ston/ston_ht.h

   1 /*!@file
   2   \brief   STON Hash Tables
   3   \details Aligned general purpose hash functions and memory definitions
   4            whose columns are provided, and whose rows, and sizes, are derived.
   5
   6              ht_size = header.ht_columns << header.ht_2pow;
   7              ht_rows = 0x1 << header.ht_2pow;
   8
   9            All generic hashtables in henge must have a power-of-two number of
  10            rows.  An ht_columns value that is also a power-of-two will result in
  11            a power-of-two sized memory imprint for the structure, making it easy
  12            to page align.
  13
  14            Elements in the columns may be of any arbitrary size.
  15
  16              typedef uint32_t my_ht_type;
  17              ht_bytes = ht_size * sizeof(my_ht_type);
  18
  19            implementation covers only 32-bit unit sizes.
  20
  21   \author  Ken Grimes
  22   \date    Feb 2017
  23   ----------------------------------------------------------------------------*/
  24 #ifndef _STON_HT_T_
  25 #define _STON_HT_T_
  26 /* Define STON_NOSTATIC to expose included function symbols */
  27 #ifndef STON_NOSTATIC
  28 #define STON_FUNC_STATIC static
  29 #else
  30 #define STON_FUNC_STATIC
  31 #endif //STON_NOSTATIC
  32 /* If GNUC is detected, uses attributes to stop inlining */
  33 #ifdef  __GNUC__
  34 #define STON_FUNC_NOINLINE __attribute__ ((noinline))
  35 #else
  36 #define STON_FUNC_NOINLINE
  37 #endif //__GNUC__
  38 /* Define STON_NOINLINE to prevent inline compiler hints */
  39 #ifndef STON_NOINLINE
  40 #define STON_FUNC_INLINE inline
  41 #else
  42 #define STON_FUNC_INLINE
  43 #endif //STON_NOINLINE
  44 /* Define STON_FUNC to override the default STON Function attributes */
  45 #ifndef STON_FUNC
  46 #define STON_FUNC STON_FUNC_STATIC STON_FUNC_INLINE
  47 #endif //STON_FUNC
  48 #ifndef STON_NOSTDIO
  49 #include <stdio.h>
  50 #include <string.h> //memcpy
  51 #include <alloca.h>
  52 #endif //STON_NOSTDIO
  53 #include <stdint.h>
  54 /* STON Hashtable Structure
  55    Hashtables are stored as dynamically sized two dimensional arrays
  56 */
  57 typedef struct ston_ht_header_t
  58 { uint16_t ht_columns;
  59   uint8_t  ht_2pow, ht_flags;
  60 }* ston_ht;
  61
  62 STON_FUNC
  63 size_t    ston_up2pow(size_t);
  64 STON_FUNC_STATIC
  65 STON_FUNC_NOINLINE
  66 ston_ht   ston_ht32_fread(FILE*,long,void*(*)(size_t));
  67 STON_FUNC
  68 ston_ht   ston_ht32_create(uint16_t,size_t,uint8_t,void*(*)(size_t));
  69 STON_FUNC
  70 uint32_t* ston_ht32_row(ston_ht,uint32_t);
  71 STON_FUNC
  72 uint32_t  ston_ht32_insert(ston_ht,uint32_t,uint16_t,uint32_t);
  73
  74 #define   ston_ht32_new(_COL,_N,_F,_FN) ston_ht32_create(_COLS,ston_up2pow(_N << 1),_F,_FN)
  75 #define   ston_ht32_entry(_HT,_KEY,_COL)        (ston_ht32_row(_HT,_KEY) + _COL)
  76 #define   ston_ht32_insertx(_HT,_KEY,_COL,_VAL) *ston_ht32_entry(_HT,_KEY,_COL) = _VAL
  77 #define   ston_ht_size(_HT)            ((_HT)->ht_columns << (_HT)->ht_2pow)
  78 #define   ston_ht_rows(_HT)            (0x1 << (_HT)->ht_2pow)
  79 #define   ston_ht_cols(_HT)            ((_HT)->ht_columns)
  80 #define   ston_ht_start(_HT)           (((uint8_t*)(_HT)) + sizeof(*(_HT)))
  81 #define   ston_ht_keyrow(_HT,_KEY)     ((_KEY) & (ston_ht_rows(ht) - 1))
  82 #define   ston_ht32_start(_HT)         ((uint32_t*)ston_ht_start(_HT))
  83 #define   ston_ht32_end(_HT)           (ston_ht32_start(_HT) + ston_ht_size(_HT))
  84 #define   ston_ht32_size(_HT)          (ston_ht_size(_HT) * sizeof(uint32_t))
  85
  86 /** @see http://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2 */
  87 STON_FUNC
  88 size_t ston_up2pow
  89 ( size_t val )
  90 { val = (val << 1) - 1;
  91   val |= val >> 1;
  92   val |= val >> 2;
  93   val |= val >> 4;
  94   val |= val >> 8;
  95   val |= val >> 16;
  96   return ++val;
  97 }
  98
  99 /* Creates a new hash table, provided a memory allocation function that takes a
 100    single size_t bytes, a column count, and a row count which determines the
 101    size of the table.
 102
 103    use ston_ht32_new to specify the exact or estimated number of unique keys
 104    held in the table.  With ston_ht32_new, the provided ht_rows is doubled, and
 105    rounded up to the nearest power of two to create a hash table with minimal
 106    collisions.
 107 */
 108 STON_FUNC
 109 ston_ht ston_ht32_create
 110 ( uint16_t ht_columns,
 111   size_t   ht_rows,
 112   uint8_t  ht_flags,
 113   void*    (*alloc_fn)(size_t)
 114 )
 115 { size_t  ht_size = ht_rows * ht_columns * sizeof(uint32_t);
 116   ston_ht ht = (ston_ht) alloc_fn(sizeof(struct ston_ht_header_t) + ht_size);
 117   if (ht != NULL)
 118     { for (ht->ht_2pow = 0; ht_size; ht->ht_2pow++)
 119         ht_size = ht_size >> 1;
 120       ht->ht_columns = ht_columns;
 121       ht->ht_flags = ht_flags;
 122     }
 123   return ht;
 124 }
 125
 126 #ifndef STON_NO_STDIO
 127 /* Reads a 32-bit hash table out of the provided file at the provide fpos, into
 128    a buffer allocated by alloc_fn.  Memory is allocated to the stack until the
 129    entire structure is verified, and all file operations are finished.
 130    Returns NULL with properly set errno on failure.
 131 */
 132 STON_FUNC_STATIC
 133 STON_FUNC_NOINLINE
 134 ston_ht ston_ht32_fread
 135 ( FILE* file,
 136   long  fpos,
 137   void* (*alloc_fn)(size_t)
 138 )
 139 { struct ston_ht_header_t header;
 140   ston_ht                 stack_ht, ht;
 141   long                    fpos_start;
 142   size_t                  table_size, alloc_size;
 143   int                     errno_local;
 144   if ((fpos_start = ftell(file)) == -1)
 145     return NULL;
 146   if (fread(&header, sizeof(header), 1, file) != 1)
 147     goto fail_seekback;
 148   table_size = ston_ht32_size(&header);
 149   alloc_size = sizeof(header) + table_size;
 150   stack_ht = (ston_ht) alloca(alloc_size);
 151   memcpy(stack_ht, &header, sizeof(header));
 152   if (fread(stack_ht + sizeof(header), table_size, 1, file) != 1)
 153     goto fail_seekback;
 154   if (fseek(file, fpos_start, SEEK_SET) != 0)
 155     return NULL;
 156   ht = (ston_ht) alloc_fn(alloc_size);
 157   if (ht != NULL)
 158     memcpy(ht, stack_ht, alloc_size);
 159   return ht;
 160  fail_seekback:
 161   /* Try to seek the file back to origin without clobbering errno */
 162   errno_local = errno;
 163   fseek(file, fpos_start, SEEK_SET);
 164   errno = errno_local;
 165   return NULL;
 166 }
 167 #endif
 168
 169 /* Returns a pointer to the row of data in the hashtable containing the provided
 170    key, inserts if not found.  Returns NULL on overflow.
 171 */
 172 STON_FUNC
 173 uint32_t* ston_ht32_row
 174 ( struct ston_ht_header_t* ht,
 175   uint32_t                key
 176 )
 177 { uint32_t* row;
 178   uint32_t* row_start = ston_ht32_start(ht);
 179   uint32_t* row_end = ston_ht32_end(ht);
 180   uint16_t  ht_cols = ston_ht_cols(ht);
 181   size_t    row_number = ston_ht_keyrow(ht,key);
 182   uint8_t   looped = 0;
 183   row = row_start + (row_number * ht_cols);
 184  next_row:
 185   if (row[0] != 0)
 186     goto populated;
 187  write_position:
 188   row[0] = key;
 189   return row;
 190  populated:
 191   if (row[0] == key)
 192     goto write_position;
 193   if (row < row_end)
 194     row += ht_cols;
 195   else if (looped)
 196       return NULL;
 197   else
 198     { looped++;
 199       row = row_start;
 200     }
 201   goto next_row;
 202 }
 203
 204 /* Inserts a value into a hashtable at the specified column, returning the
 205    previous value */
 206 STON_FUNC
 207 uint32_t ston_ht32_insert
 208 ( struct ston_ht_header_t* ht,
 209   uint32_t                key,
 210   uint16_t                column,
 211   uint32_t                value
 212 )
 213 { uint32_t* value_location, old_value;
 214   value_location = ston_ht32_entry(ht,key,column);
 215   old_value = *value_location;
 216   *value_location = value;
 217   return old_value;
 218 }
 219
 220 #ifndef STON_DHT_SIZE
 221 #define STON_DHT_SIZE 4096
 222 #endif
 223
 224 /* STON Dynamic Hashtable Structure
 225    A dynamic form of the generic hashtable implementation above which uses
 226    external allocation.
 227 */
 228 typedef struct ston_dht_header_t
 229 { uint16_t  ht_columns;
 230   uint8_t   ht_2pow, ht_flags;
 231   void*     (*ht_alloc)(size_t);
 232   void      (*ht_free)(void*);
 233   void**    page_head;
 234 }* ston_dht;
 235 #define   STON_DHT_HEADERSIZE (sizeof(struct ston_dht_header_t))
 236
 237 STON_FUNC
 238 ston_dht  ston_dht32_create(uint16_t,size_t,uint8_t,void*(*)(size_t),void(*)(void*));
 239 STON_FUNC
 240 uint32_t* ston_dht32_row(ston_dht,uint32_t);
 241 STON_FUNC
 242 uint32_t  ston_dht32_insert(ston_dht,uint32_t,uint16_t,uint32_t);
 243 STON_FUNC
 244 void      ston_dht32_free(ston_dht);
 245
 246 #define   ston_dht32_new(_COL,_N,_F,_FN)         ston_dht32_create(_COLS,ston_up2pow(_N << 1),_F,_FN)
 247 #define   ston_dht32_entry(_HT,_KEY,_COL)        (ston_dht32_row(_HT,_KEY) + _COL)
 248 #define   ston_dht32_insertx(_HT,_KEY,_COL,_VAL) *ston_dht32_col(_HT,_KEY,_COL) = _VAL
 249 #define   ston_dht_size(_HT)           (ston_ht_size(_HT))
 250 #define   ston_dht_rows(_HT)           (ston_ht_rows(_HT))
 251 #define   ston_dht_cols(_HT)           (ston_ht_cols(_HT))
 252 #define   ston_dht_keyrow(_HT,_KEY)    (ston_ht_keyrow(_HT,_KEY))
 253 #define   ston_dht_pagestart(_HT)      ((void**)(((uint8_t*)(_HT)) + STON_DHT_HEADERSIZE))
 254 #define   ston_dht_pagehead(_HT)       ((_HT)->page_head)
 255 #define   ston_dht_pagemax(_HT)        ((void**)((uint8_t*)(_HT) + STON_DHT_SIZE - sizeof(void**)))
 256 #define   ston_dht_start(_HT,_DEPTH)   ((uint8_t*)*(ston_dht_pagestart(_HT) + _DEPTH))
 257 #define   ston_dht32_start(_HT,_DEPTH) ((uint32_t*)ston_dht_start(_HT,_DEPTH))
 258 #define   ston_dht32_end(_HT,_DEPTH)   (ston_ht32_start(_HT,_DEPTH) + ston_ht_size(_HT))
 259 #define   ston_dht32_size(_HT)         (ston_dht_size(_HT) * sizeof(uint32_t))
 260 #define   ston_dht32_pagepush(_HT)     ((*(++((_HT)->page_head)) = (_HT)->ht_alloc(ston_dht32_size(_HT))))
 261 #define   ston_dht32_pagepop(_HT)      ((_HT)->ht_free((_HT)->page_head--))
 262
 263 /* Creates a new bucketted hash table, provided a memory allocation function
 264    that takes a single size_t bytes, a memory free function, a column count, and
 265    a row count which determines the size of the buckets.
 266 */
 267 STON_FUNC
 268 ston_dht ston_dht32_create
 269 ( uint16_t ht_columns,
 270   size_t   ht_rows,
 271   uint8_t  ht_flags,
 272   void*   (*ht_alloc)(size_t),
 273   void    (*ht_free)(void*)
 274 )
 275 { size_t   ht_size = ht_rows * ht_columns * sizeof(uint32_t);
 276   ston_dht ht = (ston_dht) ht_alloc(STON_DHT_SIZE);
 277   if (ht != NULL)
 278     { for (ht->ht_2pow = 0; ht_size; ht->ht_2pow++)
 279         ht_size = ht_size >> 1;
 280       ht->ht_columns = ht_columns;
 281       ht->ht_flags = ht_flags;
 282       ht->ht_alloc = ht_alloc;
 283       ht->ht_free = ht_free;
 284       ht->page_head = ston_dht_pagestart(ht);
 285       if ((*(ht->page_head) = ht->ht_alloc(ston_dht_size(ht))) == NULL)
 286         ht_free(ht);
 287     }
 288   return ht;
 289 }
 290
 291 /* Returns a pointer to the row of data in the hashtable containing the provided
 292    key, inserts if not found.  Returns NULL on overflow.
 293 */
 294 STON_FUNC
 295 uint32_t* ston_dht32_row
 296 ( struct ston_dht_header_t* ht,
 297   uint32_t                  key
 298 )
 299 { uint16_t   ht_cols = ston_dht_cols(ht);
 300   size_t     row_number = ston_dht_keyrow(ht,key);
 301   uint32_t** page = (uint32_t**)ston_dht_pagestart(ht);
 302   uint32_t** pagemax = (uint32_t**)ston_dht_pagemax(ht);
 303   uint8_t    loop_x = 0;
 304   uint8_t    loop_y = 0;
 305   uint32_t*  row,* row_end;
 306  next_page:
 307   row = *page + (row_number * ht_cols);
 308   row_end = *page + (ston_dht_size(ht) - 1);
 309  next_row:
 310   if (row[0] != 0)
 311     goto populated;
 312  write_position:
 313   row[0] = key;
 314   return row;
 315  populated:
 316   if (row[0] == key)
 317     goto write_position;
 318   if (!loop_x)
 319     { if (page < pagemax)
 320         { if (page == (uint32_t**)ston_dht_pagehead(ht))
 321             if (ston_dht32_pagepush(ht) == NULL)
 322               { ston_dht32_free(ht);
 323                 return NULL;
 324               }
 325           ++page;
 326           goto next_row;
 327         }
 328       loop_x = 1;
 329       row_number = (row_number + 1) % ston_dht_rows(ht);
 330       page = (uint32_t**)ston_dht_pagestart(ht);
 331       goto next_row;
 332     }
 333   if (row < row_end)
 334     { row += ht_cols;
 335       goto next_row;
 336     }
 337   else if (!loop_y)
 338     { loop_y = 1;
 339       row = *page;
 340       goto next_row;
 341     }
 342   if (page < pagemax)
 343     { loop_y = 0;
 344       page++;
 345       goto next_page;
 346     }
 347   return NULL;
 348 }
 349
 350 /* Inserts a value into a hashtable at the specified column, returning the
 351    previous value */
 352 STON_FUNC
 353 uint32_t ston_dht32_insert
 354 ( struct ston_dht_header_t* ht,
 355   uint32_t                  key,
 356   uint16_t                  column,
 357   uint32_t                  value
 358 )
 359 { uint32_t* value_location, old_value;
 360   value_location = ston_dht32_entry(ht,key,column);
 361   old_value = *value_location;
 362   *value_location = value;
 363   return old_value;
 364 }
 365
 366 /* Free the dynamic hash table */
 367 STON_FUNC
 368 void ston_dht32_free
 369 ( struct ston_dht_header_t* ht )
 370 { void (*ht_free)(void*) = ht->ht_free;
 371   if (ht_free != NULL)
 372     { while (ht->page_head >= ston_dht_pagestart(ht))
 373         ht_free(ht->page_head--);
 374       ht_free(ht);
 375     }
 376 }
 377
 378 #endif //_STON_HT_H_