2 * UCW Library -- Universal Sorter: Common Declarations
4 * (c) 2007 Martin Mares <mj@ucw.cz>
6 * This software may be freely distributed and used according to the terms
7 * of the GNU Lesser General Public License.
10 #ifndef _UCW_SORTER_COMMON_H
11 #define _UCW_SORTER_COMMON_H
13 #include <ucw/clists.h>
15 #ifdef CONFIG_UCW_CLEAN_ABI
16 #define asort_run ucw_asort_run
17 #define asort_start_threads ucw_asort_start_threads
18 #define asort_stop_threads ucw_asort_stop_threads
19 #define sbuck_drop ucw_sbuck_drop
20 #define sbuck_has_file ucw_sbuck_has_file
21 #define sbuck_have ucw_sbuck_have
22 #define sbuck_new ucw_sbuck_new
23 #define sbuck_read ucw_sbuck_read
24 #define sbuck_size ucw_sbuck_size
25 #define sbuck_swap_out ucw_sbuck_swap_out
26 #define sbuck_write ucw_sbuck_write
27 #define sorter_add_radix_bits ucw_sorter_add_radix_bits
28 #define sorter_alloc ucw_sorter_alloc
29 #define sorter_alloc_buf ucw_sorter_alloc_buf
30 #define sorter_bufsize ucw_sorter_bufsize
31 #define sorter_debug ucw_sorter_debug
32 #define sorter_fb_params ucw_sorter_fb_params
33 #define sorter_free_buf ucw_sorter_free_buf
34 #define sorter_max_multiway_bits ucw_sorter_max_multiway_bits
35 #define sorter_max_radix_bits ucw_sorter_max_radix_bits
36 #define sorter_min_multiway_bits ucw_sorter_min_multiway_bits
37 #define sorter_min_radix_bits ucw_sorter_min_radix_bits
38 #define sorter_prepare_buf ucw_sorter_prepare_buf
39 #define sorter_radix_threshold ucw_sorter_radix_threshold
40 #define sorter_run ucw_sorter_run
41 #define sorter_small_fb_params ucw_sorter_small_fb_params
42 #define sorter_small_input ucw_sorter_small_input
43 #define sorter_stream_bufsize ucw_sorter_stream_bufsize
44 #define sorter_thread_chunk ucw_sorter_thread_chunk
45 #define sorter_thread_threshold ucw_sorter_thread_threshold
46 #define sorter_threads ucw_sorter_threads
47 #define sorter_trace ucw_sorter_trace
48 #define sorter_trace_array ucw_sorter_trace_array
51 /* Configuration variables */
52 extern uns sorter_trace, sorter_trace_array, sorter_stream_bufsize;
53 extern uns sorter_debug, sorter_min_radix_bits, sorter_max_radix_bits, sorter_add_radix_bits;
54 extern uns sorter_min_multiway_bits, sorter_max_multiway_bits;
55 extern uns sorter_threads;
56 extern u64 sorter_bufsize, sorter_small_input;
57 extern u64 sorter_thread_threshold, sorter_thread_chunk, sorter_radix_threshold;
58 extern struct fb_params sorter_fb_params, sorter_small_fb_params;
60 #define SORT_TRACE(x...) do { if (sorter_trace) msg(L_DEBUG, x); } while(0)
61 #define SORT_XTRACE(level, x...) do { if (sorter_trace >= level) msg(L_DEBUG, x); } while(0)
64 SORT_DEBUG_NO_PRESORT = 1,
65 SORT_DEBUG_NO_JOIN = 2,
66 SORT_DEBUG_KEEP_BUCKETS = 4,
67 SORT_DEBUG_NO_RADIX = 8,
68 SORT_DEBUG_NO_MULTIWAY = 16,
69 SORT_DEBUG_ASORT_NO_RADIX = 32,
70 SORT_DEBUG_ASORT_NO_THREADS = 64
76 struct fastbuf *in_fb;
77 struct fastbuf *out_fb;
80 struct fb_params *fb_params;
87 int (*custom_presort)(struct fastbuf *dest, void *buf, size_t bufsize);
89 // Take as much as possible from the source bucket, sort it in memory and dump to destination bucket.
90 // Return 1 if there is more data available in the source bucket.
91 int (*internal_sort)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket *out, struct sort_bucket *out_only);
93 // Estimate how much input data from `b' will fit in the internal sorting buffer.
94 u64 (*internal_estimate)(struct sort_context *ctx, struct sort_bucket *b);
96 // Two-way split/merge: merge up to 2 source buckets to up to 2 destination buckets.
97 // Bucket arrays are NULL-terminated.
98 void (*twoway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket **outs);
100 // Multi-way merge: merge an arbitrary number of source buckets to a single destination bucket.
101 void (*multiway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket *out);
103 // Radix split according to hash function
104 void (*radix_split)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket **outs, uns bitpos, uns numbits);
106 // State variables of internal_sort
111 timestamp_t start_time;
113 uns total_int_time, total_pre_time, total_ext_time;
116 void sorter_run(struct sort_context *ctx);
120 void *sorter_alloc(struct sort_context *ctx, uns size);
121 void sorter_prepare_buf(struct sort_context *ctx);
122 void sorter_alloc_buf(struct sort_context *ctx);
123 void sorter_free_buf(struct sort_context *ctx);
129 struct sort_context *ctx;
133 u64 size; // Size in bytes (not valid when writing)
134 uns runs; // Number of runs, 0 if not sorted
135 uns hash_bits; // Remaining bits of the hash function
136 byte *ident; // Identifier used in debug messages
139 enum sort_bucket_flags {
140 SBF_FINAL = 1, // This bucket corresponds to the final output file (always 1 run)
141 SBF_SOURCE = 2, // Contains the source file (always 0 runs)
142 SBF_CUSTOM_PRESORT = 4, // Contains source to read via custom presorter
143 SBF_OPEN_WRITE = 256, // We are currently writing to the fastbuf
144 SBF_OPEN_READ = 512, // We are reading from the fastbuf
145 SBF_DESTROYED = 1024, // Already done with, no further references allowed
146 SBF_SWAPPED_OUT = 2048, // Swapped out to a named file
149 struct sort_bucket *sbuck_new(struct sort_context *ctx);
150 void sbuck_drop(struct sort_bucket *b);
151 int sbuck_have(struct sort_bucket *b);
152 int sbuck_has_file(struct sort_bucket *b);
153 ucw_off_t sbuck_size(struct sort_bucket *b);
154 struct fastbuf *sbuck_read(struct sort_bucket *b);
155 struct fastbuf *sbuck_write(struct sort_bucket *b);
156 void sbuck_swap_out(struct sort_bucket *b);
158 /* Contexts and helper functions for the array sorter */
160 struct asort_context {
161 // Interface between generic code in array.c and functions generated by array.h
162 void *array; // Array to sort
163 void *buffer; // Auxiliary buffer (required when radix-sorting)
164 uns num_elts; // Number of elements in the array
165 uns elt_size; // Bytes per element
166 uns hash_bits; // Remaining bits of the hash function
167 uns radix_bits; // How many bits to process in a single radix-sort pass
168 void (*quicksort)(void *array_ptr, uns num_elts);
169 void (*quicksplit)(void *array_ptr, uns num_elts, int *leftp, int *rightp);
170 void (*radix_count)(void *src_ptr, uns num_elts, uns *cnt, uns shift);
171 void (*radix_split)(void *src_ptr, void *dest_ptr, uns num_elts, uns *ptrs, uns shift);
173 // Used internally by array.c
174 struct rs_work **rs_works;
175 struct work_queue *rs_work_queue;
176 struct eltpool *eltpool;
178 // Configured limits translated from bytes to elements
179 uns thread_threshold;
184 void asort_run(struct asort_context *ctx);
185 void asort_start_threads(uns run);
186 void asort_stop_threads(void);