *
	152	FF_LAYOUT_COMP(struct pnfs_layout_segment *lseg, u32 idx)
	153	{
	154	if (idx >= FF_LAYOUT_LSEG(lseg)->mirror_array_cnt)
	155	return NULL;
	156	return FF_LAYOUT_LSEG(lseg)->mirror_array[idx];
	157	}
	158
	159	static inline u32
	160	FF_LAYOUT_MIRROR_COUNT(struct pnfs_layout_segment *lseg)
	161	{
	162	return FF_LAYOUT_LSEG(lseg)->mirror_array_cnt;
	163	}
	164
260074cd TM	165	static inline bool
	166	ff_layout_no_fallback_to_mds(struct pnfs_layout_segment *lseg)
	167	{
	168	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_IO_THRU_MDS;
	169	}
	170
fb1084e3 TH	171	static inline bool
	172	ff_layout_no_read_on_rw(struct pnfs_layout_segment *lseg)
	173	{
	174	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_READ_IO;
	175	}
	176
d67ae825 TH	177	static inline bool
	178	ff_layout_test_devid_unavailable(struct nfs4_deviceid_node *node)
	179	{
da066f3f WAA	180	/*
	181	* Flexfiles should never mark a DS unavailable, but if it does
	182	* print a (ratelimited) warning as this can affect performance.
	183	*/
	184	if (nfs4_test_deviceid_unavailable(node)) {
	185	u32 p = (u32 )node->deviceid.data;
	186
	187	pr_warn_ratelimited("NFS: flexfiles layout referencing an "
	188	"unavailable device [%x%x%x%x]\n",
	189	p[0], p[1], p[2], p[3]);
	190	return true;
	191	}
	192	return false;
d67ae825 TH	193	}
	194
	195	static inline int
	196	nfs4_ff_layout_ds_version(struct pnfs_layout_segment *lseg, u32 ds_idx)
	197	{
	198	return FF_LAYOUT_COMP(lseg, ds_idx)->mirror_ds->ds_versions[0].version;
	199	}
	200
	201	struct nfs4_ff_layout_ds *
	202	nfs4_ff_alloc_deviceid_node(struct nfs_server server, struct pnfs_device pdev,
	203	gfp_t gfp_flags);
	204	void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
	205	void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds);
	206	int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
	207	struct nfs4_ff_layout_mirror *mirror, u64 offset,
	208	u64 length, int status, enum nfs_opnum4 opnum,
	209	gfp_t gfp_flags);
5b9b3c85 TM	210	int ff_layout_encode_ds_ioerr(struct xdr_stream xdr, const struct list_head head);
	211	void ff_layout_free_ds_ioerr(struct list_head *head);
	212	unsigned int ff_layout_fetch_ds_ioerr(struct pnfs_layout_hdr *lo,
	213	const struct pnfs_layout_range *range,
	214	struct list_head *head,
	215	unsigned int maxnum);
d67ae825 TH	216	struct nfs_fh *
d67ae825 TH	217	nfs4_ff_layout_select_ds_fh(struct pnfs_layout_segment *lseg, u32 mirror_idx);
bb21ce0a TM	218	int
	219	nfs4_ff_layout_select_ds_stateid(struct pnfs_layout_segment *lseg,
	220	u32 mirror_idx,
	221	nfs4_stateid *stateid);
d67ae825 TH	222
	223	struct nfs4_pnfs_ds *
	224	nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
	225	bool fail_return);
	226
	227	struct rpc_clnt *
	228	nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg,
	229	u32 ds_idx,
	230	struct nfs_client *ds_clp,
	231	struct inode *inode);
	232	struct rpc_cred ff_layout_get_ds_cred(struct pnfs_layout_segment lseg,
	233	u32 ds_idx, struct rpc_cred *mdscred);
3b13b4b3	234	bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg);
fb1084e3 TH	235	bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg);
fb1084e3 TH	236
d67ae825	237	#endif /* FS_NFS_NFS4FLEXFILELAYOUT_H */