xref: /freebsd-src/sys/contrib/zstd/examples/streaming_compression_thread_pool.c (revision 5ff13fbc199bdf5f0572845351c68ee5ca828e71)
1f7cd7fe5SConrad Meyer /*
2*5ff13fbcSAllan Jude  * Copyright (c) Martin Liska, SUSE, Facebook, Inc.
3f7cd7fe5SConrad Meyer  * All rights reserved.
4f7cd7fe5SConrad Meyer  *
5f7cd7fe5SConrad Meyer  * This source code is licensed under both the BSD-style license (found in the
6f7cd7fe5SConrad Meyer  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7f7cd7fe5SConrad Meyer  * in the COPYING file in the root directory of this source tree).
8f7cd7fe5SConrad Meyer  * You may select, at your option, one of the above-listed licenses.
9f7cd7fe5SConrad Meyer  */
10f7cd7fe5SConrad Meyer 
11f7cd7fe5SConrad Meyer 
12f7cd7fe5SConrad Meyer #include <stdio.h>     // printf
13f7cd7fe5SConrad Meyer #include <stdlib.h>    // free
14f7cd7fe5SConrad Meyer #include <string.h>    // memset, strcat, strlen
15f7cd7fe5SConrad Meyer #include <zstd.h>      // presumes zstd library is installed
16f7cd7fe5SConrad Meyer #include "common.h"    // Helper functions, CHECK(), and CHECK_ZSTD()
17f7cd7fe5SConrad Meyer #include <pthread.h>
18f7cd7fe5SConrad Meyer 
19f7cd7fe5SConrad Meyer typedef struct compress_args
20f7cd7fe5SConrad Meyer {
21f7cd7fe5SConrad Meyer   const char *fname;
22f7cd7fe5SConrad Meyer   char *outName;
23f7cd7fe5SConrad Meyer   int cLevel;
24f7cd7fe5SConrad Meyer #if defined(ZSTD_STATIC_LINKING_ONLY)
25f7cd7fe5SConrad Meyer   ZSTD_threadPool *pool;
26f7cd7fe5SConrad Meyer #endif
27f7cd7fe5SConrad Meyer } compress_args_t;
28f7cd7fe5SConrad Meyer 
compressFile_orDie(void * data)29f7cd7fe5SConrad Meyer static void *compressFile_orDie(void *data)
30f7cd7fe5SConrad Meyer {
31*5ff13fbcSAllan Jude     const int nbThreads = 16;
32*5ff13fbcSAllan Jude 
33f7cd7fe5SConrad Meyer     compress_args_t *args = (compress_args_t *)data;
34*5ff13fbcSAllan Jude     fprintf (stderr, "Starting compression of %s with level %d, using %d threads\n", args->fname, args->cLevel, nbThreads);
35f7cd7fe5SConrad Meyer     /* Open the input and output files. */
36f7cd7fe5SConrad Meyer     FILE* const fin  = fopen_orDie(args->fname, "rb");
37f7cd7fe5SConrad Meyer     FILE* const fout = fopen_orDie(args->outName, "wb");
38f7cd7fe5SConrad Meyer     /* Create the input and output buffers.
39f7cd7fe5SConrad Meyer      * They may be any size, but we recommend using these functions to size them.
40f7cd7fe5SConrad Meyer      * Performance will only suffer significantly for very tiny buffers.
41f7cd7fe5SConrad Meyer      */
42f7cd7fe5SConrad Meyer     size_t const buffInSize = ZSTD_CStreamInSize();
43f7cd7fe5SConrad Meyer     void*  const buffIn  = malloc_orDie(buffInSize);
44f7cd7fe5SConrad Meyer     size_t const buffOutSize = ZSTD_CStreamOutSize();
45f7cd7fe5SConrad Meyer     void*  const buffOut = malloc_orDie(buffOutSize);
46f7cd7fe5SConrad Meyer 
47f7cd7fe5SConrad Meyer     /* Create the context. */
48f7cd7fe5SConrad Meyer     ZSTD_CCtx* const cctx = ZSTD_createCCtx();
49f7cd7fe5SConrad Meyer     CHECK(cctx != NULL, "ZSTD_createCCtx() failed!");
50f7cd7fe5SConrad Meyer 
51f7cd7fe5SConrad Meyer #if defined(ZSTD_STATIC_LINKING_ONLY)
52f7cd7fe5SConrad Meyer     size_t r = ZSTD_CCtx_refThreadPool(cctx, args->pool);
53f7cd7fe5SConrad Meyer     CHECK(r == 0, "ZSTD_CCtx_refThreadPool failed!");
54f7cd7fe5SConrad Meyer #endif
55f7cd7fe5SConrad Meyer 
56f7cd7fe5SConrad Meyer     /* Set any parameters you want.
57f7cd7fe5SConrad Meyer      * Here we set the compression level, and enable the checksum.
58f7cd7fe5SConrad Meyer      */
59f7cd7fe5SConrad Meyer     CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_compressionLevel, args->cLevel) );
60f7cd7fe5SConrad Meyer     CHECK_ZSTD( ZSTD_CCtx_setParameter(cctx, ZSTD_c_checksumFlag, 1) );
61*5ff13fbcSAllan Jude     ZSTD_CCtx_setParameter(cctx, ZSTD_c_nbWorkers, nbThreads);
62f7cd7fe5SConrad Meyer 
63*5ff13fbcSAllan Jude     /* This loop reads from the input file, compresses that entire chunk,
64f7cd7fe5SConrad Meyer      * and writes all output produced to the output file.
65f7cd7fe5SConrad Meyer      */
66f7cd7fe5SConrad Meyer     size_t const toRead = buffInSize;
67f7cd7fe5SConrad Meyer     for (;;) {
68f7cd7fe5SConrad Meyer         size_t read = fread_orDie(buffIn, toRead, fin);
69f7cd7fe5SConrad Meyer         /* Select the flush mode.
70f7cd7fe5SConrad Meyer          * If the read may not be finished (read == toRead) we use
71f7cd7fe5SConrad Meyer          * ZSTD_e_continue. If this is the last chunk, we use ZSTD_e_end.
72f7cd7fe5SConrad Meyer          * Zstd optimizes the case where the first flush mode is ZSTD_e_end,
73f7cd7fe5SConrad Meyer          * since it knows it is compressing the entire source in one pass.
74f7cd7fe5SConrad Meyer          */
75f7cd7fe5SConrad Meyer         int const lastChunk = (read < toRead);
76f7cd7fe5SConrad Meyer         ZSTD_EndDirective const mode = lastChunk ? ZSTD_e_end : ZSTD_e_continue;
77f7cd7fe5SConrad Meyer         /* Set the input buffer to what we just read.
78f7cd7fe5SConrad Meyer          * We compress until the input buffer is empty, each time flushing the
79f7cd7fe5SConrad Meyer          * output.
80f7cd7fe5SConrad Meyer          */
81f7cd7fe5SConrad Meyer         ZSTD_inBuffer input = { buffIn, read, 0 };
82f7cd7fe5SConrad Meyer         int finished;
83f7cd7fe5SConrad Meyer         do {
84f7cd7fe5SConrad Meyer             /* Compress into the output buffer and write all of the output to
85f7cd7fe5SConrad Meyer              * the file so we can reuse the buffer next iteration.
86f7cd7fe5SConrad Meyer              */
87f7cd7fe5SConrad Meyer             ZSTD_outBuffer output = { buffOut, buffOutSize, 0 };
88f7cd7fe5SConrad Meyer             size_t const remaining = ZSTD_compressStream2(cctx, &output , &input, mode);
89f7cd7fe5SConrad Meyer             CHECK_ZSTD(remaining);
90f7cd7fe5SConrad Meyer             fwrite_orDie(buffOut, output.pos, fout);
91f7cd7fe5SConrad Meyer             /* If we're on the last chunk we're finished when zstd returns 0,
92f7cd7fe5SConrad Meyer              * which means its consumed all the input AND finished the frame.
93f7cd7fe5SConrad Meyer              * Otherwise, we're finished when we've consumed all the input.
94f7cd7fe5SConrad Meyer              */
95f7cd7fe5SConrad Meyer             finished = lastChunk ? (remaining == 0) : (input.pos == input.size);
96f7cd7fe5SConrad Meyer         } while (!finished);
97f7cd7fe5SConrad Meyer         CHECK(input.pos == input.size,
98f7cd7fe5SConrad Meyer               "Impossible: zstd only returns 0 when the input is completely consumed!");
99f7cd7fe5SConrad Meyer 
100f7cd7fe5SConrad Meyer         if (lastChunk) {
101f7cd7fe5SConrad Meyer             break;
102f7cd7fe5SConrad Meyer         }
103f7cd7fe5SConrad Meyer     }
104f7cd7fe5SConrad Meyer 
105f7cd7fe5SConrad Meyer     fprintf (stderr, "Finishing compression of %s\n", args->outName);
106f7cd7fe5SConrad Meyer 
107f7cd7fe5SConrad Meyer     ZSTD_freeCCtx(cctx);
108f7cd7fe5SConrad Meyer     fclose_orDie(fout);
109f7cd7fe5SConrad Meyer     fclose_orDie(fin);
110f7cd7fe5SConrad Meyer     free(buffIn);
111f7cd7fe5SConrad Meyer     free(buffOut);
112f7cd7fe5SConrad Meyer     free(args->outName);
113f7cd7fe5SConrad Meyer 
114f7cd7fe5SConrad Meyer     return NULL;
115f7cd7fe5SConrad Meyer }
116f7cd7fe5SConrad Meyer 
117f7cd7fe5SConrad Meyer 
createOutFilename_orDie(const char * filename)118f7cd7fe5SConrad Meyer static char* createOutFilename_orDie(const char* filename)
119f7cd7fe5SConrad Meyer {
120f7cd7fe5SConrad Meyer     size_t const inL = strlen(filename);
121f7cd7fe5SConrad Meyer     size_t const outL = inL + 5;
122f7cd7fe5SConrad Meyer     void* const outSpace = malloc_orDie(outL);
123f7cd7fe5SConrad Meyer     memset(outSpace, 0, outL);
124f7cd7fe5SConrad Meyer     strcat(outSpace, filename);
125f7cd7fe5SConrad Meyer     strcat(outSpace, ".zst");
126f7cd7fe5SConrad Meyer     return (char*)outSpace;
127f7cd7fe5SConrad Meyer }
128f7cd7fe5SConrad Meyer 
main(int argc,const char ** argv)129f7cd7fe5SConrad Meyer int main(int argc, const char** argv)
130f7cd7fe5SConrad Meyer {
131f7cd7fe5SConrad Meyer     const char* const exeName = argv[0];
132f7cd7fe5SConrad Meyer 
133f7cd7fe5SConrad Meyer     if (argc<=3) {
134f7cd7fe5SConrad Meyer         printf("wrong arguments\n");
135f7cd7fe5SConrad Meyer         printf("usage:\n");
136f7cd7fe5SConrad Meyer         printf("%s POOL_SIZE LEVEL FILES\n", exeName);
137f7cd7fe5SConrad Meyer         return 1;
138f7cd7fe5SConrad Meyer     }
139f7cd7fe5SConrad Meyer 
140f7cd7fe5SConrad Meyer     int pool_size = atoi (argv[1]);
141f7cd7fe5SConrad Meyer     CHECK(pool_size != 0, "can't parse POOL_SIZE!");
142f7cd7fe5SConrad Meyer 
143f7cd7fe5SConrad Meyer     int level = atoi (argv[2]);
144f7cd7fe5SConrad Meyer     CHECK(level != 0, "can't parse LEVEL!");
145f7cd7fe5SConrad Meyer 
146f7cd7fe5SConrad Meyer     argc -= 3;
147f7cd7fe5SConrad Meyer     argv += 3;
148f7cd7fe5SConrad Meyer 
149f7cd7fe5SConrad Meyer #if defined(ZSTD_STATIC_LINKING_ONLY)
150f7cd7fe5SConrad Meyer     ZSTD_threadPool *pool = ZSTD_createThreadPool (pool_size);
151f7cd7fe5SConrad Meyer     CHECK(pool != NULL, "ZSTD_createThreadPool() failed!");
152f7cd7fe5SConrad Meyer     fprintf (stderr, "Using shared thread pool of size %d\n", pool_size);
153f7cd7fe5SConrad Meyer #else
154f7cd7fe5SConrad Meyer     fprintf (stderr, "All threads use its own thread pool\n");
155f7cd7fe5SConrad Meyer #endif
156f7cd7fe5SConrad Meyer 
157f7cd7fe5SConrad Meyer     pthread_t *threads = malloc_orDie(argc * sizeof(pthread_t));
158f7cd7fe5SConrad Meyer     compress_args_t *args = malloc_orDie(argc * sizeof(compress_args_t));
159f7cd7fe5SConrad Meyer 
160f7cd7fe5SConrad Meyer     for (unsigned i = 0; i < argc; i++)
161f7cd7fe5SConrad Meyer     {
162f7cd7fe5SConrad Meyer       args[i].fname = argv[i];
163f7cd7fe5SConrad Meyer       args[i].outName = createOutFilename_orDie(args[i].fname);
164f7cd7fe5SConrad Meyer       args[i].cLevel = level;
165f7cd7fe5SConrad Meyer #if defined(ZSTD_STATIC_LINKING_ONLY)
166f7cd7fe5SConrad Meyer       args[i].pool = pool;
167f7cd7fe5SConrad Meyer #endif
168f7cd7fe5SConrad Meyer 
169f7cd7fe5SConrad Meyer       pthread_create (&threads[i], NULL, compressFile_orDie, &args[i]);
170f7cd7fe5SConrad Meyer     }
171f7cd7fe5SConrad Meyer 
172f7cd7fe5SConrad Meyer     for (unsigned i = 0; i < argc; i++)
173f7cd7fe5SConrad Meyer       pthread_join (threads[i], NULL);
174f7cd7fe5SConrad Meyer 
175f7cd7fe5SConrad Meyer #if defined(ZSTD_STATIC_LINKING_ONLY)
176f7cd7fe5SConrad Meyer     ZSTD_freeThreadPool (pool);
177f7cd7fe5SConrad Meyer #endif
178f7cd7fe5SConrad Meyer 
179f7cd7fe5SConrad Meyer     return 0;
180f7cd7fe5SConrad Meyer }
181