db.1.85/btree/bt_open.c

   1 /*-
   2  * Copyright (c) 1990, 1993, 1994
   3  *      The Regents of the University of California.  All rights reserved.
   4  *
   5  * This code is derived from software contributed to Berkeley by
   6  * Mike Olson.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  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 the
  15  *    documentation and/or other materials provided with the distribution.
  16  * 3. All advertising materials mentioning features or use of this software
  17  *    must display the following acknowledgement:
  18  *      This product includes software developed by the University of
  19  *      California, Berkeley and its contributors.
  20  * 4. Neither the name of the University nor the names of its contributors
  21  *    may be used to endorse or promote products derived from this software
  22  *    without specific prior written permission.
  23  *
  24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  34  * SUCH DAMAGE.
  35  */
  36
  37 #if defined(LIBC_SCCS) && !defined(lint)
  38 static char sccsid[] = "@(#)bt_open.c   8.10 (Berkeley) 8/17/94";
  39 #endif /* LIBC_SCCS and not lint */
  40
  41 /*
  42  * Implementation of btree access method for 4.4BSD.
  43  *
  44  * The design here was originally based on that of the btree access method
  45  * used in the Postgres database system at UC Berkeley.  This implementation
  46  * is wholly independent of the Postgres code.
  47  */
  48
  49 #include <sys/param.h>
  50 #include <sys/stat.h>
  51
  52 #include <errno.h>
  53 #include <fcntl.h>
  54 #include <limits.h>
  55 #include <signal.h>
  56 #include <stdio.h>
  57 #include <stdlib.h>
  58 #include <string.h>
  59 #include <unistd.h>
  60
  61 #include <db.h>
  62 #include "btree.h"
  63
  64 #ifdef DEBUG
  65 #undef  MINPSIZE
  66 #define MINPSIZE        128
  67 #endif
  68
  69 static int byteorder __P((void));
  70 static int nroot __P((BTREE *));
  71 static int tmp __P((void));
  72
  73 /*
  74  * __BT_OPEN -- Open a btree.
  75  *
  76  * Creates and fills a DB struct, and calls the routine that actually
  77  * opens the btree.
  78  *
  79  * Parameters:
  80  *      fname:  filename (NULL for in-memory trees)
  81  *      flags:  open flag bits
  82  *      mode:   open permission bits
  83  *      b:      BTREEINFO pointer
  84  *
  85  * Returns:
  86  *      NULL on failure, pointer to DB on success.
  87  *
  88  */
  89 DB *
  90 __bt_open(fname, flags, mode, openinfo, dflags)
  91         const char *fname;
  92         int flags, mode, dflags;
  93         const BTREEINFO *openinfo;
  94 {
  95         struct stat sb;
  96         BTMETA m;
  97         BTREE *t;
  98         BTREEINFO b;
  99         DB *dbp;
 100         pgno_t ncache;
 101         ssize_t nr;
 102         int machine_lorder;
 103
 104         t = NULL;
 105
 106         /*
 107          * Intention is to make sure all of the user's selections are okay
 108          * here and then use them without checking.  Can't be complete, since
 109          * we don't know the right page size, lorder or flags until the backing
 110          * file is opened.  Also, the file's page size can cause the cachesize
 111          * to change.
 112          */
 113         machine_lorder = byteorder();
 114         if (openinfo) {
 115                 b = *openinfo;
 116
 117                 /* Flags: R_DUP. */
 118                 if (b.flags & ~(R_DUP))
 119                         goto einval;
 120
 121                 /*
 122                  * Page size must be indx_t aligned and >= MINPSIZE.  Default
 123                  * page size is set farther on, based on the underlying file
 124                  * transfer size.
 125                  */
 126                 if (b.psize &&
 127                     (b.psize < MINPSIZE || b.psize > MAX_PAGE_OFFSET + 1 ||
 128                     b.psize & sizeof(indx_t) - 1))
 129                         goto einval;
 130
 131                 /* Minimum number of keys per page; absolute minimum is 2. */
 132                 if (b.minkeypage) {
 133                         if (b.minkeypage < 2)
 134                                 goto einval;
 135                 } else
 136                         b.minkeypage = DEFMINKEYPAGE;
 137
 138                 /* If no comparison, use default comparison and prefix. */
 139                 if (b.compare == NULL) {
 140                         b.compare = __bt_defcmp;
 141                         if (b.prefix == NULL)
 142                                 b.prefix = __bt_defpfx;
 143                 }
 144
 145                 if (b.lorder == 0)
 146                         b.lorder = machine_lorder;
 147         } else {
 148                 b.compare = __bt_defcmp;
 149                 b.cachesize = 0;
 150                 b.flags = 0;
 151                 b.lorder = machine_lorder;
 152                 b.minkeypage = DEFMINKEYPAGE;
 153                 b.prefix = __bt_defpfx;
 154                 b.psize = 0;
 155         }
 156
 157         /* Check for the ubiquitous PDP-11. */
 158         if (b.lorder != BIG_ENDIAN && b.lorder != LITTLE_ENDIAN)
 159                 goto einval;
 160
 161         /* Allocate and initialize DB and BTREE structures. */
 162         if ((t = (BTREE *)malloc(sizeof(BTREE))) == NULL)
 163                 goto err;
 164         memset(t, 0, sizeof(BTREE));
 165         t->bt_fd = -1;                  /* Don't close unopened fd on error. */
 166         t->bt_lorder = b.lorder;
 167         t->bt_order = NOT;
 168         t->bt_cmp = b.compare;
 169         t->bt_pfx = b.prefix;
 170         t->bt_rfd = -1;
 171
 172         if ((t->bt_dbp = dbp = (DB *)malloc(sizeof(DB))) == NULL)
 173                 goto err;
 174         memset(t->bt_dbp, 0, sizeof(DB));
 175         if (t->bt_lorder != machine_lorder)
 176                 F_SET(t, B_NEEDSWAP);
 177
 178         dbp->type = DB_BTREE;
 179         dbp->internal = t;
 180         dbp->close = __bt_close;
 181         dbp->del = __bt_delete;
 182         dbp->fd = __bt_fd;
 183         dbp->get = __bt_get;
 184         dbp->put = __bt_put;
 185         dbp->seq = __bt_seq;
 186         dbp->sync = __bt_sync;
 187
 188         /*
 189          * If no file name was supplied, this is an in-memory btree and we
 190          * open a backing temporary file.  Otherwise, it's a disk-based tree.
 191          */
 192         if (fname) {
 193                 switch (flags & O_ACCMODE) {
 194                 case O_RDONLY:
 195                         F_SET(t, B_RDONLY);
 196                         break;
 197                 case O_RDWR:
 198                         break;
 199                 case O_WRONLY:
 200                 default:
 201                         goto einval;
 202                 }
 203
 204                 if ((t->bt_fd = open(fname, flags, mode)) < 0)
 205                         goto err;
 206
 207         } else {
 208                 if ((flags & O_ACCMODE) != O_RDWR)
 209                         goto einval;
 210                 if ((t->bt_fd = tmp()) == -1)
 211                         goto err;
 212                 F_SET(t, B_INMEM);
 213         }
 214
 215         if (fcntl(t->bt_fd, F_SETFD, 1) == -1)
 216                 goto err;
 217
 218         if (fstat(t->bt_fd, &sb))
 219                 goto err;
 220         if (sb.st_size) {
 221                 if ((nr = read(t->bt_fd, &m, sizeof(BTMETA))) < 0)
 222                         goto err;
 223                 if (nr != sizeof(BTMETA))
 224                         goto eftype;
 225
 226                 /*
 227                  * Read in the meta-data.  This can change the notion of what
 228                  * the lorder, page size and flags are, and, when the page size
 229                  * changes, the cachesize value can change too.  If the user
 230                  * specified the wrong byte order for an existing database, we
 231                  * don't bother to return an error, we just clear the NEEDSWAP
 232                  * bit.
 233                  */
 234                 if (m.magic == BTREEMAGIC)
 235                         F_CLR(t, B_NEEDSWAP);
 236                 else {
 237                         F_SET(t, B_NEEDSWAP);
 238                         M_32_SWAP(m.magic);
 239                         M_32_SWAP(m.version);
 240                         M_32_SWAP(m.psize);
 241                         M_32_SWAP(m.free);
 242                         M_32_SWAP(m.nrecs);
 243                         M_32_SWAP(m.flags);
 244                 }
 245                 if (m.magic != BTREEMAGIC || m.version != BTREEVERSION)
 246                         goto eftype;
 247                 if (m.psize < MINPSIZE || m.psize > MAX_PAGE_OFFSET + 1 ||
 248                     m.psize & sizeof(indx_t) - 1)
 249                         goto eftype;
 250                 if (m.flags & ~SAVEMETA)
 251                         goto eftype;
 252                 b.psize = m.psize;
 253                 F_SET(t, m.flags);
 254                 t->bt_free = m.free;
 255                 t->bt_nrecs = m.nrecs;
 256         } else {
 257                 /*
 258                  * Set the page size to the best value for I/O to this file.
 259                  * Don't overflow the page offset type.
 260                  */
 261                 if (b.psize == 0) {
 262                         b.psize = sb.st_blksize;
 263                         if (b.psize < MINPSIZE)
 264                                 b.psize = MINPSIZE;
 265                         if (b.psize > MAX_PAGE_OFFSET + 1)
 266                                 b.psize = MAX_PAGE_OFFSET + 1;
 267                 }
 268
 269                 /* Set flag if duplicates permitted. */
 270                 if (!(b.flags & R_DUP))
 271                         F_SET(t, B_NODUPS);
 272
 273                 t->bt_free = P_INVALID;
 274                 t->bt_nrecs = 0;
 275                 F_SET(t, B_METADIRTY);
 276         }
 277
 278         t->bt_psize = b.psize;
 279
 280         /* Set the cache size; must be a multiple of the page size. */
 281         if (b.cachesize && b.cachesize & b.psize - 1)
 282                 b.cachesize += (~b.cachesize & b.psize - 1) + 1;
 283         if (b.cachesize < b.psize * MINCACHE)
 284                 b.cachesize = b.psize * MINCACHE;
 285
 286         /* Calculate number of pages to cache. */
 287         ncache = (b.cachesize + t->bt_psize - 1) / t->bt_psize;
 288
 289         /*
 290          * The btree data structure requires that at least two keys can fit on
 291          * a page, but other than that there's no fixed requirement.  The user
 292          * specified a minimum number per page, and we translated that into the
 293          * number of bytes a key/data pair can use before being placed on an
 294          * overflow page.  This calculation includes the page header, the size
 295          * of the index referencing the leaf item and the size of the leaf item
 296          * structure.  Also, don't let the user specify a minkeypage such that
 297          * a key/data pair won't fit even if both key and data are on overflow
 298          * pages.
 299          */
 300         t->bt_ovflsize = (t->bt_psize - BTDATAOFF) / b.minkeypage -
 301             (sizeof(indx_t) + NBLEAFDBT(0, 0));
 302         if (t->bt_ovflsize < NBLEAFDBT(NOVFLSIZE, NOVFLSIZE) + sizeof(indx_t))
 303                 t->bt_ovflsize =
 304                     NBLEAFDBT(NOVFLSIZE, NOVFLSIZE) + sizeof(indx_t);
 305
 306         /* Initialize the buffer pool. */
 307         if ((t->bt_mp =
 308             mpool_open(NULL, t->bt_fd, t->bt_psize, ncache)) == NULL)
 309                 goto err;
 310         if (!F_ISSET(t, B_INMEM))
 311                 mpool_filter(t->bt_mp, __bt_pgin, __bt_pgout, t);
 312
 313         /* Create a root page if new tree. */
 314         if (nroot(t) == RET_ERROR)
 315                 goto err;
 316
 317         /* Global flags. */
 318         if (dflags & DB_LOCK)
 319                 F_SET(t, B_DB_LOCK);
 320         if (dflags & DB_SHMEM)
 321                 F_SET(t, B_DB_SHMEM);
 322         if (dflags & DB_TXN)
 323                 F_SET(t, B_DB_TXN);
 324
 325         return (dbp);
 326
 327 einval: errno = EINVAL;
 328         goto err;
 329
 330 eftype: errno = EFTYPE;
 331         goto err;
 332
 333 err:    if (t) {
 334                 if (t->bt_dbp)
 335                         free(t->bt_dbp);
 336                 if (t->bt_fd != -1)
 337                         (void)close(t->bt_fd);
 338                 free(t);
 339         }
 340         return (NULL);
 341 }
 342
 343 /*
 344  * NROOT -- Create the root of a new tree.
 345  *
 346  * Parameters:
 347  *      t:      tree
 348  *
 349  * Returns:
 350  *      RET_ERROR, RET_SUCCESS
 351  */
 352 static int
 353 nroot(t)
 354         BTREE *t;
 355 {
 356         PAGE *meta, *root;
 357         pgno_t npg;
 358
 359         if ((meta = mpool_get(t->bt_mp, 0, 0)) != NULL) {
 360                 mpool_put(t->bt_mp, meta, 0);
 361                 return (RET_SUCCESS);
 362         }
 363         if (errno != EINVAL)            /* It's OK to not exist. */
 364                 return (RET_ERROR);
 365         errno = 0;
 366
 367         if ((meta = mpool_new(t->bt_mp, &npg)) == NULL)
 368                 return (RET_ERROR);
 369
 370         if ((root = mpool_new(t->bt_mp, &npg)) == NULL)
 371                 return (RET_ERROR);
 372
 373         if (npg != P_ROOT)
 374                 return (RET_ERROR);
 375         root->pgno = npg;
 376         root->prevpg = root->nextpg = P_INVALID;
 377         root->lower = BTDATAOFF;
 378         root->upper = t->bt_psize;
 379         root->flags = P_BLEAF;
 380         memset(meta, 0, t->bt_psize);
 381         mpool_put(t->bt_mp, meta, MPOOL_DIRTY);
 382         mpool_put(t->bt_mp, root, MPOOL_DIRTY);
 383         return (RET_SUCCESS);
 384 }
 385
 386 static int
 387 tmp()
 388 {
 389         sigset_t set, oset;
 390         int fd;
 391         char *envtmp;
 392         char path[MAXPATHLEN];
 393
 394         envtmp = getenv("TMPDIR");
 395         (void)snprintf(path,
 396             sizeof(path), "%s/bt.XXXXXX", envtmp ? envtmp : "/tmp");
 397
 398         (void)sigfillset(&set);
 399         (void)sigprocmask(SIG_BLOCK, &set, &oset);
 400         if ((fd = mkstemp(path)) != -1)
 401                 (void)unlink(path);
 402         (void)sigprocmask(SIG_SETMASK, &oset, NULL);
 403         return(fd);
 404 }
 405
 406 static int
 407 byteorder()
 408 {
 409         u_int32_t x;
 410         u_char *p;
 411
 412         x = 0x01020304;
 413         p = (u_char *)&x;
 414         switch (*p) {
 415         case 1:
 416                 return (BIG_ENDIAN);
 417         case 4:
 418                 return (LITTLE_ENDIAN);
 419         default:
 420                 return (0);
 421         }
 422 }
 423
 424 int
 425 __bt_fd(dbp)
 426         const DB *dbp;
 427 {
 428         BTREE *t;
 429
 430         t = dbp->internal;
 431
 432         /* Toss any page pinned across calls. */
 433         if (t->bt_pinned != NULL) {
 434                 mpool_put(t->bt_mp, t->bt_pinned, 0);
 435                 t->bt_pinned = NULL;
 436         }
 437
 438         /* In-memory database can't have a file descriptor. */
 439         if (F_ISSET(t, B_INMEM)) {
 440                 errno = ENOENT;
 441                 return (-1);
 442         }
 443         return (t->bt_fd);
 444 }