md_p.h

   1 /*
   2    md_p.h : physical layout of Linux RAID devices
   3           Copyright (C) 1996-98 Ingo Molnar, Gadi Oxman
   4
   5    This program is free software; you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation; either version 2, or (at your option)
   8    any later version.
   9
  10    You should have received a copy of the GNU General Public License
  11    (for example /usr/src/linux/COPYING); if not, write to the Free
  12    Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  13 */
  14
  15 #ifndef _MD_P_H
  16 #define _MD_P_H
  17
  18 /*
  19  * RAID superblock.
  20  *
  21  * The RAID superblock maintains some statistics on each RAID configuration.
  22  * Each real device in the RAID set contains it near the end of the device.
  23  * Some of the ideas are copied from the ext2fs implementation.
  24  *
  25  * We currently use 4096 bytes as follows:
  26  *
  27  *      word offset     function
  28  *
  29  *         0  -    31   Constant generic RAID device information.
  30  *        32  -    63   Generic state information.
  31  *        64  -   127   Personality specific information.
  32  *       128  -   511   12 32-words descriptors of the disks in the raid set.
  33  *       512  -   911   Reserved.
  34  *       912  -  1023   Disk specific descriptor.
  35  */
  36
  37 /*
  38  * If x is the real device size in bytes, we return an apparent size of:
  39  *
  40  *      y = (x & ~(MD_RESERVED_BYTES - 1)) - MD_RESERVED_BYTES
  41  *
  42  * and place the 4kB superblock at offset y.
  43  */
  44 #define MD_RESERVED_BYTES               (64 * 1024)
  45 #define MD_RESERVED_SECTORS             (MD_RESERVED_BYTES / 512)
  46 #define MD_RESERVED_BLOCKS              (MD_RESERVED_BYTES / BLOCK_SIZE)
  47
  48 #define MD_NEW_SIZE_SECTORS(x)          ((x & ~(MD_RESERVED_SECTORS - 1)) - MD_RESERVED_SECTORS)
  49 #define MD_NEW_SIZE_BLOCKS(x)           ((x & ~(MD_RESERVED_BLOCKS - 1)) - MD_RESERVED_BLOCKS)
  50
  51 #define MD_SB_BYTES                     4096
  52 #define MD_SB_WORDS                     (MD_SB_BYTES / 4)
  53 #define MD_SB_BLOCKS                    (MD_SB_BYTES / BLOCK_SIZE)
  54 #define MD_SB_SECTORS                   (MD_SB_BYTES / 512)
  55
  56 /*
  57  * The following are counted in 32-bit words
  58  */
  59 #define MD_SB_GENERIC_OFFSET            0
  60 #define MD_SB_PERSONALITY_OFFSET        64
  61 #define MD_SB_DISKS_OFFSET              128
  62 #define MD_SB_DESCRIPTOR_OFFSET         992
  63
  64 #define MD_SB_GENERIC_CONSTANT_WORDS    32
  65 #define MD_SB_GENERIC_STATE_WORDS       32
  66 #define MD_SB_GENERIC_WORDS             (MD_SB_GENERIC_CONSTANT_WORDS + MD_SB_GENERIC_STATE_WORDS)
  67 #define MD_SB_PERSONALITY_WORDS         64
  68 #define MD_SB_DESCRIPTOR_WORDS          32
  69 #define MD_SB_DISKS                     27
  70 #define MD_SB_DISKS_WORDS               (MD_SB_DISKS*MD_SB_DESCRIPTOR_WORDS)
  71 #define MD_SB_RESERVED_WORDS            (1024 - MD_SB_GENERIC_WORDS - MD_SB_PERSONALITY_WORDS - MD_SB_DISKS_WORDS - MD_SB_DESCRIPTOR_WORDS)
  72 #define MD_SB_EQUAL_WORDS               (MD_SB_GENERIC_WORDS + MD_SB_PERSONALITY_WORDS + MD_SB_DISKS_WORDS)
  73
  74 /*
  75  * Device "operational" state bits
  76  */
  77 #define MD_DISK_FAULTY          0 /* disk is faulty / operational */
  78 #define MD_DISK_ACTIVE          1 /* disk is running but may not be in sync */
  79 #define MD_DISK_SYNC            2 /* disk is in sync with the raid set */
  80 #define MD_DISK_REMOVED         3 /* disk is in sync with the raid set */
  81 #define MD_DISK_CLUSTER_ADD     4 /* Initiate a disk add across the cluster
  82                                    * For clustered enviroments only.
  83                                    */
  84 #define MD_DISK_CANDIDATE       5 /* disk is added as spare (local) until confirmed
  85                                    * For clustered enviroments only.
  86                                    */
  87
  88 #define MD_DISK_WRITEMOSTLY     9 /* disk is "write-mostly" is RAID1 config.
  89                                    * read requests will only be sent here in
  90                                    * dire need
  91                                    */
  92 #define MD_DISK_FAILFAST        10 /* Fewer retries, more failures */
  93
  94 #define MD_DISK_REPLACEMENT     17
  95 #define MD_DISK_JOURNAL         18 /* disk is used as the write journal in RAID-5/6 */
  96
  97 #define MD_DISK_ROLE_SPARE      0xffff
  98 #define MD_DISK_ROLE_FAULTY     0xfffe
  99 #define MD_DISK_ROLE_JOURNAL    0xfffd
 100 #define MD_DISK_ROLE_MAX        0xff00 /* max value of regular disk role */
 101
 102 typedef struct mdp_device_descriptor_s {
 103         __u32 number;           /* 0 Device number in the entire set          */
 104         __u32 major;            /* 1 Device major number                      */
 105         __u32 minor;            /* 2 Device minor number                      */
 106         __u32 raid_disk;        /* 3 The role of the device in the raid set   */
 107         __u32 state;            /* 4 Operational state                        */
 108         __u32 reserved[MD_SB_DESCRIPTOR_WORDS - 5];
 109 } mdp_disk_t;
 110
 111 #define MD_SB_MAGIC             0xa92b4efc
 112
 113 /*
 114  * Superblock state bits
 115  */
 116 #define MD_SB_CLEAN             0
 117 #define MD_SB_ERRORS            1
 118 #define MD_SB_BBM_ERRORS        2
 119 #define MD_SB_BLOCK_CONTAINER_RESHAPE 3 /* block container wide reshapes */
 120 #define MD_SB_BLOCK_VOLUME      4 /* block activation of array, other arrays
 121                                    * in container can be activated */
 122 #define MD_SB_CLUSTERED         5 /* MD is clustered  */
 123 #define MD_SB_BITMAP_PRESENT    8 /* bitmap may be present nearby */
 124
 125 typedef struct mdp_superblock_s {
 126         /*
 127          * Constant generic information
 128          */
 129         __u32 md_magic;         /*  0 MD identifier                           */
 130         __u32 major_version;    /*  1 major version to which the set conforms */
 131         __u32 minor_version;    /*  2 minor version ...                       */
 132         __u32 patch_version;    /*  3 patchlevel version ...                  */
 133         __u32 gvalid_words;     /*  4 Number of used words in this section    */
 134         __u32 set_uuid0;        /*  5 Raid set identifier                     */
 135         __u32 ctime;            /*  6 Creation time                           */
 136         __u32 level;            /*  7 Raid personality                        */
 137         __u32 size;             /*  8 Apparent size of each individual disk   */
 138         __u32 nr_disks;         /*  9 total disks in the raid set             */
 139         __u32 raid_disks;       /* 10 disks in a fully functional raid set    */
 140         __u32 md_minor;         /* 11 preferred MD minor device number        */
 141         __u32 not_persistent;   /* 12 does it have a persistent superblock    */
 142         __u32 set_uuid1;        /* 13 Raid set identifier #2                  */
 143         __u32 set_uuid2;        /* 14 Raid set identifier #3                  */
 144         __u32 set_uuid3;        /* 15 Raid set identifier #4                  */
 145         __u32 gstate_creserved[MD_SB_GENERIC_CONSTANT_WORDS - 16];
 146
 147         /*
 148          * Generic state information
 149          */
 150         __u32 utime;            /*  0 Superblock update time                  */
 151         __u32 state;            /*  1 State bits (clean, ...)                 */
 152         __u32 active_disks;     /*  2 Number of currently active disks        */
 153         __u32 working_disks;    /*  3 Number of working disks                 */
 154         __u32 failed_disks;     /*  4 Number of failed disks                  */
 155         __u32 spare_disks;      /*  5 Number of spare disks                   */
 156         __u32 sb_csum;          /*  6 checksum of the whole superblock        */
 157 #if  __BYTE_ORDER ==  __BIG_ENDIAN
 158         __u32 events_hi;        /*  7 high-order of superblock update count   */
 159         __u32 events_lo;        /*  8 low-order of superblock update count    */
 160         __u32 cp_events_hi;     /*  9 high-order of checkpoint update count   */
 161         __u32 cp_events_lo;     /* 10 low-order of checkpoint update count    */
 162 #else
 163         __u32 events_lo;        /*  7 low-order of superblock update count    */
 164         __u32 events_hi;        /*  8 high-order of superblock update count   */
 165         __u32 cp_events_lo;     /*  9 low-order of checkpoint update count    */
 166         __u32 cp_events_hi;     /* 10 high-order of checkpoint update count   */
 167 #endif
 168         __u32 recovery_cp;      /* 11 recovery checkpoint sector count        */
 169         /* There are only valid for minor_version > 90 */
 170         __u64 reshape_position; /* 12,13 next address in array-space for reshape */
 171         __u32 new_level;        /* 14 new level we are reshaping to           */
 172         __u32 delta_disks;      /* 15 change in number of raid_disks          */
 173         __u32 new_layout;       /* 16 new layout                              */
 174         __u32 new_chunk;        /* 17 new chunk size (bytes)                  */
 175         __u32 gstate_sreserved[MD_SB_GENERIC_STATE_WORDS - 18];
 176
 177         /*
 178          * Personality information
 179          */
 180         __u32 layout;           /*  0 the array's physical layout             */
 181         __u32 chunk_size;       /*  1 chunk size in bytes                     */
 182         __u32 root_pv;          /*  2 LV root PV */
 183         __u32 root_block;       /*  3 LV root block */
 184         __u32 pstate_reserved[MD_SB_PERSONALITY_WORDS - 4];
 185
 186         /*
 187          * Disks information
 188          */
 189         mdp_disk_t disks[MD_SB_DISKS];
 190
 191         /*
 192          * Reserved
 193          */
 194         __u32 reserved[MD_SB_RESERVED_WORDS];
 195
 196         /*
 197          * Active descriptor
 198          */
 199         mdp_disk_t this_disk;
 200
 201 } mdp_super_t;
 202
 203 #ifdef __TINYC__
 204 typedef unsigned long long __u64;
 205 #endif
 206
 207 static inline __u64 md_event(mdp_super_t *sb) {
 208         __u64 ev = sb->events_hi;
 209         return (ev<<32)| sb->events_lo;
 210 }
 211
 212 struct r5l_payload_header {
 213         __u16 type;
 214         __u16 flags;
 215 } __attribute__ ((__packed__));
 216
 217 enum r5l_payload_type {
 218         R5LOG_PAYLOAD_DATA = 0,
 219         R5LOG_PAYLOAD_PARITY = 1,
 220         R5LOG_PAYLOAD_FLUSH = 2,
 221 };
 222
 223 struct r5l_payload_data_parity {
 224         struct r5l_payload_header header;
 225         __u32 size; /* sector. data/parity size. each 4k has a checksum */
 226         __u64 location; /* sector. For data, it's raid sector. For
 227                                 parity, it's stripe sector */
 228         __u32 checksum[];
 229 } __attribute__ ((__packed__));
 230
 231 enum r5l_payload_data_parity_flag {
 232         R5LOG_PAYLOAD_FLAG_DISCARD = 1, /* payload is discard */
 233         /*
 234          * RESHAPED/RESHAPING is only set when there is reshape activity. Note,
 235          * both data/parity of a stripe should have the same flag set
 236          *
 237          * RESHAPED: reshape is running, and this stripe finished reshape
 238          * RESHAPING: reshape is running, and this stripe isn't reshaped
 239          * */
 240         R5LOG_PAYLOAD_FLAG_RESHAPED = 2,
 241         R5LOG_PAYLOAD_FLAG_RESHAPING = 3,
 242 };
 243
 244 struct r5l_payload_flush {
 245         struct r5l_payload_header header;
 246         __u32 size; /* flush_stripes size, bytes */
 247         __u64 flush_stripes[];
 248 } __attribute__ ((__packed__));
 249
 250 enum r5l_payload_flush_flag {
 251         R5LOG_PAYLOAD_FLAG_FLUSH_STRIPE = 1, /* data represents whole stripe */
 252 };
 253
 254 struct r5l_meta_block {
 255         __u32 magic;
 256         __u32 checksum;
 257         __u8 version;
 258         __u8 __zero_pading_1;
 259         __u16 __zero_pading_2;
 260         __u32 meta_size; /* whole size of the block */
 261
 262         __u64 seq;
 263         __u64 position; /* sector, start from rdev->data_offset, current position */
 264         struct r5l_payload_header payloads[];
 265 } __attribute__ ((__packed__));
 266
 267 #define R5LOG_VERSION 0x1
 268 #define R5LOG_MAGIC 0x6433c509
 269
 270 struct ppl_header_entry {
 271         __u64 data_sector;      /* raid sector of the new data */
 272         __u32 pp_size;          /* length of partial parity */
 273         __u32 data_size;        /* length of data */
 274         __u32 parity_disk;      /* member disk containing parity */
 275         __u32 checksum;         /* checksum of this entry's partial parity */
 276 } __attribute__ ((__packed__));
 277
 278 #define PPL_HEADER_SIZE 4096
 279 #define PPL_HDR_RESERVED 512
 280 #define PPL_HDR_ENTRY_SPACE \
 281         (PPL_HEADER_SIZE - PPL_HDR_RESERVED - 4 * sizeof(__u32) - sizeof(__u64))
 282 #define PPL_HDR_MAX_ENTRIES \
 283         (PPL_HDR_ENTRY_SPACE / sizeof(struct ppl_header_entry))
 284
 285 struct ppl_header {
 286         __u8 reserved[PPL_HDR_RESERVED];/* reserved space, fill with 0xff */
 287         __u32 signature;                /* signature (family number of volume) */
 288         __u32 padding;                  /* zero pad */
 289         __u64 generation;               /* generation number of the header */
 290         __u32 entries_count;            /* number of entries in entry array */
 291         __u32 checksum;                 /* checksum of the header */
 292         struct ppl_header_entry entries[PPL_HDR_MAX_ENTRIES];
 293 } __attribute__ ((__packed__));
 294
 295 #endif