source/lib/ldb/common/qsort.c

   1 /* Copyright (C) 1991,1992,1996,1997,1999,2004 Free Software Foundation, Inc.
   2    This file is part of the GNU C Library.
   3    Written by Douglas C. Schmidt (schmidt@ics.uci.edu).
   4
   5    The GNU C Library is free software; you can redistribute it and/or
   6    modify it under the terms of the GNU Lesser General Public
   7    License as published by the Free Software Foundation; either
   8    version 2.1 of the License, or (at your option) any later version.
   9
  10    The GNU C Library is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13    Lesser General Public License for more details.
  14
  15    You should have received a copy of the GNU Lesser General Public
  16    License along with the GNU C Library; if not, write to the Free
  17    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  18    02111-1307 USA.  */
  19
  20 /* If you consider tuning this algorithm, you should consult first:
  21    Engineering a sort function; Jon Bentley and M. Douglas McIlroy;
  22    Software - Practice and Experience; Vol. 23 (11), 1249-1265, 1993.  */
  23
  24 /* Modified to be used in samba4 by
  25  * Simo Sorce <idra@samba.org>          2005
  26  */
  27
  28 #include "includes.h"
  29 #include "ldb/include/includes.h"
  30
  31 /* Byte-wise swap two items of size SIZE. */
  32 #define SWAP(a, b, size)                                                      \
  33   do                                                                          \
  34     {                                                                         \
  35       register size_t __size = (size);                                        \
  36       register char *__a = (a), *__b = (b);                                   \
  37       do                                                                      \
  38         {                                                                     \
  39           char __tmp = *__a;                                                  \
  40           *__a++ = *__b;                                                      \
  41           *__b++ = __tmp;                                                     \
  42         } while (--__size > 0);                                               \
  43     } while (0)
  44
  45 /* Discontinue quicksort algorithm when partition gets below this size.
  46    This particular magic number was chosen to work best on a Sun 4/260. */
  47 #define MAX_THRESH 4
  48
  49 /* Stack node declarations used to store unfulfilled partition obligations. */
  50 typedef struct
  51   {
  52     char *lo;
  53     char *hi;
  54   } stack_node;
  55
  56 /* The next 4 #defines implement a very fast in-line stack abstraction. */
  57 /* The stack needs log (total_elements) entries (we could even subtract
  58    log(MAX_THRESH)).  Since total_elements has type size_t, we get as
  59    upper bound for log (total_elements):
  60    bits per byte (CHAR_BIT) * sizeof(size_t).  */
  61 #ifndef CHAR_BIT
  62 #define CHAR_BIT 8
  63 #endif
  64 #define STACK_SIZE      (CHAR_BIT * sizeof(size_t))
  65 #define PUSH(low, high) ((void) ((top->lo = (low)), (top->hi = (high)), ++top))
  66 #define POP(low, high)  ((void) (--top, (low = top->lo), (high = top->hi)))
  67 #define STACK_NOT_EMPTY (stack < top)
  68
  69
  70 /* Order size using quicksort.  This implementation incorporates
  71    four optimizations discussed in Sedgewick:
  72
  73    1. Non-recursive, using an explicit stack of pointer that store the
  74       next array partition to sort.  To save time, this maximum amount
  75       of space required to store an array of SIZE_MAX is allocated on the
  76       stack.  Assuming a 32-bit (64 bit) integer for size_t, this needs
  77       only 32 * sizeof(stack_node) == 256 bytes (for 64 bit: 1024 bytes).
  78       Pretty cheap, actually.
  79
  80    2. Chose the pivot element using a median-of-three decision tree.
  81       This reduces the probability of selecting a bad pivot value and
  82       eliminates certain extraneous comparisons.
  83
  84    3. Only quicksorts TOTAL_ELEMS / MAX_THRESH partitions, leaving
  85       insertion sort to order the MAX_THRESH items within each partition.
  86       This is a big win, since insertion sort is faster for small, mostly
  87       sorted array segments.
  88
  89    4. The larger of the two sub-partitions is always pushed onto the
  90       stack first, with the algorithm then concentrating on the
  91       smaller partition.  This *guarantees* no more than log (total_elems)
  92       stack size is needed (actually O(1) in this case)!  */
  93
  94 void ldb_qsort (void *const pbase, size_t total_elems, size_t size,
  95                 void *opaque, ldb_qsort_cmp_fn_t cmp)
  96 {
  97   register char *base_ptr = (char *) pbase;
  98
  99   const size_t max_thresh = MAX_THRESH * size;
 100
 101   if (total_elems == 0)
 102     /* Avoid lossage with unsigned arithmetic below.  */
 103     return;
 104
 105   if (total_elems > MAX_THRESH)
 106     {
 107       char *lo = base_ptr;
 108       char *hi = &lo[size * (total_elems - 1)];
 109       stack_node stack[STACK_SIZE];
 110       stack_node *top = stack;
 111
 112       PUSH (NULL, NULL);
 113
 114       while (STACK_NOT_EMPTY)
 115         {
 116           char *left_ptr;
 117           char *right_ptr;
 118
 119           /* Select median value from among LO, MID, and HI. Rearrange
 120              LO and HI so the three values are sorted. This lowers the
 121              probability of picking a pathological pivot value and
 122              skips a comparison for both the LEFT_PTR and RIGHT_PTR in
 123              the while loops. */
 124
 125           char *mid = lo + size * ((hi - lo) / size >> 1);
 126
 127           if ((*cmp) ((void *) mid, (void *) lo, opaque) < 0)
 128             SWAP (mid, lo, size);
 129           if ((*cmp) ((void *) hi, (void *) mid, opaque) < 0)
 130             SWAP (mid, hi, size);
 131           else
 132             goto jump_over;
 133           if ((*cmp) ((void *) mid, (void *) lo, opaque) < 0)
 134             SWAP (mid, lo, size);
 135         jump_over:;
 136
 137           left_ptr  = lo + size;
 138           right_ptr = hi - size;
 139
 140           /* Here's the famous ``collapse the walls'' section of quicksort.
 141              Gotta like those tight inner loops!  They are the main reason
 142              that this algorithm runs much faster than others. */
 143           do
 144             {
 145               while ((*cmp) ((void *) left_ptr, (void *) mid, opaque) < 0)
 146                 left_ptr += size;
 147
 148               while ((*cmp) ((void *) mid, (void *) right_ptr, opaque) < 0)
 149                 right_ptr -= size;
 150
 151               if (left_ptr < right_ptr)
 152                 {
 153                   SWAP (left_ptr, right_ptr, size);
 154                   if (mid == left_ptr)
 155                     mid = right_ptr;
 156                   else if (mid == right_ptr)
 157                     mid = left_ptr;
 158                   left_ptr += size;
 159                   right_ptr -= size;
 160                 }
 161               else if (left_ptr == right_ptr)
 162                 {
 163                   left_ptr += size;
 164                   right_ptr -= size;
 165                   break;
 166                 }
 167             }
 168           while (left_ptr <= right_ptr);
 169
 170           /* Set up pointers for next iteration.  First determine whether
 171              left and right partitions are below the threshold size.  If so,
 172              ignore one or both.  Otherwise, push the larger partition's
 173              bounds on the stack and continue sorting the smaller one. */
 174
 175           if ((size_t) (right_ptr - lo) <= max_thresh)
 176             {
 177               if ((size_t) (hi - left_ptr) <= max_thresh)
 178                 /* Ignore both small partitions. */
 179                 POP (lo, hi);
 180               else
 181                 /* Ignore small left partition. */
 182                 lo = left_ptr;
 183             }
 184           else if ((size_t) (hi - left_ptr) <= max_thresh)
 185             /* Ignore small right partition. */
 186             hi = right_ptr;
 187           else if ((right_ptr - lo) > (hi - left_ptr))
 188             {
 189               /* Push larger left partition indices. */
 190               PUSH (lo, right_ptr);
 191               lo = left_ptr;
 192             }
 193           else
 194             {
 195               /* Push larger right partition indices. */
 196               PUSH (left_ptr, hi);
 197               hi = right_ptr;
 198             }
 199         }
 200     }
 201
 202   /* Once the BASE_PTR array is partially sorted by quicksort the rest
 203      is completely sorted using insertion sort, since this is efficient
 204      for partitions below MAX_THRESH size. BASE_PTR points to the beginning
 205      of the array to sort, and END_PTR points at the very last element in
 206      the array (*not* one beyond it!). */
 207
 208 #define min(x, y) ((x) < (y) ? (x) : (y))
 209
 210   {
 211     char *const end_ptr = &base_ptr[size * (total_elems - 1)];
 212     char *tmp_ptr = base_ptr;
 213     char *thresh = min(end_ptr, base_ptr + max_thresh);
 214     register char *run_ptr;
 215
 216     /* Find smallest element in first threshold and place it at the
 217        array's beginning.  This is the smallest array element,
 218        and the operation speeds up insertion sort's inner loop. */
 219
 220     for (run_ptr = tmp_ptr + size; run_ptr <= thresh; run_ptr += size)
 221       if ((*cmp) ((void *) run_ptr, (void *) tmp_ptr, opaque) < 0)
 222         tmp_ptr = run_ptr;
 223
 224     if (tmp_ptr != base_ptr)
 225       SWAP (tmp_ptr, base_ptr, size);
 226
 227     /* Insertion sort, running from left-hand-side up to right-hand-side.  */
 228
 229     run_ptr = base_ptr + size;
 230     while ((run_ptr += size) <= end_ptr)
 231       {
 232         tmp_ptr = run_ptr - size;
 233         while ((*cmp) ((void *) run_ptr, (void *) tmp_ptr, opaque) < 0)
 234           tmp_ptr -= size;
 235
 236         tmp_ptr += size;
 237         if (tmp_ptr != run_ptr)
 238           {
 239             char *trav;
 240
 241             trav = run_ptr + size;
 242             while (--trav >= run_ptr)
 243               {
 244                 char c = *trav;
 245                 char *hi, *lo;
 246
 247                 for (hi = lo = trav; (lo -= size) >= tmp_ptr; hi = lo)
 248                   *hi = *lo;
 249                 *hi = c;
 250               }
 251           }
 252       }
 253   }
 254 }