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
   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 static void *
 198 bcache_data (const void *addr, int length, struct bcache *bcache)
 199 {
 200   unsigned long full_hash;
 201   unsigned short half_hash;
 202   int hash_index;
 203   struct bstring *s;
 204
 205   /* If our average chain length is too high, expand the hash table.  */
 206   if (bcache->unique_count >= bcache->num_buckets * CHAIN_LENGTH_THRESHOLD)
 207     expand_hash_table (bcache);
 208
 209   bcache->total_count++;
 210   bcache->total_size += length;
 211
 212   full_hash = hash (addr, length);
 213   half_hash = (full_hash >> 16);
 214   hash_index = full_hash % bcache->num_buckets;
 215
 216   /* Search the hash bucket for a string identical to the caller's.
 217      As a short-circuit first compare the upper part of each hash
 218      values.  */
 219   for (s = bcache->bucket[hash_index]; s; s = s->next)
 220     {
 221       if (s->half_hash == half_hash)
 222         {
 223           if (s->length == length
 224               && ! memcmp (&s->d.data, addr, length))
 225             return &s->d.data;
 226           else
 227             bcache->half_hash_miss_count++;
 228         }
 229     }
 230
 231   /* The user's string isn't in the list.  Insert it after *ps.  */
 232   {
 233     struct bstring *new
 234       = obstack_alloc (&bcache->cache, BSTRING_SIZE (length));
 235     memcpy (&new->d.data, addr, length);
 236     new->length = length;
 237     new->next = bcache->bucket[hash_index];
 238     new->half_hash = half_hash;
 239     bcache->bucket[hash_index] = new;
 240
 241     bcache->unique_count++;
 242     bcache->unique_size += length;
 243     bcache->structure_size += BSTRING_SIZE (length);
 244
 245     return &new->d.data;
 246   }
 247 }
 248
 249 void *
 250 deprecated_bcache (const void *addr, int length, struct bcache *bcache)
 251 {
 252   return bcache_data (addr, length, bcache);
 253 }
 254
 255 const void *
 256 bcache (const void *addr, int length, struct bcache *bcache)
 257 {
 258   return bcache_data (addr, length, bcache);
 259 }
 260 \f
 261 /* Allocating and freeing bcaches.  */
 262
 263 struct bcache *
 264 bcache_xmalloc (void)
 265 {
 266   /* Allocate the bcache pre-zeroed.  */
 267   struct bcache *b = XCALLOC (1, struct bcache);
 268   /* We could use obstack_specify_allocation here instead, but
 269      gdb_obstack.h specifies the allocation/deallocation
 270      functions.  */
 271   obstack_init (&b->cache);
 272   return b;
 273 }
 274
 275 /* Free all the storage associated with BCACHE.  */
 276 void
 277 bcache_xfree (struct bcache *bcache)
 278 {
 279   if (bcache == NULL)
 280     return;
 281   obstack_free (&bcache->cache, 0);
 282   xfree (bcache->bucket);
 283   xfree (bcache);
 284 }
 285
 286
 287 \f
 288 /* Printing statistics.  */
 289
 290 static int
 291 compare_ints (const void *ap, const void *bp)
 292 {
 293   /* Because we know we're comparing two ints which are positive,
 294      there's no danger of overflow here.  */
 295   return * (int *) ap - * (int *) bp;
 296 }
 297
 298
 299 static void
 300 print_percentage (int portion, int total)
 301 {
 302   if (total == 0)
 303     /* i18n: Like "Percentage of duplicates, by count: (not applicable)" */
 304     printf_filtered (_("(not applicable)\n"));
 305   else
 306     printf_filtered ("%3d%%\n", (int) (portion * 100.0 / total));
 307 }
 308
 309
 310 /* Print statistics on BCACHE's memory usage and efficacity at
 311    eliminating duplication.  NAME should describe the kind of data
 312    BCACHE holds.  Statistics are printed using `printf_filtered' and
 313    its ilk.  */
 314 void
 315 print_bcache_statistics (struct bcache *c, char *type)
 316 {
 317   int occupied_buckets;
 318   int max_chain_length;
 319   int median_chain_length;
 320   int max_entry_size;
 321   int median_entry_size;
 322
 323   /* Count the number of occupied buckets, tally the various string
 324      lengths, and measure chain lengths.  */
 325   {
 326     unsigned int b;
 327     int *chain_length = XCALLOC (c->num_buckets + 1, int);
 328     int *entry_size = XCALLOC (c->unique_count + 1, int);
 329     int stringi = 0;
 330
 331     occupied_buckets = 0;
 332
 333     for (b = 0; b < c->num_buckets; b++)
 334       {
 335         struct bstring *s = c->bucket[b];
 336
 337         chain_length[b] = 0;
 338
 339         if (s)
 340           {
 341             occupied_buckets++;
 342
 343             while (s)
 344               {
 345                 gdb_assert (b < c->num_buckets);
 346                 chain_length[b]++;
 347                 gdb_assert (stringi < c->unique_count);
 348                 entry_size[stringi++] = s->length;
 349                 s = s->next;
 350               }
 351           }
 352       }
 353
 354     /* To compute the median, we need the set of chain lengths sorted.  */
 355     qsort (chain_length, c->num_buckets, sizeof (chain_length[0]),
 356            compare_ints);
 357     qsort (entry_size, c->unique_count, sizeof (entry_size[0]),
 358            compare_ints);
 359
 360     if (c->num_buckets > 0)
 361       {
 362         max_chain_length = chain_length[c->num_buckets - 1];
 363         median_chain_length = chain_length[c->num_buckets / 2];
 364       }
 365     else
 366       {
 367         max_chain_length = 0;
 368         median_chain_length = 0;
 369       }
 370     if (c->unique_count > 0)
 371       {
 372         max_entry_size = entry_size[c->unique_count - 1];
 373         median_entry_size = entry_size[c->unique_count / 2];
 374       }
 375     else
 376       {
 377         max_entry_size = 0;
 378         median_entry_size = 0;
 379       }
 380
 381     xfree (chain_length);
 382     xfree (entry_size);
 383   }
 384
 385   printf_filtered (_("  Cached '%s' statistics:\n"), type);
 386   printf_filtered (_("    Total object count:  %ld\n"), c->total_count);
 387   printf_filtered (_("    Unique object count: %lu\n"), c->unique_count);
 388   printf_filtered (_("    Percentage of duplicates, by count: "));
 389   print_percentage (c->total_count - c->unique_count, c->total_count);
 390   printf_filtered ("\n");
 391
 392   printf_filtered (_("    Total object size:   %ld\n"), c->total_size);
 393   printf_filtered (_("    Unique object size:  %ld\n"), c->unique_size);
 394   printf_filtered (_("    Percentage of duplicates, by size:  "));
 395   print_percentage (c->total_size - c->unique_size, c->total_size);
 396   printf_filtered ("\n");
 397
 398   printf_filtered (_("    Max entry size:     %d\n"), max_entry_size);
 399   printf_filtered (_("    Average entry size: "));
 400   if (c->unique_count > 0)
 401     printf_filtered ("%ld\n", c->unique_size / c->unique_count);
 402   else
 403     /* i18n: "Average entry size: (not applicable)" */
 404     printf_filtered (_("(not applicable)\n"));
 405   printf_filtered (_("    Median entry size:  %d\n"), median_entry_size);
 406   printf_filtered ("\n");
 407
 408   printf_filtered (_("    Total memory used by bcache, including overhead: %ld\n"),
 409                    c->structure_size);
 410   printf_filtered (_("    Percentage memory overhead: "));
 411   print_percentage (c->structure_size - c->unique_size, c->unique_size);
 412   printf_filtered (_("    Net memory savings:         "));
 413   print_percentage (c->total_size - c->structure_size, c->total_size);
 414   printf_filtered ("\n");
 415
 416   printf_filtered (_("    Hash table size:           %3d\n"), c->num_buckets);
 417   printf_filtered (_("    Hash table expands:        %lu\n"),
 418                    c->expand_count);
 419   printf_filtered (_("    Hash table hashes:         %lu\n"),
 420                    c->total_count + c->expand_hash_count);
 421   printf_filtered (_("    Half hash misses:          %lu\n"),
 422                    c->half_hash_miss_count);
 423   printf_filtered (_("    Hash table population:     "));
 424   print_percentage (occupied_buckets, c->num_buckets);
 425   printf_filtered (_("    Median hash chain length:  %3d\n"),
 426                    median_chain_length);
 427   printf_filtered (_("    Average hash chain length: "));
 428   if (c->num_buckets > 0)
 429     printf_filtered ("%3lu\n", c->unique_count / c->num_buckets);
 430   else
 431     /* i18n: "Average hash chain length: (not applicable)" */
 432     printf_filtered (_("(not applicable)\n"));
 433   printf_filtered (_("    Maximum hash chain length: %3d\n"), max_chain_length);
 434   printf_filtered ("\n");
 435 }
 436
 437 int
 438 bcache_memory_used (struct bcache *bcache)
 439 {
 440   return obstack_memory_used (&bcache->cache);
 441 }