gdb/bcache.c

   1 /* Implement a cached obstack.
   2    Written by Fred Fish <fnf@cygnus.com>
   3    Rewritten by Jim Blandy <jimb@cygnus.com>
   4
   5    Copyright (C) 1999, 2000, 2002, 2003, 2007, 2008, 2009, 2010
   6    Free Software Foundation, Inc.
   7
   8    This file is part of GDB.
   9
  10    This program is free software; you can redistribute it and/or modify
  11    it under the terms of the GNU General Public License as published by
  12    the Free Software Foundation; either version 3 of the License, or
  13    (at your option) any later version.
  14
  15    This program is distributed in the hope that it will be useful,
  16    but WITHOUT ANY WARRANTY; without even the implied warranty of
  17    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18    GNU General Public License for more details.
  19
  20    You should have received a copy of the GNU General Public License
  21    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  22
  23 #include "defs.h"
  24 #include "gdb_obstack.h"
  25 #include "bcache.h"
  26 #include "gdb_string.h"         /* For memcpy declaration */
  27 #include "gdb_assert.h"
  28
  29 #include <stddef.h>
  30 #include <stdlib.h>
  31
  32 /* The type used to hold a single bcache string.  The user data is
  33    stored in d.data.  Since it can be any type, it needs to have the
  34    same alignment as the most strict alignment of any type on the host
  35    machine.  I don't know of any really correct way to do this in
  36    stock ANSI C, so just do it the same way obstack.h does.  */
  37
  38 struct bstring
  39 {
  40   /* Hash chain.  */
  41   struct bstring *next;
  42   /* Assume the data length is no more than 64k.  */
  43   unsigned short length;
  44   /* The half hash hack.  This contains the upper 16 bits of the hash
  45      value and is used as a pre-check when comparing two strings and
  46      avoids the need to do length or memcmp calls.  It proves to be
  47      roughly 100% effective.  */
  48   unsigned short half_hash;
  49
  50   union
  51   {
  52     char data[1];
  53     double dummy;
  54   }
  55   d;
  56 };
  57
  58
  59 /* The structure for a bcache itself.  The bcache is initialized, in
  60    bcache_xmalloc(), by filling it with zeros and then setting the
  61    corresponding obstack's malloc() and free() methods.  */
  62
  63 struct bcache
  64 {
  65   /* All the bstrings are allocated here.  */
  66   struct obstack cache;
  67
  68   /* How many hash buckets we're using.  */
  69   unsigned int num_buckets;
  70
  71   /* Hash buckets.  This table is allocated using malloc, so when we
  72      grow the table we can return the old table to the system.  */
  73   struct bstring **bucket;
  74
  75   /* Statistics.  */
  76   unsigned long unique_count;   /* number of unique strings */
  77   long total_count;     /* total number of strings cached, including dups */
  78   long unique_size;     /* size of unique strings, in bytes */
  79   long total_size;      /* total number of bytes cached, including dups */
  80   long structure_size;  /* total size of bcache, including infrastructure */
  81   /* Number of times that the hash table is expanded and hence
  82      re-built, and the corresponding number of times that a string is
  83      [re]hashed as part of entering it into the expanded table.  The
  84      total number of hashes can be computed by adding TOTAL_COUNT to
  85      expand_hash_count.  */
  86   unsigned long expand_count;
  87   unsigned long expand_hash_count;
  88   /* Number of times that the half-hash compare hit (compare the upper
  89      16 bits of hash values) hit, but the corresponding combined
  90      length/data compare missed.  */
  91   unsigned long half_hash_miss_count;
  92 };
  93
  94 /* The old hash function was stolen from SDBM. This is what DB 3.0 uses now,
  95  * and is better than the old one.
  96  */
  97 \f
  98 unsigned long
  99 hash(const void *addr, int length)
 100 {
 101                 const unsigned char *k, *e;
 102                 unsigned long h;
 103
 104                 k = (const unsigned char *)addr;
 105                 e = k+length;
 106                 for (h=0; k< e;++k)
 107                 {
 108                                 h *=16777619;
 109                                 h ^= *k;
 110                 }
 111                 return (h);
 112 }
 113 \f
 114 /* Growing the bcache's hash table.  */
 115
 116 /* If the average chain length grows beyond this, then we want to
 117    resize our hash table.  */
 118 #define CHAIN_LENGTH_THRESHOLD (5)
 119
 120 static void
 121 expand_hash_table (struct bcache *bcache)
 122 {
 123   /* A table of good hash table sizes.  Whenever we grow, we pick the
 124      next larger size from this table.  sizes[i] is close to 1 << (i+10),
 125      so we roughly double the table size each time.  After we fall off
 126      the end of this table, we just double.  Don't laugh --- there have
 127      been executables sighted with a gigabyte of debug info.  */
 128   static unsigned long sizes[] = {
 129     1021, 2053, 4099, 8191, 16381, 32771,
 130     65537, 131071, 262144, 524287, 1048573, 2097143,
 131     4194301, 8388617, 16777213, 33554467, 67108859, 134217757,
 132     268435459, 536870923, 1073741827, 2147483659UL
 133   };
 134   unsigned int new_num_buckets;
 135   struct bstring **new_buckets;
 136   unsigned int i;
 137
 138   /* Count the stats.  Every unique item needs to be re-hashed and
 139      re-entered.  */
 140   bcache->expand_count++;
 141   bcache->expand_hash_count += bcache->unique_count;
 142
 143   /* Find the next size.  */
 144   new_num_buckets = bcache->num_buckets * 2;
 145   for (i = 0; i < (sizeof (sizes) / sizeof (sizes[0])); i++)
 146     if (sizes[i] > bcache->num_buckets)
 147       {
 148         new_num_buckets = sizes[i];
 149         break;
 150       }
 151
 152   /* Allocate the new table.  */
 153   {
 154     size_t new_size = new_num_buckets * sizeof (new_buckets[0]);
 155     new_buckets = (struct bstring **) xmalloc (new_size);
 156     memset (new_buckets, 0, new_size);
 157
 158     bcache->structure_size -= (bcache->num_buckets
 159                                * sizeof (bcache->bucket[0]));
 160     bcache->structure_size += new_size;
 161   }
 162
 163   /* Rehash all existing strings.  */
 164   for (i = 0; i < bcache->num_buckets; i++)
 165     {
 166       struct bstring *s, *next;
 167
 168       for (s = bcache->bucket[i]; s; s = next)
 169         {
 170           struct bstring **new_bucket;
 171           next = s->next;
 172
 173           new_bucket = &new_buckets[(hash (&s->d.data, s->length)
 174                                      % new_num_buckets)];
 175           s->next = *new_bucket;
 176           *new_bucket = s;
 177         }
 178     }
 179
 180   /* Plug in the new table.  */
 181   if (bcache->bucket)
 182     xfree (bcache->bucket);
 183   bcache->bucket = new_buckets;
 184   bcache->num_buckets = new_num_buckets;
 185 }
 186
 187 \f
 188 /* Looking up things in the bcache.  */
 189
 190 /* The number of bytes needed to allocate a struct bstring whose data
 191    is N bytes long.  */
 192 #define BSTRING_SIZE(n) (offsetof (struct bstring, d.data) + (n))
 193
 194 /* Find a copy of the LENGTH bytes at ADDR in BCACHE.  If BCACHE has
 195    never seen those bytes before, add a copy of them to BCACHE.  In
 196    either case, return a pointer to BCACHE's copy of that string.  */
 197 const void *
 198 bcache (const void *addr, int length, struct bcache *bcache)
 199 {
 200   return bcache_full (addr, length, bcache, NULL);
 201 }
 202
 203 /* Find a copy of the LENGTH bytes at ADDR in BCACHE.  If BCACHE has
 204    never seen those bytes before, add a copy of them to BCACHE.  In
 205    either case, return a pointer to BCACHE's copy of that string.  If
 206    optional ADDED is not NULL, return 1 in case of new entry or 0 if
 207    returning an old entry.  */
 208
 209 const void *
 210 bcache_full (const void *addr, int length, struct bcache *bcache, int *added)
 211 {
 212   unsigned long full_hash;
 213   unsigned short half_hash;
 214   int hash_index;
 215   struct bstring *s;
 216
 217   if (added)
 218     *added = 0;
 219
 220   /* Lazily initialize the obstack.  This can save quite a bit of
 221      memory in some cases.  */
 222   if (bcache->total_count == 0)
 223     {
 224       /* We could use obstack_specify_allocation here instead, but
 225          gdb_obstack.h specifies the allocation/deallocation
 226          functions.  */
 227       obstack_init (&bcache->cache);
 228     }
 229
 230   /* If our average chain length is too high, expand the hash table.  */
 231   if (bcache->unique_count >= bcache->num_buckets * CHAIN_LENGTH_THRESHOLD)
 232     expand_hash_table (bcache);
 233
 234   bcache->total_count++;
 235   bcache->total_size += length;
 236
 237   full_hash = hash (addr, length);
 238   half_hash = (full_hash >> 16);
 239   hash_index = full_hash % bcache->num_buckets;
 240
 241   /* Search the hash bucket for a string identical to the caller's.
 242      As a short-circuit first compare the upper part of each hash
 243      values.  */
 244   for (s = bcache->bucket[hash_index]; s; s = s->next)
 245     {
 246       if (s->half_hash == half_hash)
 247         {
 248           if (s->length == length
 249               && ! memcmp (&s->d.data, addr, length))
 250             return &s->d.data;
 251           else
 252             bcache->half_hash_miss_count++;
 253         }
 254     }
 255
 256   /* The user's string isn't in the list.  Insert it after *ps.  */
 257   {
 258     struct bstring *new
 259       = obstack_alloc (&bcache->cache, BSTRING_SIZE (length));
 260     memcpy (&new->d.data, addr, length);
 261     new->length = length;
 262     new->next = bcache->bucket[hash_index];
 263     new->half_hash = half_hash;
 264     bcache->bucket[hash_index] = new;
 265
 266     bcache->unique_count++;
 267     bcache->unique_size += length;
 268     bcache->structure_size += BSTRING_SIZE (length);
 269
 270     if (added)
 271       *added = 1;
 272
 273     return &new->d.data;
 274   }
 275 }
 276 \f
 277 /* Allocating and freeing bcaches.  */
 278
 279 struct bcache *
 280 bcache_xmalloc (void)
 281 {
 282   /* Allocate the bcache pre-zeroed.  */
 283   struct bcache *b = XCALLOC (1, struct bcache);
 284   return b;
 285 }
 286
 287 /* Free all the storage associated with BCACHE.  */
 288 void
 289 bcache_xfree (struct bcache *bcache)
 290 {
 291   if (bcache == NULL)
 292     return;
 293   /* Only free the obstack if we actually initialized it.  */
 294   if (bcache->total_count > 0)
 295     obstack_free (&bcache->cache, 0);
 296   xfree (bcache->bucket);
 297   xfree (bcache);
 298 }
 299
 300
 301 \f
 302 /* Printing statistics.  */
 303
 304 static void
 305 print_percentage (int portion, int total)
 306 {
 307   if (total == 0)
 308     /* i18n: Like "Percentage of duplicates, by count: (not applicable)" */
 309     printf_filtered (_("(not applicable)\n"));
 310   else
 311     printf_filtered ("%3d%%\n", (int) (portion * 100.0 / total));
 312 }
 313
 314
 315 /* Print statistics on BCACHE's memory usage and efficacity at
 316    eliminating duplication.  NAME should describe the kind of data
 317    BCACHE holds.  Statistics are printed using `printf_filtered' and
 318    its ilk.  */
 319 void
 320 print_bcache_statistics (struct bcache *c, char *type)
 321 {
 322   int occupied_buckets;
 323   int max_chain_length;
 324   int median_chain_length;
 325   int max_entry_size;
 326   int median_entry_size;
 327
 328   /* Count the number of occupied buckets, tally the various string
 329      lengths, and measure chain lengths.  */
 330   {
 331     unsigned int b;
 332     int *chain_length = XCALLOC (c->num_buckets + 1, int);
 333     int *entry_size = XCALLOC (c->unique_count + 1, int);
 334     int stringi = 0;
 335
 336     occupied_buckets = 0;
 337
 338     for (b = 0; b < c->num_buckets; b++)
 339       {
 340         struct bstring *s = c->bucket[b];
 341
 342         chain_length[b] = 0;
 343
 344         if (s)
 345           {
 346             occupied_buckets++;
 347
 348             while (s)
 349               {
 350                 gdb_assert (b < c->num_buckets);
 351                 chain_length[b]++;
 352                 gdb_assert (stringi < c->unique_count);
 353                 entry_size[stringi++] = s->length;
 354                 s = s->next;
 355               }
 356           }
 357       }
 358
 359     /* To compute the median, we need the set of chain lengths sorted.  */
 360     qsort (chain_length, c->num_buckets, sizeof (chain_length[0]),
 361            compare_positive_ints);
 362     qsort (entry_size, c->unique_count, sizeof (entry_size[0]),
 363            compare_positive_ints);
 364
 365     if (c->num_buckets > 0)
 366       {
 367         max_chain_length = chain_length[c->num_buckets - 1];
 368         median_chain_length = chain_length[c->num_buckets / 2];
 369       }
 370     else
 371       {
 372         max_chain_length = 0;
 373         median_chain_length = 0;
 374       }
 375     if (c->unique_count > 0)
 376       {
 377         max_entry_size = entry_size[c->unique_count - 1];
 378         median_entry_size = entry_size[c->unique_count / 2];
 379       }
 380     else
 381       {
 382         max_entry_size = 0;
 383         median_entry_size = 0;
 384       }
 385
 386     xfree (chain_length);
 387     xfree (entry_size);
 388   }
 389
 390   printf_filtered (_("  Cached '%s' statistics:\n"), type);
 391   printf_filtered (_("    Total object count:  %ld\n"), c->total_count);
 392   printf_filtered (_("    Unique object count: %lu\n"), c->unique_count);
 393   printf_filtered (_("    Percentage of duplicates, by count: "));
 394   print_percentage (c->total_count - c->unique_count, c->total_count);
 395   printf_filtered ("\n");
 396
 397   printf_filtered (_("    Total object size:   %ld\n"), c->total_size);
 398   printf_filtered (_("    Unique object size:  %ld\n"), c->unique_size);
 399   printf_filtered (_("    Percentage of duplicates, by size:  "));
 400   print_percentage (c->total_size - c->unique_size, c->total_size);
 401   printf_filtered ("\n");
 402
 403   printf_filtered (_("    Max entry size:     %d\n"), max_entry_size);
 404   printf_filtered (_("    Average entry size: "));
 405   if (c->unique_count > 0)
 406     printf_filtered ("%ld\n", c->unique_size / c->unique_count);
 407   else
 408     /* i18n: "Average entry size: (not applicable)" */
 409     printf_filtered (_("(not applicable)\n"));
 410   printf_filtered (_("    Median entry size:  %d\n"), median_entry_size);
 411   printf_filtered ("\n");
 412
 413   printf_filtered (_("    Total memory used by bcache, including overhead: %ld\n"),
 414                    c->structure_size);
 415   printf_filtered (_("    Percentage memory overhead: "));
 416   print_percentage (c->structure_size - c->unique_size, c->unique_size);
 417   printf_filtered (_("    Net memory savings:         "));
 418   print_percentage (c->total_size - c->structure_size, c->total_size);
 419   printf_filtered ("\n");
 420
 421   printf_filtered (_("    Hash table size:           %3d\n"), c->num_buckets);
 422   printf_filtered (_("    Hash table expands:        %lu\n"),
 423                    c->expand_count);
 424   printf_filtered (_("    Hash table hashes:         %lu\n"),
 425                    c->total_count + c->expand_hash_count);
 426   printf_filtered (_("    Half hash misses:          %lu\n"),
 427                    c->half_hash_miss_count);
 428   printf_filtered (_("    Hash table population:     "));
 429   print_percentage (occupied_buckets, c->num_buckets);
 430   printf_filtered (_("    Median hash chain length:  %3d\n"),
 431                    median_chain_length);
 432   printf_filtered (_("    Average hash chain length: "));
 433   if (c->num_buckets > 0)
 434     printf_filtered ("%3lu\n", c->unique_count / c->num_buckets);
 435   else
 436     /* i18n: "Average hash chain length: (not applicable)" */
 437     printf_filtered (_("(not applicable)\n"));
 438   printf_filtered (_("    Maximum hash chain length: %3d\n"), max_chain_length);
 439   printf_filtered ("\n");
 440 }
 441
 442 int
 443 bcache_memory_used (struct bcache *bcache)
 444 {
 445   if (bcache->total_count == 0)
 446     return 0;
 447   return obstack_memory_used (&bcache->cache);
 448 }