xref: /freebsd/sys/contrib/zstd/programs/util.c (revision b9f654b163bce26de79705e77b872427c9f2afa1)
1 /*
2  * Copyright (c) 2016-present, Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
3  * All rights reserved.
4  *
5  * This source code is licensed under both the BSD-style license (found in the
6  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7  * in the COPYING file in the root directory of this source tree).
8  * You may select, at your option, one of the above-listed licenses.
9  */
10 
11 #if defined (__cplusplus)
12 extern "C" {
13 #endif
14 
15 
16 /*-****************************************
17 *  Dependencies
18 ******************************************/
19 #include "util.h"       /* note : ensure that platform.h is included first ! */
20 #include <errno.h>
21 #include <assert.h>
22 
23 
24 int UTIL_fileExist(const char* filename)
25 {
26     stat_t statbuf;
27 #if defined(_MSC_VER)
28     int const stat_error = _stat64(filename, &statbuf);
29 #else
30     int const stat_error = stat(filename, &statbuf);
31 #endif
32     return !stat_error;
33 }
34 
35 int UTIL_isRegularFile(const char* infilename)
36 {
37     stat_t statbuf;
38     return UTIL_getFileStat(infilename, &statbuf); /* Only need to know whether it is a regular file */
39 }
40 
41 int UTIL_getFileStat(const char* infilename, stat_t *statbuf)
42 {
43     int r;
44 #if defined(_MSC_VER)
45     r = _stat64(infilename, statbuf);
46     if (r || !(statbuf->st_mode & S_IFREG)) return 0;   /* No good... */
47 #else
48     r = stat(infilename, statbuf);
49     if (r || !S_ISREG(statbuf->st_mode)) return 0;   /* No good... */
50 #endif
51     return 1;
52 }
53 
54 int UTIL_setFileStat(const char *filename, stat_t *statbuf)
55 {
56     int res = 0;
57     struct utimbuf timebuf;
58 
59     if (!UTIL_isRegularFile(filename))
60         return -1;
61 
62     timebuf.actime = time(NULL);
63     timebuf.modtime = statbuf->st_mtime;
64     res += utime(filename, &timebuf);  /* set access and modification times */
65 
66 #if !defined(_WIN32)
67     res += chown(filename, statbuf->st_uid, statbuf->st_gid);  /* Copy ownership */
68 #endif
69 
70     res += chmod(filename, statbuf->st_mode & 07777);  /* Copy file permissions */
71 
72     errno = 0;
73     return -res; /* number of errors is returned */
74 }
75 
76 U32 UTIL_isDirectory(const char* infilename)
77 {
78     int r;
79     stat_t statbuf;
80 #if defined(_MSC_VER)
81     r = _stat64(infilename, &statbuf);
82     if (!r && (statbuf.st_mode & _S_IFDIR)) return 1;
83 #else
84     r = stat(infilename, &statbuf);
85     if (!r && S_ISDIR(statbuf.st_mode)) return 1;
86 #endif
87     return 0;
88 }
89 
90 int UTIL_isSameFile(const char* file1, const char* file2)
91 {
92 #if defined(_MSC_VER)
93     /* note : Visual does not support file identification by inode.
94      *        The following work-around is limited to detecting exact name repetition only,
95      *        aka `filename` is considered different from `subdir/../filename` */
96     return !strcmp(file1, file2);
97 #else
98     stat_t file1Stat;
99     stat_t file2Stat;
100     return UTIL_getFileStat(file1, &file1Stat)
101         && UTIL_getFileStat(file2, &file2Stat)
102         && (file1Stat.st_dev == file2Stat.st_dev)
103         && (file1Stat.st_ino == file2Stat.st_ino);
104 #endif
105 }
106 
107 U32 UTIL_isLink(const char* infilename)
108 {
109 /* macro guards, as defined in : https://linux.die.net/man/2/lstat */
110 #ifndef __STRICT_ANSI__
111 #if defined(_BSD_SOURCE) \
112     || (defined(_XOPEN_SOURCE) && (_XOPEN_SOURCE >= 500)) \
113     || (defined(_XOPEN_SOURCE) && defined(_XOPEN_SOURCE_EXTENDED)) \
114     || (defined(_POSIX_C_SOURCE) && (_POSIX_C_SOURCE >= 200112L)) \
115     || (defined(__APPLE__) && defined(__MACH__)) \
116     || defined(__OpenBSD__) \
117     || defined(__FreeBSD__)
118     int r;
119     stat_t statbuf;
120     r = lstat(infilename, &statbuf);
121     if (!r && S_ISLNK(statbuf.st_mode)) return 1;
122 #endif
123 #endif
124     (void)infilename;
125     return 0;
126 }
127 
128 U64 UTIL_getFileSize(const char* infilename)
129 {
130     if (!UTIL_isRegularFile(infilename)) return UTIL_FILESIZE_UNKNOWN;
131     {   int r;
132 #if defined(_MSC_VER)
133         struct __stat64 statbuf;
134         r = _stat64(infilename, &statbuf);
135         if (r || !(statbuf.st_mode & S_IFREG)) return UTIL_FILESIZE_UNKNOWN;
136 #elif defined(__MINGW32__) && defined (__MSVCRT__)
137         struct _stati64 statbuf;
138         r = _stati64(infilename, &statbuf);
139         if (r || !(statbuf.st_mode & S_IFREG)) return UTIL_FILESIZE_UNKNOWN;
140 #else
141         struct stat statbuf;
142         r = stat(infilename, &statbuf);
143         if (r || !S_ISREG(statbuf.st_mode)) return UTIL_FILESIZE_UNKNOWN;
144 #endif
145         return (U64)statbuf.st_size;
146     }
147 }
148 
149 
150 U64 UTIL_getTotalFileSize(const char* const * const fileNamesTable, unsigned nbFiles)
151 {
152     U64 total = 0;
153     int error = 0;
154     unsigned n;
155     for (n=0; n<nbFiles; n++) {
156         U64 const size = UTIL_getFileSize(fileNamesTable[n]);
157         error |= (size == UTIL_FILESIZE_UNKNOWN);
158         total += size;
159     }
160     return error ? UTIL_FILESIZE_UNKNOWN : total;
161 }
162 
163 #ifdef _WIN32
164 int UTIL_prepareFileList(const char *dirName, char** bufStart, size_t* pos, char** bufEnd, int followLinks)
165 {
166     char* path;
167     int dirLength, fnameLength, pathLength, nbFiles = 0;
168     WIN32_FIND_DATAA cFile;
169     HANDLE hFile;
170 
171     dirLength = (int)strlen(dirName);
172     path = (char*) malloc(dirLength + 3);
173     if (!path) return 0;
174 
175     memcpy(path, dirName, dirLength);
176     path[dirLength] = '\\';
177     path[dirLength+1] = '*';
178     path[dirLength+2] = 0;
179 
180     hFile=FindFirstFileA(path, &cFile);
181     if (hFile == INVALID_HANDLE_VALUE) {
182         UTIL_DISPLAYLEVEL(1, "Cannot open directory '%s'\n", dirName);
183         return 0;
184     }
185     free(path);
186 
187     do {
188         fnameLength = (int)strlen(cFile.cFileName);
189         path = (char*) malloc(dirLength + fnameLength + 2);
190         if (!path) { FindClose(hFile); return 0; }
191         memcpy(path, dirName, dirLength);
192         path[dirLength] = '\\';
193         memcpy(path+dirLength+1, cFile.cFileName, fnameLength);
194         pathLength = dirLength+1+fnameLength;
195         path[pathLength] = 0;
196         if (cFile.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
197             if ( strcmp (cFile.cFileName, "..") == 0
198               || strcmp (cFile.cFileName, ".") == 0 )
199                 continue;
200             /* Recursively call "UTIL_prepareFileList" with the new path. */
201             nbFiles += UTIL_prepareFileList(path, bufStart, pos, bufEnd, followLinks);
202             if (*bufStart == NULL) { free(path); FindClose(hFile); return 0; }
203         } else if ( (cFile.dwFileAttributes & FILE_ATTRIBUTE_NORMAL)
204                  || (cFile.dwFileAttributes & FILE_ATTRIBUTE_ARCHIVE)
205                  || (cFile.dwFileAttributes & FILE_ATTRIBUTE_COMPRESSED) ) {
206             if (*bufStart + *pos + pathLength >= *bufEnd) {
207                 ptrdiff_t const newListSize = (*bufEnd - *bufStart) + LIST_SIZE_INCREASE;
208                 *bufStart = (char*)UTIL_realloc(*bufStart, newListSize);
209                 if (*bufStart == NULL) { free(path); FindClose(hFile); return 0; }
210                 *bufEnd = *bufStart + newListSize;
211             }
212             if (*bufStart + *pos + pathLength < *bufEnd) {
213                 memcpy(*bufStart + *pos, path, pathLength+1 /* include final \0 */);
214                 *pos += pathLength + 1;
215                 nbFiles++;
216             }
217         }
218         free(path);
219     } while (FindNextFileA(hFile, &cFile));
220 
221     FindClose(hFile);
222     return nbFiles;
223 }
224 
225 #elif defined(__linux__) || (PLATFORM_POSIX_VERSION >= 200112L)  /* opendir, readdir require POSIX.1-2001 */
226 
227 int UTIL_prepareFileList(const char *dirName, char** bufStart, size_t* pos, char** bufEnd, int followLinks)
228 {
229     DIR *dir;
230     struct dirent *entry;
231     char* path;
232     int dirLength, fnameLength, pathLength, nbFiles = 0;
233 
234     if (!(dir = opendir(dirName))) {
235         UTIL_DISPLAYLEVEL(1, "Cannot open directory '%s': %s\n", dirName, strerror(errno));
236         return 0;
237     }
238 
239     dirLength = (int)strlen(dirName);
240     errno = 0;
241     while ((entry = readdir(dir)) != NULL) {
242         if (strcmp (entry->d_name, "..") == 0 ||
243             strcmp (entry->d_name, ".") == 0) continue;
244         fnameLength = (int)strlen(entry->d_name);
245         path = (char*) malloc(dirLength + fnameLength + 2);
246         if (!path) { closedir(dir); return 0; }
247         memcpy(path, dirName, dirLength);
248 
249         path[dirLength] = '/';
250         memcpy(path+dirLength+1, entry->d_name, fnameLength);
251         pathLength = dirLength+1+fnameLength;
252         path[pathLength] = 0;
253 
254         if (!followLinks && UTIL_isLink(path)) {
255             UTIL_DISPLAYLEVEL(2, "Warning : %s is a symbolic link, ignoring\n", path);
256             continue;
257         }
258 
259         if (UTIL_isDirectory(path)) {
260             nbFiles += UTIL_prepareFileList(path, bufStart, pos, bufEnd, followLinks);  /* Recursively call "UTIL_prepareFileList" with the new path. */
261             if (*bufStart == NULL) { free(path); closedir(dir); return 0; }
262         } else {
263             if (*bufStart + *pos + pathLength >= *bufEnd) {
264                 ptrdiff_t newListSize = (*bufEnd - *bufStart) + LIST_SIZE_INCREASE;
265                 *bufStart = (char*)UTIL_realloc(*bufStart, newListSize);
266                 *bufEnd = *bufStart + newListSize;
267                 if (*bufStart == NULL) { free(path); closedir(dir); return 0; }
268             }
269             if (*bufStart + *pos + pathLength < *bufEnd) {
270                 memcpy(*bufStart + *pos, path, pathLength + 1);  /* with final \0 */
271                 *pos += pathLength + 1;
272                 nbFiles++;
273             }
274         }
275         free(path);
276         errno = 0; /* clear errno after UTIL_isDirectory, UTIL_prepareFileList */
277     }
278 
279     if (errno != 0) {
280         UTIL_DISPLAYLEVEL(1, "readdir(%s) error: %s\n", dirName, strerror(errno));
281         free(*bufStart);
282         *bufStart = NULL;
283     }
284     closedir(dir);
285     return nbFiles;
286 }
287 
288 #else
289 
290 int UTIL_prepareFileList(const char *dirName, char** bufStart, size_t* pos, char** bufEnd, int followLinks)
291 {
292     (void)bufStart; (void)bufEnd; (void)pos; (void)followLinks;
293     UTIL_DISPLAYLEVEL(1, "Directory %s ignored (compiled without _WIN32 or _POSIX_C_SOURCE)\n", dirName);
294     return 0;
295 }
296 
297 #endif /* #ifdef _WIN32 */
298 
299 /*
300  * UTIL_createFileList - takes a list of files and directories (params: inputNames, inputNamesNb), scans directories,
301  *                       and returns a new list of files (params: return value, allocatedBuffer, allocatedNamesNb).
302  * After finishing usage of the list the structures should be freed with UTIL_freeFileList(params: return value, allocatedBuffer)
303  * In case of error UTIL_createFileList returns NULL and UTIL_freeFileList should not be called.
304  */
305 const char**
306 UTIL_createFileList(const char **inputNames, unsigned inputNamesNb,
307                     char** allocatedBuffer, unsigned* allocatedNamesNb,
308                     int followLinks)
309 {
310     size_t pos;
311     unsigned i, nbFiles;
312     char* buf = (char*)malloc(LIST_SIZE_INCREASE);
313     char* bufend = buf + LIST_SIZE_INCREASE;
314     const char** fileTable;
315 
316     if (!buf) return NULL;
317 
318     for (i=0, pos=0, nbFiles=0; i<inputNamesNb; i++) {
319         if (!UTIL_isDirectory(inputNames[i])) {
320             size_t const len = strlen(inputNames[i]);
321             if (buf + pos + len >= bufend) {
322                 ptrdiff_t newListSize = (bufend - buf) + LIST_SIZE_INCREASE;
323                 buf = (char*)UTIL_realloc(buf, newListSize);
324                 bufend = buf + newListSize;
325                 if (!buf) return NULL;
326             }
327             if (buf + pos + len < bufend) {
328                 memcpy(buf+pos, inputNames[i], len+1);  /* with final \0 */
329                 pos += len + 1;
330                 nbFiles++;
331             }
332         } else {
333             nbFiles += UTIL_prepareFileList(inputNames[i], &buf, &pos, &bufend, followLinks);
334             if (buf == NULL) return NULL;
335     }   }
336 
337     if (nbFiles == 0) { free(buf); return NULL; }
338 
339     fileTable = (const char**)malloc((nbFiles+1) * sizeof(const char*));
340     if (!fileTable) { free(buf); return NULL; }
341 
342     for (i=0, pos=0; i<nbFiles; i++) {
343         fileTable[i] = buf + pos;
344         pos += strlen(fileTable[i]) + 1;
345     }
346 
347     if (buf + pos > bufend) { free(buf); free((void*)fileTable); return NULL; }
348 
349     *allocatedBuffer = buf;
350     *allocatedNamesNb = nbFiles;
351 
352     return fileTable;
353 }
354 
355 
356 /*-****************************************
357 *  Console log
358 ******************************************/
359 int g_utilDisplayLevel;
360 
361 
362 
363 /*-****************************************
364 *  count the number of physical cores
365 ******************************************/
366 
367 #if defined(_WIN32) || defined(WIN32)
368 
369 #include <windows.h>
370 
371 typedef BOOL(WINAPI* LPFN_GLPI)(PSYSTEM_LOGICAL_PROCESSOR_INFORMATION, PDWORD);
372 
373 int UTIL_countPhysicalCores(void)
374 {
375     static int numPhysicalCores = 0;
376     if (numPhysicalCores != 0) return numPhysicalCores;
377 
378     {   LPFN_GLPI glpi;
379         BOOL done = FALSE;
380         PSYSTEM_LOGICAL_PROCESSOR_INFORMATION buffer = NULL;
381         PSYSTEM_LOGICAL_PROCESSOR_INFORMATION ptr = NULL;
382         DWORD returnLength = 0;
383         size_t byteOffset = 0;
384 
385         glpi = (LPFN_GLPI)GetProcAddress(GetModuleHandle(TEXT("kernel32")),
386                                          "GetLogicalProcessorInformation");
387 
388         if (glpi == NULL) {
389             goto failed;
390         }
391 
392         while(!done) {
393             DWORD rc = glpi(buffer, &returnLength);
394             if (FALSE == rc) {
395                 if (GetLastError() == ERROR_INSUFFICIENT_BUFFER) {
396                     if (buffer)
397                         free(buffer);
398                     buffer = (PSYSTEM_LOGICAL_PROCESSOR_INFORMATION)malloc(returnLength);
399 
400                     if (buffer == NULL) {
401                         perror("zstd");
402                         exit(1);
403                     }
404                 } else {
405                     /* some other error */
406                     goto failed;
407                 }
408             } else {
409                 done = TRUE;
410             }
411         }
412 
413         ptr = buffer;
414 
415         while (byteOffset + sizeof(SYSTEM_LOGICAL_PROCESSOR_INFORMATION) <= returnLength) {
416 
417             if (ptr->Relationship == RelationProcessorCore) {
418                 numPhysicalCores++;
419             }
420 
421             ptr++;
422             byteOffset += sizeof(SYSTEM_LOGICAL_PROCESSOR_INFORMATION);
423         }
424 
425         free(buffer);
426 
427         return numPhysicalCores;
428     }
429 
430 failed:
431     /* try to fall back on GetSystemInfo */
432     {   SYSTEM_INFO sysinfo;
433         GetSystemInfo(&sysinfo);
434         numPhysicalCores = sysinfo.dwNumberOfProcessors;
435         if (numPhysicalCores == 0) numPhysicalCores = 1; /* just in case */
436     }
437     return numPhysicalCores;
438 }
439 
440 #elif defined(__APPLE__)
441 
442 #include <sys/sysctl.h>
443 
444 /* Use apple-provided syscall
445  * see: man 3 sysctl */
446 int UTIL_countPhysicalCores(void)
447 {
448     static S32 numPhysicalCores = 0; /* apple specifies int32_t */
449     if (numPhysicalCores != 0) return numPhysicalCores;
450 
451     {   size_t size = sizeof(S32);
452         int const ret = sysctlbyname("hw.physicalcpu", &numPhysicalCores, &size, NULL, 0);
453         if (ret != 0) {
454             if (errno == ENOENT) {
455                 /* entry not present, fall back on 1 */
456                 numPhysicalCores = 1;
457             } else {
458                 perror("zstd: can't get number of physical cpus");
459                 exit(1);
460             }
461         }
462 
463         return numPhysicalCores;
464     }
465 }
466 
467 #elif defined(__linux__)
468 
469 /* parse /proc/cpuinfo
470  * siblings / cpu cores should give hyperthreading ratio
471  * otherwise fall back on sysconf */
472 int UTIL_countPhysicalCores(void)
473 {
474     static int numPhysicalCores = 0;
475 
476     if (numPhysicalCores != 0) return numPhysicalCores;
477 
478     numPhysicalCores = (int)sysconf(_SC_NPROCESSORS_ONLN);
479     if (numPhysicalCores == -1) {
480         /* value not queryable, fall back on 1 */
481         return numPhysicalCores = 1;
482     }
483 
484     /* try to determine if there's hyperthreading */
485     {   FILE* const cpuinfo = fopen("/proc/cpuinfo", "r");
486 #define BUF_SIZE 80
487         char buff[BUF_SIZE];
488 
489         int siblings = 0;
490         int cpu_cores = 0;
491         int ratio = 1;
492 
493         if (cpuinfo == NULL) {
494             /* fall back on the sysconf value */
495             return numPhysicalCores;
496         }
497 
498         /* assume the cpu cores/siblings values will be constant across all
499          * present processors */
500         while (!feof(cpuinfo)) {
501             if (fgets(buff, BUF_SIZE, cpuinfo) != NULL) {
502                 if (strncmp(buff, "siblings", 8) == 0) {
503                     const char* const sep = strchr(buff, ':');
504                     if (*sep == '\0') {
505                         /* formatting was broken? */
506                         goto failed;
507                     }
508 
509                     siblings = atoi(sep + 1);
510                 }
511                 if (strncmp(buff, "cpu cores", 9) == 0) {
512                     const char* const sep = strchr(buff, ':');
513                     if (*sep == '\0') {
514                         /* formatting was broken? */
515                         goto failed;
516                     }
517 
518                     cpu_cores = atoi(sep + 1);
519                 }
520             } else if (ferror(cpuinfo)) {
521                 /* fall back on the sysconf value */
522                 goto failed;
523             }
524         }
525         if (siblings && cpu_cores) {
526             ratio = siblings / cpu_cores;
527         }
528 failed:
529         fclose(cpuinfo);
530         return numPhysicalCores = numPhysicalCores / ratio;
531     }
532 }
533 
534 #elif defined(__FreeBSD__)
535 
536 #include <sys/param.h>
537 #include <sys/sysctl.h>
538 
539 /* Use physical core sysctl when available
540  * see: man 4 smp, man 3 sysctl */
541 int UTIL_countPhysicalCores(void)
542 {
543     static int numPhysicalCores = 0; /* freebsd sysctl is native int sized */
544     if (numPhysicalCores != 0) return numPhysicalCores;
545 
546 #if __FreeBSD_version >= 1300008
547     {   size_t size = sizeof(numPhysicalCores);
548         int ret = sysctlbyname("kern.smp.cores", &numPhysicalCores, &size, NULL, 0);
549         if (ret == 0) return numPhysicalCores;
550         if (errno != ENOENT) {
551             perror("zstd: can't get number of physical cpus");
552             exit(1);
553         }
554         /* sysctl not present, fall through to older sysconf method */
555     }
556 #endif
557 
558     numPhysicalCores = (int)sysconf(_SC_NPROCESSORS_ONLN);
559     if (numPhysicalCores == -1) {
560         /* value not queryable, fall back on 1 */
561         numPhysicalCores = 1;
562     }
563     return numPhysicalCores;
564 }
565 
566 #elif defined(__NetBSD__) || defined(__OpenBSD__) || defined(__DragonFly__)
567 
568 /* Use POSIX sysconf
569  * see: man 3 sysconf */
570 int UTIL_countPhysicalCores(void)
571 {
572     static int numPhysicalCores = 0;
573 
574     if (numPhysicalCores != 0) return numPhysicalCores;
575 
576     numPhysicalCores = (int)sysconf(_SC_NPROCESSORS_ONLN);
577     if (numPhysicalCores == -1) {
578         /* value not queryable, fall back on 1 */
579         return numPhysicalCores = 1;
580     }
581     return numPhysicalCores;
582 }
583 
584 #else
585 
586 int UTIL_countPhysicalCores(void)
587 {
588     /* assume 1 */
589     return 1;
590 }
591 
592 #endif
593 
594 #if defined (__cplusplus)
595 }
596 #endif
597