Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
tjh.dev
kernel
os
linux
1/*
2 * linux/fs/nfs/blocklayout/blocklayout.h
3 *
4 * Module for the NFSv4.1 pNFS block layout driver.
5 *
6 * Copyright (c) 2006 The Regents of the University of Michigan.
7 * All rights reserved.
8 *
9 * Andy Adamson <andros@citi.umich.edu>
10 * Fred Isaman <iisaman@umich.edu>
11 *
12 * permission is granted to use, copy, create derivative works and
13 * redistribute this software and such derivative works for any purpose,
14 * so long as the name of the university of michigan is not used in
15 * any advertising or publicity pertaining to the use or distribution
16 * of this software without specific, written prior authorization. if
17 * the above copyright notice or any other identification of the
18 * university of michigan is included in any copy of any portion of
19 * this software, then the disclaimer below must also be included.
20 *
21 * this software is provided as is, without representation from the
22 * university of michigan as to its fitness for any purpose, and without
23 * warranty by the university of michigan of any kind, either express
24 * or implied, including without limitation the implied warranties of
25 * merchantability and fitness for a particular purpose. the regents
26 * of the university of michigan shall not be liable for any damages,
27 * including special, indirect, incidental, or consequential damages,
28 * with respect to any claim arising out or in connection with the use
29 * of the software, even if it has been or is hereafter advised of the
30 * possibility of such damages.
31 */
32#ifndef FS_NFS_NFS4BLOCKLAYOUT_H
33#define FS_NFS_NFS4BLOCKLAYOUT_H
34
35#include <linux/device-mapper.h>
36#include <linux/nfs_fs.h>
37#include <linux/sunrpc/rpc_pipe_fs.h>
38
39#include "../pnfs.h"
40#include "../netns.h"
41
42#define PAGE_CACHE_SECTORS (PAGE_CACHE_SIZE >> SECTOR_SHIFT)
43#define PAGE_CACHE_SECTOR_SHIFT (PAGE_CACHE_SHIFT - SECTOR_SHIFT)
44#define SECTOR_SIZE (1 << SECTOR_SHIFT)
45
46struct block_mount_id {
47 spinlock_t bm_lock; /* protects list */
48 struct list_head bm_devlist; /* holds pnfs_block_dev */
49};
50
51struct pnfs_block_dev {
52 struct list_head bm_node;
53 struct nfs4_deviceid bm_mdevid; /* associated devid */
54 struct block_device *bm_mdev; /* meta device itself */
55 struct net *net;
56};
57
58enum exstate4 {
59 PNFS_BLOCK_READWRITE_DATA = 0,
60 PNFS_BLOCK_READ_DATA = 1,
61 PNFS_BLOCK_INVALID_DATA = 2, /* mapped, but data is invalid */
62 PNFS_BLOCK_NONE_DATA = 3 /* unmapped, it's a hole */
63};
64
65#define MY_MAX_TAGS (15) /* tag bitnums used must be less than this */
66
67struct my_tree {
68 sector_t mtt_step_size; /* Internal sector alignment */
69 struct list_head mtt_stub; /* Should be a radix tree */
70};
71
72struct pnfs_inval_markings {
73 spinlock_t im_lock;
74 struct my_tree im_tree; /* Sectors that need LAYOUTCOMMIT */
75 sector_t im_block_size; /* Server blocksize in sectors */
76 struct list_head im_extents; /* Short extents for INVAL->RW conversion */
77};
78
79struct pnfs_inval_tracking {
80 struct list_head it_link;
81 int it_sector;
82 int it_tags;
83};
84
85/* sector_t fields are all in 512-byte sectors */
86struct pnfs_block_extent {
87 struct kref be_refcnt;
88 struct list_head be_node; /* link into lseg list */
89 struct nfs4_deviceid be_devid; /* FIXME: could use device cache instead */
90 struct block_device *be_mdev;
91 sector_t be_f_offset; /* the starting offset in the file */
92 sector_t be_length; /* the size of the extent */
93 sector_t be_v_offset; /* the starting offset in the volume */
94 enum exstate4 be_state; /* the state of this extent */
95 struct pnfs_inval_markings *be_inval; /* tracks INVAL->RW transition */
96};
97
98/* Shortened extent used by LAYOUTCOMMIT */
99struct pnfs_block_short_extent {
100 struct list_head bse_node;
101 struct nfs4_deviceid bse_devid;
102 struct block_device *bse_mdev;
103 sector_t bse_f_offset; /* the starting offset in the file */
104 sector_t bse_length; /* the size of the extent */
105};
106
107static inline void
108BL_INIT_INVAL_MARKS(struct pnfs_inval_markings *marks, sector_t blocksize)
109{
110 spin_lock_init(&marks->im_lock);
111 INIT_LIST_HEAD(&marks->im_tree.mtt_stub);
112 INIT_LIST_HEAD(&marks->im_extents);
113 marks->im_block_size = blocksize;
114 marks->im_tree.mtt_step_size = min((sector_t)PAGE_CACHE_SECTORS,
115 blocksize);
116}
117
118enum extentclass4 {
119 RW_EXTENT = 0, /* READWRTE and INVAL */
120 RO_EXTENT = 1, /* READ and NONE */
121 EXTENT_LISTS = 2,
122};
123
124static inline int bl_choose_list(enum exstate4 state)
125{
126 if (state == PNFS_BLOCK_READ_DATA || state == PNFS_BLOCK_NONE_DATA)
127 return RO_EXTENT;
128 else
129 return RW_EXTENT;
130}
131
132struct pnfs_block_layout {
133 struct pnfs_layout_hdr bl_layout;
134 struct pnfs_inval_markings bl_inval; /* tracks INVAL->RW transition */
135 spinlock_t bl_ext_lock; /* Protects list manipulation */
136 struct list_head bl_extents[EXTENT_LISTS]; /* R and RW extents */
137 struct list_head bl_commit; /* Needs layout commit */
138 struct list_head bl_committing; /* Layout committing */
139 unsigned int bl_count; /* entries in bl_commit */
140 sector_t bl_blocksize; /* Server blocksize in sectors */
141};
142
143#define BLK_ID(lo) ((struct block_mount_id *)(NFS_SERVER(lo->plh_inode)->pnfs_ld_data))
144
145static inline struct pnfs_block_layout *
146BLK_LO2EXT(struct pnfs_layout_hdr *lo)
147{
148 return container_of(lo, struct pnfs_block_layout, bl_layout);
149}
150
151static inline struct pnfs_block_layout *
152BLK_LSEG2EXT(struct pnfs_layout_segment *lseg)
153{
154 return BLK_LO2EXT(lseg->pls_layout);
155}
156
157struct bl_pipe_msg {
158 struct rpc_pipe_msg msg;
159 wait_queue_head_t *bl_wq;
160};
161
162struct bl_msg_hdr {
163 u8 type;
164 u16 totallen; /* length of entire message, including hdr itself */
165};
166
167#define BL_DEVICE_UMOUNT 0x0 /* Umount--delete devices */
168#define BL_DEVICE_MOUNT 0x1 /* Mount--create devices*/
169#define BL_DEVICE_REQUEST_INIT 0x0 /* Start request */
170#define BL_DEVICE_REQUEST_PROC 0x1 /* User level process succeeds */
171#define BL_DEVICE_REQUEST_ERR 0x2 /* User level process fails */
172
173/* blocklayoutdev.c */
174ssize_t bl_pipe_downcall(struct file *, const char __user *, size_t);
175void bl_pipe_destroy_msg(struct rpc_pipe_msg *);
176void nfs4_blkdev_put(struct block_device *bdev);
177struct pnfs_block_dev *nfs4_blk_decode_device(struct nfs_server *server,
178 struct pnfs_device *dev);
179int nfs4_blk_process_layoutget(struct pnfs_layout_hdr *lo,
180 struct nfs4_layoutget_res *lgr, gfp_t gfp_flags);
181
182/* blocklayoutdm.c */
183void bl_free_block_dev(struct pnfs_block_dev *bdev);
184
185/* extents.c */
186struct pnfs_block_extent *
187bl_find_get_extent(struct pnfs_block_layout *bl, sector_t isect,
188 struct pnfs_block_extent **cow_read);
189int bl_mark_sectors_init(struct pnfs_inval_markings *marks,
190 sector_t offset, sector_t length);
191void bl_put_extent(struct pnfs_block_extent *be);
192struct pnfs_block_extent *bl_alloc_extent(void);
193int bl_is_sector_init(struct pnfs_inval_markings *marks, sector_t isect);
194int encode_pnfs_block_layoutupdate(struct pnfs_block_layout *bl,
195 struct xdr_stream *xdr,
196 const struct nfs4_layoutcommit_args *arg);
197void clean_pnfs_block_layoutupdate(struct pnfs_block_layout *bl,
198 const struct nfs4_layoutcommit_args *arg,
199 int status);
200int bl_add_merge_extent(struct pnfs_block_layout *bl,
201 struct pnfs_block_extent *new);
202int bl_mark_for_commit(struct pnfs_block_extent *be,
203 sector_t offset, sector_t length,
204 struct pnfs_block_short_extent *new);
205int bl_push_one_short_extent(struct pnfs_inval_markings *marks);
206struct pnfs_block_short_extent *
207bl_pop_one_short_extent(struct pnfs_inval_markings *marks);
208void bl_free_short_extents(struct pnfs_inval_markings *marks, int num_to_free);
209
210#endif /* FS_NFS_NFS4BLOCKLAYOUT_H */