xref: /netbsd-src/external/bsd/zstd/dist/examples/streaming_compression_thread_pool.c (revision 3117ece4fc4a4ca4489ba793710b60b0d26bab6c)
1*3117ece4Schristos /*
2*3117ece4Schristos  * Copyright (c) Martin Liska, SUSE, Meta Platforms, Inc. and affiliates.
3*3117ece4Schristos  * All rights reserved.
4*3117ece4Schristos  *
5*3117ece4Schristos  * This source code is licensed under both the BSD-style license (found in the
6*3117ece4Schristos  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7*3117ece4Schristos  * in the COPYING file in the root directory of this source tree).
8*3117ece4Schristos  * You may select, at your option, one of the above-listed licenses.
9*3117ece4Schristos  */
10*3117ece4Schristos 
11*3117ece4Schristos 
12*3117ece4Schristos #include <stdio.h>     // printf
13*3117ece4Schristos #include <stdlib.h>    // free
14*3117ece4Schristos #include <string.h>    // memset, strcat, strlen
15*3117ece4Schristos #include <zstd.h>      // presumes zstd library is installed
16*3117ece4Schristos #include "common.h"    // Helper functions, CHECK(), and CHECK_ZSTD()
17*3117ece4Schristos #include <pthread.h>
18*3117ece4Schristos 
19*3117ece4Schristos typedef struct compress_args
20*3117ece4Schristos {
21*3117ece4Schristos   const char *fname;
22*3117ece4Schristos   char *outName;
23*3117ece4Schristos   int cLevel;
24*3117ece4Schristos #if defined(ZSTD_STATIC_LINKING_ONLY)
25*3117ece4Schristos   ZSTD_threadPool *pool;
26*3117ece4Schristos #endif
27*3117ece4Schristos } compress_args_t;
28*3117ece4Schristos 
29*3117ece4Schristos static void *compressFile_orDie(void *data)
30*3117ece4Schristos {
31*3117ece4Schristos     const int nbThreads = 16;
32*3117ece4Schristos 
33*3117ece4Schristos     compress_args_t *args = (compress_args_t *)data;
34*3117ece4Schristos     fprintf (stderr, "Starting compression of %s with level %d, using %d threads\n", args->fname, args->cLevel, nbThreads);
35*3117ece4Schristos     /* Open the input and output files. */
36*3117ece4Schristos     FILE* const fin  = fopen_orDie(args->fname, "rb");
37*3117ece4Schristos     FILE* const fout = fopen_orDie(args->outName, "wb");
38*3117ece4Schristos     /* Create the input and output buffers.
39*3117ece4Schristos      * They may be any size, but we recommend using these functions to size them.
40*3117ece4Schristos      * Performance will only suffer significantly for very tiny buffers.
41*3117ece4Schristos      */
42*3117ece4Schristos     size_t const buffInSize = ZSTD_CStreamInSize();
43*3117ece4Schristos     void*  const buffIn  = malloc_orDie(buffInSize);
44*3117ece4Schristos     size_t const buffOutSize = ZSTD_CStreamOutSize();
45*3117ece4Schristos     void*  const buffOut = malloc_orDie(buffOutSize);
46*3117ece4Schristos 
47*3117ece4Schristos     /* Create the context. */
48*3117ece4Schristos     ZSTD_CCtx* const cctx = ZSTD_createCCtx();
49*3117ece4Schristos     CHECK(cctx != NULL, "ZSTD_createCCtx() failed!");
50*3117ece4Schristos 
51*3117ece4Schristos #if defined(ZSTD_STATIC_LINKING_ONLY)
52*3117ece4Schristos     size_t r = ZSTD_CCtx_refThreadPool(cctx, args->pool);
53*3117ece4Schristos     CHECK(r == 0, "ZSTD_CCtx_refThreadPool failed!");
54*3117ece4Schristos #endif
55*3117ece4Schristos 
56*3117ece4Schristos     /* Set any parameters you want.
57*3117ece4Schristos      * Here we set the compression level, and enable the checksum.
58*3117ece4Schristos      */
59*3117ece4Schristos     CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_compressionLevel, args->cLevel) );
60*3117ece4Schristos     CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_checksumFlag, 1) );
61*3117ece4Schristos     ZSTD_CCtx_setParameter(cctx, ZSTD_c_nbWorkers, nbThreads);
62*3117ece4Schristos 
63*3117ece4Schristos     /* This loop reads from the input file, compresses that entire chunk,
64*3117ece4Schristos      * and writes all output produced to the output file.
65*3117ece4Schristos      */
66*3117ece4Schristos     size_t const toRead = buffInSize;
67*3117ece4Schristos     for (;;) {
68*3117ece4Schristos         size_t read = fread_orDie(buffIn, toRead, fin);
69*3117ece4Schristos         /* Select the flush mode.
70*3117ece4Schristos          * If the read may not be finished (read == toRead) we use
71*3117ece4Schristos          * ZSTD_e_continue. If this is the last chunk, we use ZSTD_e_end.
72*3117ece4Schristos          * Zstd optimizes the case where the first flush mode is ZSTD_e_end,
73*3117ece4Schristos          * since it knows it is compressing the entire source in one pass.
74*3117ece4Schristos          */
75*3117ece4Schristos         int const lastChunk = (read < toRead);
76*3117ece4Schristos         ZSTD_EndDirective const mode = lastChunk ? ZSTD_e_end : ZSTD_e_continue;
77*3117ece4Schristos         /* Set the input buffer to what we just read.
78*3117ece4Schristos          * We compress until the input buffer is empty, each time flushing the
79*3117ece4Schristos          * output.
80*3117ece4Schristos          */
81*3117ece4Schristos         ZSTD_inBuffer input = { buffIn, read, 0 };
82*3117ece4Schristos         int finished;
83*3117ece4Schristos         do {
84*3117ece4Schristos             /* Compress into the output buffer and write all of the output to
85*3117ece4Schristos              * the file so we can reuse the buffer next iteration.
86*3117ece4Schristos              */
87*3117ece4Schristos             ZSTD_outBuffer output = { buffOut, buffOutSize, 0 };
88*3117ece4Schristos             size_t const remaining = ZSTD_compressStream2(cctx, &output , &input, mode);
89*3117ece4Schristos             CHECK_ZSTD(remaining);
90*3117ece4Schristos             fwrite_orDie(buffOut, output.pos, fout);
91*3117ece4Schristos             /* If we're on the last chunk we're finished when zstd returns 0,
92*3117ece4Schristos              * which means its consumed all the input AND finished the frame.
93*3117ece4Schristos              * Otherwise, we're finished when we've consumed all the input.
94*3117ece4Schristos              */
95*3117ece4Schristos             finished = lastChunk ? (remaining == 0) : (input.pos == input.size);
96*3117ece4Schristos         } while (!finished);
97*3117ece4Schristos         CHECK(input.pos == input.size,
98*3117ece4Schristos               "Impossible: zstd only returns 0 when the input is completely consumed!");
99*3117ece4Schristos 
100*3117ece4Schristos         if (lastChunk) {
101*3117ece4Schristos             break;
102*3117ece4Schristos         }
103*3117ece4Schristos     }
104*3117ece4Schristos 
105*3117ece4Schristos     fprintf (stderr, "Finishing compression of %s\n", args->outName);
106*3117ece4Schristos 
107*3117ece4Schristos     ZSTD_freeCCtx(cctx);
108*3117ece4Schristos     fclose_orDie(fout);
109*3117ece4Schristos     fclose_orDie(fin);
110*3117ece4Schristos     free(buffIn);
111*3117ece4Schristos     free(buffOut);
112*3117ece4Schristos     free(args->outName);
113*3117ece4Schristos 
114*3117ece4Schristos     return NULL;
115*3117ece4Schristos }
116*3117ece4Schristos 
117*3117ece4Schristos 
118*3117ece4Schristos static char* createOutFilename_orDie(const char* filename)
119*3117ece4Schristos {
120*3117ece4Schristos     size_t const inL = strlen(filename);
121*3117ece4Schristos     size_t const outL = inL + 5;
122*3117ece4Schristos     void* const outSpace = malloc_orDie(outL);
123*3117ece4Schristos     memset(outSpace, 0, outL);
124*3117ece4Schristos     strcat(outSpace, filename);
125*3117ece4Schristos     strcat(outSpace, ".zst");
126*3117ece4Schristos     return (char*)outSpace;
127*3117ece4Schristos }
128*3117ece4Schristos 
129*3117ece4Schristos int main(int argc, const char** argv)
130*3117ece4Schristos {
131*3117ece4Schristos     const char* const exeName = argv[0];
132*3117ece4Schristos 
133*3117ece4Schristos     if (argc<=3) {
134*3117ece4Schristos         printf("wrong arguments\n");
135*3117ece4Schristos         printf("usage:\n");
136*3117ece4Schristos         printf("%s POOL_SIZE LEVEL FILES\n", exeName);
137*3117ece4Schristos         return 1;
138*3117ece4Schristos     }
139*3117ece4Schristos 
140*3117ece4Schristos     int pool_size = atoi (argv[1]);
141*3117ece4Schristos     CHECK(pool_size != 0, "can't parse POOL_SIZE!");
142*3117ece4Schristos 
143*3117ece4Schristos     int level = atoi (argv[2]);
144*3117ece4Schristos     CHECK(level != 0, "can't parse LEVEL!");
145*3117ece4Schristos 
146*3117ece4Schristos     argc -= 3;
147*3117ece4Schristos     argv += 3;
148*3117ece4Schristos 
149*3117ece4Schristos #if defined(ZSTD_STATIC_LINKING_ONLY)
150*3117ece4Schristos     ZSTD_threadPool *pool = ZSTD_createThreadPool (pool_size);
151*3117ece4Schristos     CHECK(pool != NULL, "ZSTD_createThreadPool() failed!");
152*3117ece4Schristos     fprintf (stderr, "Using shared thread pool of size %d\n", pool_size);
153*3117ece4Schristos #else
154*3117ece4Schristos     fprintf (stderr, "All threads use its own thread pool\n");
155*3117ece4Schristos #endif
156*3117ece4Schristos 
157*3117ece4Schristos     pthread_t *threads = malloc_orDie(argc * sizeof(pthread_t));
158*3117ece4Schristos     compress_args_t *args = malloc_orDie(argc * sizeof(compress_args_t));
159*3117ece4Schristos 
160*3117ece4Schristos     for (unsigned i = 0; i < argc; i++)
161*3117ece4Schristos     {
162*3117ece4Schristos       args[i].fname = argv[i];
163*3117ece4Schristos       args[i].outName = createOutFilename_orDie(args[i].fname);
164*3117ece4Schristos       args[i].cLevel = level;
165*3117ece4Schristos #if defined(ZSTD_STATIC_LINKING_ONLY)
166*3117ece4Schristos       args[i].pool = pool;
167*3117ece4Schristos #endif
168*3117ece4Schristos 
169*3117ece4Schristos       pthread_create (&threads[i], NULL, compressFile_orDie, &args[i]);
170*3117ece4Schristos     }
171*3117ece4Schristos 
172*3117ece4Schristos     for (unsigned i = 0; i < argc; i++)
173*3117ece4Schristos       pthread_join (threads[i], NULL);
174*3117ece4Schristos 
175*3117ece4Schristos #if defined(ZSTD_STATIC_LINKING_ONLY)
176*3117ece4Schristos     ZSTD_freeThreadPool (pool);
177*3117ece4Schristos #endif
178*3117ece4Schristos 
179*3117ece4Schristos     return 0;
180*3117ece4Schristos }
181