sys/vfs/hammer/hammer_undo.c

   1 /*
   2  * Copyright (c) 2008 The DragonFly Project.  All rights reserved.
   3  *
   4  * This code is derived from software contributed to The DragonFly Project
   5  * by Matthew Dillon <dillon@backplane.com>
   6  *
   7  * Redistribution and use in source and binary forms, with or without
   8  * modification, are permitted provided that the following conditions
   9  * are met:
  10  *
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in
  15  *    the documentation and/or other materials provided with the
  16  *    distribution.
  17  * 3. Neither the name of The DragonFly Project nor the names of its
  18  *    contributors may be used to endorse or promote products derived
  19  *    from this software without specific, prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
  24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
  25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
  27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
  29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
  30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
  31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  32  * SUCH DAMAGE.
  33  *
  34  * $DragonFly: src/sys/vfs/hammer/hammer_undo.c,v 1.17 2008/06/17 04:02:38 dillon Exp $
  35  */
  36
  37 /*
  38  * HAMMER undo - undo buffer/FIFO management.
  39  */
  40
  41 #include "hammer.h"
  42
  43 static int hammer_und_rb_compare(hammer_undo_t node1, hammer_undo_t node2);
  44
  45 RB_GENERATE2(hammer_und_rb_tree, hammer_undo, rb_node,
  46              hammer_und_rb_compare, hammer_off_t, offset);
  47
  48 /*
  49  * Convert a zone-3 undo offset into a zone-2 buffer offset.
  50  */
  51 hammer_off_t
  52 hammer_undo_lookup(hammer_mount_t hmp, hammer_off_t zone3_off, int *errorp)
  53 {
  54         hammer_volume_t root_volume;
  55         hammer_blockmap_t undomap;
  56         hammer_off_t result_offset;
  57         int i;
  58
  59         KKASSERT((zone3_off & HAMMER_OFF_ZONE_MASK) == HAMMER_ZONE_UNDO);
  60         root_volume = hammer_get_root_volume(hmp, errorp);
  61         if (*errorp)
  62                 return(0);
  63         undomap = &hmp->blockmap[HAMMER_ZONE_UNDO_INDEX];
  64         KKASSERT(HAMMER_ZONE_DECODE(undomap->alloc_offset) == HAMMER_ZONE_UNDO_INDEX);
  65         KKASSERT (zone3_off < undomap->alloc_offset);
  66
  67         i = (zone3_off & HAMMER_OFF_SHORT_MASK) / HAMMER_LARGEBLOCK_SIZE;
  68         result_offset = root_volume->ondisk->vol0_undo_array[i] +
  69                         (zone3_off & HAMMER_LARGEBLOCK_MASK64);
  70
  71         hammer_rel_volume(root_volume, 0);
  72         return(result_offset);
  73 }
  74
  75 /*
  76  * Generate an UNDO record for the block of data at the specified zone1
  77  * or zone2 offset.
  78  *
  79  * The recovery code will execute UNDOs in reverse order, allowing overlaps.
  80  * All the UNDOs are executed together so if we already laid one down we
  81  * do not have to lay another one down for the same range.
  82  */
  83 int
  84 hammer_generate_undo(hammer_transaction_t trans, hammer_io_t io,
  85                      hammer_off_t zone_off, void *base, int len)
  86 {
  87         hammer_mount_t hmp;
  88         hammer_volume_t root_volume;
  89         hammer_blockmap_t undomap;
  90         hammer_buffer_t buffer = NULL;
  91         hammer_fifo_undo_t undo;
  92         hammer_fifo_tail_t tail;
  93         hammer_off_t next_offset;
  94         int error;
  95         int bytes;
  96
  97         hmp = trans->hmp;
  98
  99         /*
 100          * Enter the offset into our undo history.  If there is an existing
 101          * undo we do not have to generate a new one.
 102          */
 103         if (hammer_enter_undo_history(hmp, zone_off, len) == EALREADY)
 104                 return(0);
 105
 106         root_volume = trans->rootvol;
 107         undomap = &hmp->blockmap[HAMMER_ZONE_UNDO_INDEX];
 108
 109         /* no undo recursion */
 110         hammer_modify_volume(NULL, root_volume, NULL, 0);
 111
 112         hammer_lock_ex(&hmp->undo_lock);
 113 again:
 114         /*
 115          * Allocate space in the FIFO
 116          */
 117         bytes = ((len + HAMMER_HEAD_ALIGN_MASK) & ~HAMMER_HEAD_ALIGN_MASK) +
 118                 sizeof(struct hammer_fifo_undo) +
 119                 sizeof(struct hammer_fifo_tail);
 120         if (hammer_undo_space(hmp) < bytes + HAMMER_BUFSIZE*2)
 121                 panic("hammer: insufficient undo FIFO space!");
 122
 123         next_offset = undomap->next_offset;
 124
 125         /*
 126          * Wrap next_offset
 127          */
 128         if (undomap->next_offset == undomap->alloc_offset) {
 129                 next_offset = HAMMER_ZONE_ENCODE(HAMMER_ZONE_UNDO_INDEX, 0);
 130                 undomap->next_offset = next_offset;
 131                 hkprintf("undo zone's next_offset wrapped\n");
 132         }
 133
 134         /*
 135          * This is a tail-chasing FIFO, when we hit the start of a new
 136          * buffer we don't have to read it in.
 137          */
 138         if ((next_offset & HAMMER_BUFMASK) == 0)
 139                 undo = hammer_bnew(hmp, next_offset, &error, &buffer);
 140         else
 141                 undo = hammer_bread(hmp, next_offset, &error, &buffer);
 142         hammer_modify_buffer(NULL, buffer, NULL, 0);
 143
 144         KKASSERT(undomap->next_offset == next_offset);
 145
 146         /*
 147          * The FIFO entry would cross a buffer boundary, PAD to the end
 148          * of the buffer and try again.  Due to our data alignment, the
 149          * worst case (smallest) PAD record is 8 bytes.  PAD records only
 150          * populate the first 8 bytes of hammer_fifo_head and the tail may
 151          * be at the same offset as the head.
 152          */
 153         if ((next_offset ^ (next_offset + bytes)) & ~HAMMER_BUFMASK64) {
 154                 bytes = HAMMER_BUFSIZE - ((int)next_offset & HAMMER_BUFMASK);
 155                 tail = (void *)((char *)undo + bytes - sizeof(*tail));
 156                 if ((void *)undo != (void *)tail) {
 157                         tail->tail_signature = HAMMER_TAIL_SIGNATURE;
 158                         tail->tail_type = HAMMER_HEAD_TYPE_PAD;
 159                         tail->tail_size = bytes;
 160                 }
 161                 undo->head.hdr_signature = HAMMER_HEAD_SIGNATURE;
 162                 undo->head.hdr_type = HAMMER_HEAD_TYPE_PAD;
 163                 undo->head.hdr_size = bytes;
 164                 /* NO CRC */
 165                 undomap->next_offset += bytes;
 166                 hammer_modify_buffer_done(buffer);
 167                 goto again;
 168         }
 169         if (hammer_debug_general & 0x0080)
 170                 kprintf("undo %016llx %d %d\n", next_offset, bytes, len);
 171
 172         /*
 173          * We're good, create the entry.
 174          */
 175         undo->head.hdr_signature = HAMMER_HEAD_SIGNATURE;
 176         undo->head.hdr_type = HAMMER_HEAD_TYPE_UNDO;
 177         undo->head.hdr_size = bytes;
 178         undo->head.reserved01 = 0;
 179         undo->head.hdr_crc = 0;
 180         undo->undo_offset = zone_off;
 181         undo->undo_data_bytes = len;
 182         bcopy(base, undo + 1, len);
 183
 184         tail = (void *)((char *)undo + bytes - sizeof(*tail));
 185         tail->tail_signature = HAMMER_TAIL_SIGNATURE;
 186         tail->tail_type = HAMMER_HEAD_TYPE_UNDO;
 187         tail->tail_size = bytes;
 188
 189         KKASSERT(bytes >= sizeof(undo->head));
 190         undo->head.hdr_crc = crc32(undo, HAMMER_FIFO_HEAD_CRCOFF) ^
 191                              crc32(&undo->head + 1, bytes - sizeof(undo->head));
 192         undomap->next_offset += bytes;
 193
 194         hammer_modify_buffer_done(buffer);
 195         hammer_modify_volume_done(root_volume);
 196
 197         hammer_unlock(&hmp->undo_lock);
 198
 199         if (buffer)
 200                 hammer_rel_buffer(buffer, 0);
 201         return(error);
 202 }
 203
 204 /*
 205  * UNDO HISTORY API
 206  *
 207  * It is not necessary to layout an undo record for the same address space
 208  * multiple times.  Maintain a cache of recent undo's.
 209  */
 210
 211 /*
 212  * Enter an undo into the history.  Return EALREADY if the request completely
 213  * covers a previous request.
 214  */
 215 int
 216 hammer_enter_undo_history(hammer_mount_t hmp, hammer_off_t offset, int bytes)
 217 {
 218         hammer_undo_t node;
 219         hammer_undo_t onode;
 220
 221         node = RB_LOOKUP(hammer_und_rb_tree, &hmp->rb_undo_root, offset);
 222         if (node) {
 223                 TAILQ_REMOVE(&hmp->undo_lru_list, node, lru_entry);
 224                 TAILQ_INSERT_TAIL(&hmp->undo_lru_list, node, lru_entry);
 225                 if (bytes <= node->bytes)
 226                         return(EALREADY);
 227                 node->bytes = bytes;
 228                 return(0);
 229         }
 230         if (hmp->undo_alloc != HAMMER_MAX_UNDOS) {
 231                 node = &hmp->undos[hmp->undo_alloc++];
 232         } else {
 233                 node = TAILQ_FIRST(&hmp->undo_lru_list);
 234                 TAILQ_REMOVE(&hmp->undo_lru_list, node, lru_entry);
 235                 RB_REMOVE(hammer_und_rb_tree, &hmp->rb_undo_root, node);
 236         }
 237         node->offset = offset;
 238         node->bytes = bytes;
 239         TAILQ_INSERT_TAIL(&hmp->undo_lru_list, node, lru_entry);
 240         onode = RB_INSERT(hammer_und_rb_tree, &hmp->rb_undo_root, node);
 241         KKASSERT(onode == NULL);
 242         return(0);
 243 }
 244
 245 void
 246 hammer_clear_undo_history(hammer_mount_t hmp)
 247 {
 248         RB_INIT(&hmp->rb_undo_root);
 249         TAILQ_INIT(&hmp->undo_lru_list);
 250         hmp->undo_alloc = 0;
 251 }
 252
 253 /*
 254  * Misc helper routines.  Return available space and total space.
 255  */
 256 int64_t
 257 hammer_undo_used(hammer_mount_t hmp)
 258 {
 259         hammer_blockmap_t rootmap;
 260         int64_t max_bytes;
 261         int64_t bytes;
 262
 263         rootmap = &hmp->blockmap[HAMMER_ZONE_UNDO_INDEX];
 264
 265         if (rootmap->first_offset <= rootmap->next_offset) {
 266                 bytes = rootmap->next_offset - rootmap->first_offset;
 267         } else {
 268                 bytes = rootmap->alloc_offset - rootmap->first_offset +
 269                         (rootmap->next_offset & HAMMER_OFF_LONG_MASK);
 270         }
 271         max_bytes = rootmap->alloc_offset & HAMMER_OFF_SHORT_MASK;
 272         KKASSERT(bytes <= max_bytes);
 273         return(bytes);
 274 }
 275
 276 int64_t
 277 hammer_undo_space(hammer_mount_t hmp)
 278 {
 279         hammer_blockmap_t rootmap;
 280         int64_t max_bytes;
 281
 282         rootmap = &hmp->blockmap[HAMMER_ZONE_UNDO_INDEX];
 283         max_bytes = rootmap->alloc_offset & HAMMER_OFF_SHORT_MASK;
 284         return(max_bytes - hammer_undo_used(hmp));
 285 }
 286
 287 int64_t
 288 hammer_undo_max(hammer_mount_t hmp)
 289 {
 290         hammer_blockmap_t rootmap;
 291         int64_t max_bytes;
 292
 293         rootmap = &hmp->blockmap[HAMMER_ZONE_UNDO_INDEX];
 294         max_bytes = rootmap->alloc_offset & HAMMER_OFF_SHORT_MASK;
 295
 296         return(max_bytes);
 297 }
 298
 299 static int
 300 hammer_und_rb_compare(hammer_undo_t node1, hammer_undo_t node2)
 301 {
 302         if (node1->offset < node2->offset)
 303                 return(-1);
 304         if (node1->offset > node2->offset)
 305                 return(1);
 306         return(0);
 307 }
 308