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 #ifdef STON_HT_FREAD
  49 #include <stdio.h>
  50 #include <errno.h>
  51 #include <alloca.h>
  52 STON_FUNC_STATIC
  53 STON_FUNC_NOINLINE
  54 ston_ht   ston_ht32_fread(FILE*,long,void*(*)(size_t));
  55 #else
  56 #include <stddef.h>
  57 #endif //STON_HT_FREAD
  58 #include <stdint.h>
  59 #include <string.h> //mem*
  60 /* STON Hashtable Structure
  61    Hashtables are stored as dynamically sized two dimensional arrays
  62 */
  63 typedef struct ston_ht_header_t
  64 { uint16_t ht_columns;
  65   uint8_t  ht_2pow, ht_flags;
  66 }ston_ht_h,* ston_ht;
  67 #define   STON_HT_HEADERSIZE (sizeof(struct ston_ht_header_t))
  68
  69 STON_FUNC
  70 uint32_t  ston_up2pow(uint32_t);
  71 STON_FUNC
  72 uint8_t   ston_trailing0(uint32_t);
  73 STON_FUNC
  74 ston_ht   ston_ht32_create(struct ston_ht_header_t,void*(*)(size_t));
  75 STON_FUNC
  76 uint32_t* ston_ht32_row(ston_ht,uint32_t);
  77 STON_FUNC
  78 uint32_t  ston_ht32_insert(ston_ht,uint32_t,uint16_t,uint32_t);
  79 STON_FUNC
  80 size_t    ston_ht32_insertx(ston_ht,uint32_t,uint32_t*,size_t,size_t);
  81
  82 #define   ston_ht32_new(_COL,_N,_F,_FN) (ston_ht32_create((ston_ht_h){_COL,ston_trailing0(ston_up2pow(_N << 1)),_F},_FN))
  83 #define   ston_ht32_entry(_HT,_KEY,_COL) (ston_ht32_row(_HT,_KEY) + _COL)
  84 #define   ston_ht_size(_HT)            ((_HT)->ht_columns << (_HT)->ht_2pow)
  85 #define   ston_ht_rows(_HT)            (0x1 << (_HT)->ht_2pow)
  86 #define   ston_ht_cols(_HT)            ((_HT)->ht_columns)
  87 #define   ston_ht_start(_HT)           ((uint8_t*)((_HT) + 1))
  88 #define   ston_ht_keyrow(_HT,_KEY)     ((_KEY) & (ston_ht_rows(ht) - 1))
  89 #define   ston_ht32_start(_HT)         ((uint32_t*)ston_ht_start(_HT))
  90 #define   ston_ht32_end(_HT)           (ston_ht32_start(_HT) + ston_ht_size(_HT))
  91 #define   ston_ht32_size(_HT)          (ston_ht_size(_HT) * sizeof(uint32_t))
  92
  93 /** @see http://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2 */
  94 STON_FUNC
  95 uint32_t ston_up2pow
  96 ( uint32_t val )
  97 { val = (val << 1) - 1;
  98   val |= val >> 1;
  99   val |= val >> 2;
 100   val |= val >> 4;
 101   val |= val >> 8;
 102   val |= val >> 16;
 103   return ++val;
 104 }
 105
 106 /** @see https://graphics.stanford.edu/~seander/bithacks.html#ZerosOnRightParallel */
 107 STON_FUNC
 108 uint8_t ston_trailing0
 109 ( uint32_t v )
 110 { uint8_t c = 32;
 111   v &= -(int32_t)v;
 112   if (v) c--;
 113   if (v & 0x0000FFFF) c -= 16;
 114   if (v & 0x00FF00FF) c -= 8;
 115   if (v & 0x0F0F0F0F) c -= 4;
 116   if (v & 0x33333333) c -= 2;
 117   if (v & 0x55555555) c -= 1;
 118   return c;
 119 }
 120
 121 /* Creates a new hash table, provided a memory allocation function that takes a
 122    single size_t bytes, a column count, and a row count which determines the
 123    size of the table.
 124
 125    use ston_ht32_new to specify the exact or estimated number of unique keys
 126    held in the table.  With ston_ht32_new, the provided ht_rows is doubled, and
 127    rounded up to the nearest power of two to create a hash table with minimal
 128    collisions.
 129 */
 130 STON_FUNC
 131 ston_ht ston_ht32_create
 132 ( struct ston_ht_header_t ht_header,
 133   void*                   (*alloc_fn)(size_t)
 134 )
 135 { size_t  ht_bytes = ston_ht32_size(&ht_header);
 136   ston_ht ht = (ston_ht) alloc_fn(STON_HT_HEADERSIZE + ht_bytes);
 137   if (ht != NULL)
 138     { memcpy(ht,&ht_header,STON_HT_HEADERSIZE);
 139       memset(ht + 1, 0, ht_bytes);
 140     }
 141   return ht;
 142 }
 143
 144 #ifdef STON_HT_FREAD
 145 /* Reads a 32-bit hash table out of the provided file at the provide fpos, into
 146    a buffer allocated by alloc_fn.  Memory is allocated to the stack until the
 147    entire structure is verified, and all file operations are finished.
 148    Returns NULL with properly set errno on failure.
 149 */
 150 ston_ht ston_ht32_fread
 151 ( FILE* file,
 152   long  fpos,
 153   void* (*alloc_fn)(size_t)
 154 )
 155 { struct ston_ht_header_t header;
 156   ston_ht                 stack_ht, ht;
 157   long                    fpos_start;
 158   size_t                  table_size, alloc_size;
 159   int                     errno_local;
 160   if ((fpos_start = ftell(file)) == -1)
 161     return NULL;
 162   if (fread(&header, sizeof(header), 1, file) != 1)
 163     goto fail_seekback;
 164   table_size = ston_ht32_size(&header);
 165   alloc_size = sizeof(header) + table_size;
 166   stack_ht = (ston_ht) alloca(alloc_size);
 167   memcpy(stack_ht, &header, sizeof(header));
 168   if (fread(stack_ht + sizeof(header), table_size, 1, file) != 1)
 169     goto fail_seekback;
 170   if (fseek(file, fpos_start, SEEK_SET) != 0)
 171     return NULL;
 172   ht = (ston_ht) alloc_fn(alloc_size);
 173   if (ht != NULL)
 174     memcpy(ht, stack_ht, alloc_size);
 175   return ht;
 176  fail_seekback:
 177   /* Try to seek the file back to origin without clobbering errno */
 178   errno_local = errno;
 179   fseek(file, fpos_start, SEEK_SET);
 180   errno = errno_local;
 181   return NULL;
 182 }
 183 #endif
 184
 185 /* Returns a pointer to the row of data in the hashtable containing the provided
 186    key, inserts if not found.  Returns NULL on overflow.
 187 */
 188 STON_FUNC
 189 uint32_t* ston_ht32_row
 190 ( struct ston_ht_header_t* ht,
 191   uint32_t                key
 192 )
 193 { uint32_t* row;
 194   uint32_t* row_start = ston_ht32_start(ht);
 195   uint32_t* row_end = ston_ht32_end(ht);
 196   uint16_t  ht_cols = ston_ht_cols(ht);
 197   size_t    row_number = ston_ht_keyrow(ht,key);
 198   uint8_t   looped = 0;
 199   row = row_start + (row_number * ht_cols);
 200  next_row:
 201   if (row[0] != 0)
 202     goto populated;
 203  write_position:
 204   row[0] = key;
 205   return row;
 206  populated:
 207   if (row[0] == key)
 208     goto write_position;
 209   if (row + ht_cols < row_end)
 210     row += ht_cols;
 211   else if (looped)
 212       return NULL;
 213   else
 214     { looped++;
 215       row = row_start;
 216     }
 217   goto next_row;
 218 }
 219
 220 /* Inserts a value into a hashtable at the specified column, returning the
 221    previous value */
 222 STON_FUNC
 223 uint32_t ston_ht32_insert
 224 ( struct ston_ht_header_t* ht,
 225   uint32_t                key,
 226   uint16_t                column,
 227   uint32_t                value
 228 )
 229 { uint32_t* value_location, old_value;
 230   value_location = ston_ht32_entry(ht,key,column);
 231   old_value = *value_location;
 232   *value_location = value;
 233   return old_value;
 234 }
 235
 236 /* Inserts a row of units into a hashtable, starting with the specified column.
 237    Returns the number of elements that were written.  This function will not
 238    overflow internal buffers, but will return a short count (lower than the
 239    provided 'units') when truncation of source data occurs. */
 240 STON_FUNC
 241 size_t
 242 ston_ht32_insertx
 243 ( struct ston_ht_header_t* ht,
 244   uint32_t                 key,
 245   uint32_t*                data_src,
 246   size_t                   start_column,
 247   size_t                   units
 248 )
 249 { uint32_t* data_row = ston_ht32_row(ht,key);
 250   uint32_t* data_limit = data_row + ston_ht_cols(ht);
 251   uint32_t* data_trg = data_row + start_column;
 252   if (data_row == NULL)
 253     return 0;
 254   while (units-- && data_trg < data_limit)
 255     *data_trg++ = *data_src++;
 256   return (size_t)(data_trg - data_row);
 257 }
 258
 259
 260 #ifndef STON_DHT_SIZE
 261 #define STON_DHT_SIZE 4096
 262 #endif
 263
 264 /* STON Dynamic Hashtable Structure
 265    A dynamic form of the generic hashtable implementation above which uses
 266    external allocation.
 267 */
 268 typedef struct ston_dht_header_t
 269 { uint16_t  ht_columns;
 270   uint8_t   ht_2pow, ht_flags;
 271   void*     (*ht_alloc)(size_t);
 272   void      (*ht_free)(void*);
 273   void**    page_head;
 274 }ston_dht_h,* ston_dht;
 275 #define   STON_DHT_HEADERSIZE (sizeof(struct ston_dht_header_t))
 276
 277 STON_FUNC
 278 ston_dht  ston_dht32_create(struct ston_ht_header_t,void*(*)(size_t),void(*)(void*));
 279 STON_FUNC
 280 uint32_t* ston_dht32_row(ston_dht,uint32_t);
 281 STON_FUNC
 282 uint32_t  ston_dht32_insert(ston_dht,uint32_t,uint16_t,uint32_t);
 283 STON_FUNC
 284 size_t    ston_dht32_insertx(ston_dht,uint32_t,uint32_t*,size_t,size_t);
 285 STON_FUNC
 286 ston_dht  ston_dht32_free(ston_dht);
 287
 288 #define   ston_dht32_new(_COL,_N,_F,_ALLOC,_FREE) (ston_dht32_create((ston_ht_h){_COL,ston_trailing0(ston_up2pow(_N << 1)),_F},_ALLOC,_FREE))
 289 #define   ston_dht32_entry(_HT,_KEY,_COL)        (ston_dht32_row(_HT,_KEY) + _COL)
 290 #define   ston_dht_size(_HT)           (ston_ht_size(_HT))
 291 #define   ston_dht_rows(_HT)           (ston_ht_rows(_HT))
 292 #define   ston_dht_cols(_HT)           (ston_ht_cols(_HT))
 293 #define   ston_dht_keyrow(_HT,_KEY)    (ston_ht_keyrow(_HT,_KEY))
 294 #define   ston_dht_pagestart(_HT)      ((void**)(((uint8_t*)(_HT)) + STON_DHT_HEADERSIZE))
 295 #define   ston_dht_pagehead(_HT)       ((_HT)->page_head)
 296 #define   ston_dht_pagemax(_HT)        ((void**)((uint8_t*)(_HT) + STON_DHT_SIZE - sizeof(void**)))
 297 #define   ston_dht_start(_HT,_DEPTH)   ((uint8_t*)*(ston_dht_pagestart(_HT) + _DEPTH))
 298 #define   ston_dht32_start(_HT,_DEPTH) ((uint32_t*)ston_dht_start(_HT,_DEPTH))
 299 #define   ston_dht32_end(_HT,_DEPTH)   (ston_ht32_start(_HT,_DEPTH) + ston_ht_size(_HT))
 300 #define   ston_dht32_size(_HT)         (ston_dht_size(_HT) * sizeof(uint32_t))
 301 #define   ston_dht32_pagepush(_HT)     ((*(++((_HT)->page_head)) = (_HT)->ht_alloc(ston_dht32_size(_HT))))
 302 #define   ston_dht32_pagepop(_HT)      ((_HT)->ht_free((_HT)->page_head--))
 303
 304 /* Creates a new bucketted hash table, provided a memory allocation function
 305    that takes a single size_t bytes, a memory free function, a column count, and
 306    a row count which determines the size of the buckets.
 307 */
 308 STON_FUNC
 309 ston_dht ston_dht32_create
 310 ( struct ston_ht_header_t ht_header,
 311   void*                   (*ht_alloc)(size_t),
 312   void                    (*ht_free)(void*)
 313 )
 314 { size_t   ht_bytes = ston_dht32_size(&ht_header);
 315   ston_dht ht = (ston_dht) ht_alloc(STON_DHT_SIZE);
 316   if (ht != NULL)
 317     { memcpy(ht, &ht_header, sizeof(ht_header));
 318       ht->ht_alloc = ht_alloc;
 319       ht->ht_free = ht_free;
 320       ht->page_head = ston_dht_pagestart(ht);
 321       if ((*(ht->page_head) = ht->ht_alloc(ht_bytes)) == NULL)
 322         if (ht_free != NULL)
 323           ht_free(ht);
 324     }
 325   return ht;
 326 }
 327
 328 /* Returns a pointer to the row of data in the hashtable containing the provided
 329    key, inserts if not found.  Returns NULL on overflow.
 330 */
 331 STON_FUNC
 332 uint32_t* ston_dht32_row
 333 ( struct ston_dht_header_t* ht,
 334   uint32_t                  key
 335 )
 336 { uint16_t   ht_cols = ston_dht_cols(ht);
 337   size_t     row_number = ston_dht_keyrow(ht,key);
 338   uint32_t** page = (uint32_t**)ston_dht_pagestart(ht);
 339   uint32_t** pagemax = (uint32_t**)ston_dht_pagemax(ht);
 340   uint8_t    loop_x = 0;
 341   uint8_t    loop_y = 0;
 342   uint32_t*  row,* row_end;
 343  next_page:
 344   row = *page + (row_number * ht_cols);
 345   row_end = *page + (ston_dht_size(ht) - 1);
 346  next_row:
 347   if (row[0] != 0)
 348     goto populated;
 349  write_position:
 350   row[0] = key;
 351   return row;
 352  populated:
 353   if (row[0] == key)
 354     goto write_position;
 355   if (!loop_x)
 356     { if (page < pagemax)
 357         { if (page == (uint32_t**)ston_dht_pagehead(ht))
 358             if (ston_dht32_pagepush(ht) == NULL)
 359               { ston_dht32_free(ht);
 360                 return NULL;
 361               }
 362           ++page;
 363           goto next_row;
 364         }
 365       loop_x = 1;
 366       row_number = (row_number + 1) % ston_dht_rows(ht);
 367       page = (uint32_t**)ston_dht_pagestart(ht);
 368       goto next_row;
 369     }
 370   if (row + ht_cols < row_end)
 371     { row += ht_cols;
 372       goto next_row;
 373     }
 374   else if (!loop_y)
 375     { loop_y = 1;
 376       row = *page;
 377       goto next_row;
 378     }
 379   if (page < pagemax)
 380     { loop_y = 0;
 381       page++;
 382       goto next_page;
 383     }
 384   return NULL;
 385 }
 386
 387 /* Inserts a value into a hashtable at the specified column, returning the
 388    previous value */
 389 STON_FUNC
 390 uint32_t ston_dht32_insert
 391 ( struct ston_dht_header_t* ht,
 392   uint32_t                  key,
 393   uint16_t                  column,
 394   uint32_t                  value
 395 )
 396 { uint32_t* value_location, old_value;
 397   value_location = ston_dht32_entry(ht,key,column);
 398   old_value = *value_location;
 399   *value_location = value;
 400   return old_value;
 401 }
 402
 403 /* Free the dynamic hash table */
 404 STON_FUNC
 405 struct ston_dht_header_t* ston_dht32_free
 406 ( struct ston_dht_header_t* ht )
 407 { void (*ht_free)(void*) = ht->ht_free;
 408   if (ht_free != NULL)
 409     { while (ht->page_head >= ston_dht_pagestart(ht))
 410         { ht_free(*(ht->page_head));
 411           ht->page_head--;
 412         }
 413       ht_free(ht);
 414       return NULL;
 415     }
 416   return ht;
 417 }
 418
 419 /* Insert multiple values, returning the number of bytes written */
 420 STON_FUNC
 421 size_t
 422 ston_dht32_insertx
 423 ( struct ston_dht_header_t* ht,
 424   uint32_t                  key,
 425   uint32_t*                 data_src,
 426   size_t                    start_column,
 427   size_t                    units
 428 )
 429 { uint32_t* data_row = ston_dht32_row(ht,key);
 430   uint32_t* data_limit = data_row + ston_dht_cols(ht);
 431   uint32_t* data_trg = data_row + start_column;
 432   if (data_row == NULL)
 433     return 0;
 434   while (units-- && data_trg < data_limit)
 435     *data_trg++ = *data_src++;
 436   return (size_t)(data_trg - data_row);
 437 }
 438
 439
 440 #endif //_STON_HT_H_