Coverage Report

Created: 2023-06-07 07:17

/src/icu/source/common/umapfile.cpp
Line
Count
Source (jump to first uncovered line)
1
// © 2016 and later: Unicode, Inc. and others.
2
// License & terms of use: http://www.unicode.org/copyright.html
3
/*
4
******************************************************************************
5
*
6
*   Copyright (C) 1999-2013, International Business Machines
7
*   Corporation and others.  All Rights Reserved.
8
*
9
******************************************************************************/
10
11
12
/*----------------------------------------------------------------------------
13
 *
14
 *       Memory mapped file wrappers for use by the ICU Data Implementation
15
 *       All of the platform-specific implementation for mapping data files
16
 *         is here.  The rest of the ICU Data implementation uses only the
17
 *         wrapper functions.
18
 *
19
 *----------------------------------------------------------------------------*/
20
/* Defines _XOPEN_SOURCE for access to POSIX functions.
21
 * Must be before any other #includes. */
22
#include "uposixdefs.h"
23
24
#include "unicode/putil.h"
25
#include "udatamem.h"
26
#include "umapfile.h"
27
28
/* memory-mapping base definitions ------------------------------------------ */
29
30
#if MAP_IMPLEMENTATION==MAP_WIN32
31
#ifndef WIN32_LEAN_AND_MEAN
32
#   define WIN32_LEAN_AND_MEAN
33
#endif
34
#   define VC_EXTRALEAN
35
#   define NOUSER
36
#   define NOSERVICE
37
#   define NOIME
38
#   define NOMCX
39
#   include <windows.h>
40
#   include "cmemory.h"
41
42
    typedef HANDLE MemoryMap;
43
44
#   define IS_MAP(map) ((map)!=NULL)
45
#elif MAP_IMPLEMENTATION==MAP_POSIX || MAP_IMPLEMENTATION==MAP_390DLL
46
    typedef size_t MemoryMap;
47
48
#   define IS_MAP(map) ((map)!=0)
49
50
#   include <unistd.h>
51
#   include <sys/mman.h>
52
#   include <sys/stat.h>
53
#   include <fcntl.h>
54
55
#   ifndef MAP_FAILED
56
#       define MAP_FAILED ((void*)-1)
57
#   endif
58
59
#   if MAP_IMPLEMENTATION==MAP_390DLL
60
        /*   No memory mapping for 390 batch mode.  Fake it using dll loading.  */
61
#       include <dll.h>
62
#       include "cstring.h"
63
#       include "cmemory.h"
64
#       include "unicode/udata.h"
65
#       define LIB_PREFIX "lib"
66
#       define LIB_SUFFIX ".dll"
67
        /* This is inconvienient until we figure out what to do with U_ICUDATA_NAME in utypes.h */
68
#       define U_ICUDATA_ENTRY_NAME "icudt" U_ICU_VERSION_SHORT U_LIB_SUFFIX_C_NAME_STRING "_dat"
69
#   endif
70
#elif MAP_IMPLEMENTATION==MAP_STDIO
71
#   include <stdio.h>
72
#   include "cmemory.h"
73
74
    typedef void *MemoryMap;
75
76
#   define IS_MAP(map) ((map)!=NULL)
77
#endif
78
79
/*----------------------------------------------------------------------------*
80
 *                                                                            *
81
 *   Memory Mapped File support.  Platform dependent implementation of        *
82
 *                           functions used by the rest of the implementation.*
83
 *                                                                            *
84
 *----------------------------------------------------------------------------*/
85
#if MAP_IMPLEMENTATION==MAP_NONE
86
    U_CFUNC UBool
87
    uprv_mapFile(UDataMemory *pData, const char *path) {
88
        UDataMemory_init(pData); /* Clear the output struct. */
89
        return FALSE;            /* no file access */
90
    }
91
92
    U_CFUNC void uprv_unmapFile(UDataMemory *pData) {
93
        /* nothing to do */
94
    }
95
#elif MAP_IMPLEMENTATION==MAP_WIN32
96
    U_CFUNC UBool
97
    uprv_mapFile(
98
         UDataMemory *pData,    /* Fill in with info on the result doing the mapping. */
99
                                /*   Output only; any original contents are cleared.  */
100
         const char *path       /* File path to be opened/mapped                      */
101
         )
102
    {
103
        HANDLE map;
104
        HANDLE file;
105
        SECURITY_ATTRIBUTES mappingAttributes;
106
        SECURITY_ATTRIBUTES *mappingAttributesPtr = NULL;
107
        SECURITY_DESCRIPTOR securityDesc;
108
109
        UDataMemory_init(pData); /* Clear the output struct.        */
110
111
        /* open the input file */
112
#if U_PLATFORM_HAS_WINUWP_API == 0
113
        file=CreateFileA(path, GENERIC_READ, FILE_SHARE_READ, NULL,
114
            OPEN_EXISTING,
115
            FILE_ATTRIBUTE_NORMAL|FILE_FLAG_RANDOM_ACCESS, NULL);
116
#else
117
        // First we need to go from char to UTF-16
118
        // u_UCharsToChars could work but it requires length.
119
        WCHAR utf16Path[MAX_PATH];
120
        int32_t i;
121
        for (i = 0; i < UPRV_LENGTHOF(utf16Path); i++)
122
        {
123
            utf16Path[i] = path[i];
124
            if (path[i] == '\0')
125
            {
126
                break;
127
            }
128
        }
129
        if (i >= UPRV_LENGTHOF(utf16Path))
130
        {
131
            // Ran out of room, unlikely but be safe
132
            utf16Path[UPRV_LENGTHOF(utf16Path) - 1] = '\0';
133
        }
134
135
        // TODO: Is it worth setting extended parameters to specify random access?
136
        file = CreateFile2(utf16Path, GENERIC_READ, FILE_SHARE_READ, OPEN_EXISTING, NULL);
137
#endif
138
        if(file==INVALID_HANDLE_VALUE) {
139
            return FALSE;
140
        }
141
142
        /* Declare and initialize a security descriptor.
143
           This is required for multiuser systems on Windows 2000 SP4 and beyond */
144
        // TODO: UWP does not have this function and I do not think it is required?
145
#if U_PLATFORM_HAS_WINUWP_API == 0
146
        if (InitializeSecurityDescriptor(&securityDesc, SECURITY_DESCRIPTOR_REVISION)) {
147
            /* give the security descriptor a Null Dacl done using the  "TRUE, (PACL)NULL" here */
148
            if (SetSecurityDescriptorDacl(&securityDesc, TRUE, (PACL)NULL, FALSE)) {
149
                /* Make the security attributes point to the security descriptor */
150
                uprv_memset(&mappingAttributes, 0, sizeof(mappingAttributes));
151
                mappingAttributes.nLength = sizeof(mappingAttributes);
152
                mappingAttributes.lpSecurityDescriptor = &securityDesc;
153
                mappingAttributes.bInheritHandle = FALSE; /* object uninheritable */
154
                mappingAttributesPtr = &mappingAttributes;
155
            }
156
        }
157
        /* else creating security descriptors can fail when we are on Windows 98,
158
           and mappingAttributesPtr == NULL for that case. */
159
160
        /* create an unnamed Windows file-mapping object for the specified file */
161
        map=CreateFileMapping(file, mappingAttributesPtr, PAGE_READONLY, 0, 0, NULL);
162
#else
163
        map = CreateFileMappingFromApp(file, NULL, PAGE_READONLY, 0, NULL);
164
#endif
165
        CloseHandle(file);
166
        if(map==NULL) {
167
            return FALSE;
168
        }
169
170
        /* map a view of the file into our address space */
171
        pData->pHeader=(const DataHeader *)MapViewOfFile(map, FILE_MAP_READ, 0, 0, 0);
172
        if(pData->pHeader==NULL) {
173
            CloseHandle(map);
174
            return FALSE;
175
        }
176
        pData->map=map;
177
        return TRUE;
178
    }
179
180
    U_CFUNC void
181
    uprv_unmapFile(UDataMemory *pData) {
182
        if(pData!=NULL && pData->map!=NULL) {
183
            UnmapViewOfFile(pData->pHeader);
184
            CloseHandle(pData->map);
185
            pData->pHeader=NULL;
186
            pData->map=NULL;
187
        }
188
    }
189
190
191
192
#elif MAP_IMPLEMENTATION==MAP_POSIX
193
    U_CFUNC UBool
194
0
    uprv_mapFile(UDataMemory *pData, const char *path) {
195
0
        int fd;
196
0
        int length;
197
0
        struct stat mystat;
198
0
        void *data;
199
200
0
        UDataMemory_init(pData); /* Clear the output struct.        */
201
202
        /* determine the length of the file */
203
0
        if(stat(path, &mystat)!=0 || mystat.st_size<=0) {
204
0
            return FALSE;
205
0
        }
206
0
        length=mystat.st_size;
207
208
        /* open the file */
209
0
        fd=open(path, O_RDONLY);
210
0
        if(fd==-1) {
211
0
            return FALSE;
212
0
        }
213
214
        /* get a view of the mapping */
215
0
#if U_PLATFORM != U_PF_HPUX
216
0
        data=mmap(0, length, PROT_READ, MAP_SHARED,  fd, 0);
217
#else
218
        data=mmap(0, length, PROT_READ, MAP_PRIVATE, fd, 0);
219
#endif
220
0
        close(fd); /* no longer needed */
221
0
        if(data==MAP_FAILED) {
222
0
            return FALSE;
223
0
        }
224
225
0
        pData->map = (char *)data + length;
226
0
        pData->pHeader=(const DataHeader *)data;
227
0
        pData->mapAddr = data;
228
#if U_PLATFORM == U_PF_IPHONE
229
        posix_madvise(data, length, POSIX_MADV_RANDOM);
230
#endif
231
0
        return TRUE;
232
0
    }
233
234
    U_CFUNC void
235
2.38k
    uprv_unmapFile(UDataMemory *pData) {
236
2.38k
        if(pData!=NULL && pData->map!=NULL) {
237
0
            size_t dataLen = (char *)pData->map - (char *)pData->mapAddr;
238
0
            if(munmap(pData->mapAddr, dataLen)==-1) {
239
0
            }
240
0
            pData->pHeader=NULL;
241
0
            pData->map=0;
242
0
            pData->mapAddr=NULL;
243
0
        }
244
2.38k
    }
245
246
247
248
#elif MAP_IMPLEMENTATION==MAP_STDIO
249
    /* copy of the filestrm.c/T_FileStream_size() implementation */
250
    static int32_t
251
    umap_fsize(FILE *f) {
252
        int32_t savedPos = ftell(f);
253
        int32_t size = 0;
254
255
        /*Changes by Bertrand A. D. doesn't affect the current position
256
        goes to the end of the file before ftell*/
257
        fseek(f, 0, SEEK_END);
258
        size = (int32_t)ftell(f);
259
        fseek(f, savedPos, SEEK_SET);
260
        return size;
261
    }
262
263
    U_CFUNC UBool
264
    uprv_mapFile(UDataMemory *pData, const char *path) {
265
        FILE *file;
266
        int32_t fileLength;
267
        void *p;
268
269
        UDataMemory_init(pData); /* Clear the output struct.        */
270
        /* open the input file */
271
        file=fopen(path, "rb");
272
        if(file==NULL) {
273
            return FALSE;
274
        }
275
276
        /* get the file length */
277
        fileLength=umap_fsize(file);
278
        if(ferror(file) || fileLength<=20) {
279
            fclose(file);
280
            return FALSE;
281
        }
282
283
        /* allocate the memory to hold the file data */
284
        p=uprv_malloc(fileLength);
285
        if(p==NULL) {
286
            fclose(file);
287
            return FALSE;
288
        }
289
290
        /* read the file */
291
        if(fileLength!=fread(p, 1, fileLength, file)) {
292
            uprv_free(p);
293
            fclose(file);
294
            return FALSE;
295
        }
296
297
        fclose(file);
298
        pData->map=p;
299
        pData->pHeader=(const DataHeader *)p;
300
        pData->mapAddr=p;
301
        return TRUE;
302
    }
303
304
    U_CFUNC void
305
    uprv_unmapFile(UDataMemory *pData) {
306
        if(pData!=NULL && pData->map!=NULL) {
307
            uprv_free(pData->map);
308
            pData->map     = NULL;
309
            pData->mapAddr = NULL;
310
            pData->pHeader = NULL;
311
        }
312
    }
313
314
315
#elif MAP_IMPLEMENTATION==MAP_390DLL
316
    /*  390 specific Library Loading.
317
     *  This is the only platform left that dynamically loads an ICU Data Library.
318
     *  All other platforms use .data files when dynamic loading is required, but
319
     *  this turn out to be awkward to support in 390 batch mode.
320
     *
321
     *  The idea here is to hide the fact that 390 is using dll loading from the
322
     *   rest of ICU, and make it look like there is file loading happening.
323
     *
324
     */
325
326
    static char *strcpy_returnEnd(char *dest, const char *src)
327
    {
328
        while((*dest=*src)!=0) {
329
            ++dest;
330
            ++src;
331
        }
332
        return dest;
333
    }
334
    
335
    /*------------------------------------------------------------------------------
336
     *                                                                              
337
     *  computeDirPath   given a user-supplied path of an item to be opened,             
338
     *                         compute and return 
339
     *                            - the full directory path to be used 
340
     *                              when opening the file.
341
     *                            - Pointer to null at end of above returned path    
342
     *
343
     *                       Parameters:
344
     *                          path:        input path.  Buffer is not altered.
345
     *                          pathBuffer:  Output buffer.  Any contents are overwritten.
346
     *
347
     *                       Returns:
348
     *                          Pointer to null termination in returned pathBuffer.
349
     *
350
     *                    TODO:  This works the way ICU historically has, but the
351
     *                           whole data fallback search path is so complicated that
352
     *                           proabably almost no one will ever really understand it,
353
     *                           the potential for confusion is large.  (It's not just 
354
     *                           this one function, but the whole scheme.)
355
     *                            
356
     *------------------------------------------------------------------------------*/
357
    static char *uprv_computeDirPath(const char *path, char *pathBuffer)
358
    {
359
        char   *finalSlash;       /* Ptr to last dir separator in input path, or null if none. */
360
        int32_t pathLen;          /* Length of the returned directory path                     */
361
        
362
        finalSlash = 0;
363
        if (path != 0) {
364
            finalSlash = uprv_strrchr(path, U_FILE_SEP_CHAR);
365
        }
366
        
367
        *pathBuffer = 0;
368
        if (finalSlash == 0) {
369
        /* No user-supplied path.  
370
            * Copy the ICU_DATA path to the path buffer and return that*/
371
            const char *icuDataDir;
372
            icuDataDir=u_getDataDirectory();
373
            if(icuDataDir!=NULL && *icuDataDir!=0) {
374
                return strcpy_returnEnd(pathBuffer, icuDataDir);
375
            } else {
376
                /* there is no icuDataDir either.  Just return the empty pathBuffer. */
377
                return pathBuffer;
378
            }
379
        } 
380
        
381
        /* User supplied path did contain a directory portion.
382
        * Copy it to the output path buffer */
383
        pathLen = (int32_t)(finalSlash - path + 1);
384
        uprv_memcpy(pathBuffer, path, pathLen);
385
        *(pathBuffer+pathLen) = 0;
386
        return pathBuffer+pathLen;
387
    }
388
    
389
390
#   define DATA_TYPE "dat"
391
392
    U_CFUNC UBool uprv_mapFile(UDataMemory *pData, const char *path) {
393
        const char *inBasename;
394
        char *basename;
395
        char pathBuffer[1024];
396
        const DataHeader *pHeader;
397
        dllhandle *handle;
398
        void *val=0;
399
400
        inBasename=uprv_strrchr(path, U_FILE_SEP_CHAR);
401
        if(inBasename==NULL) {
402
            inBasename = path;
403
        } else {
404
            inBasename++;
405
        }
406
        basename=uprv_computeDirPath(path, pathBuffer);
407
        if(uprv_strcmp(inBasename, U_ICUDATA_NAME".dat") != 0) {
408
            /* must mmap file... for build */
409
            int fd;
410
            int length;
411
            struct stat mystat;
412
            void *data;
413
            UDataMemory_init(pData); /* Clear the output struct. */
414
415
            /* determine the length of the file */
416
            if(stat(path, &mystat)!=0 || mystat.st_size<=0) {
417
                return FALSE;
418
            }
419
            length=mystat.st_size;
420
421
            /* open the file */
422
            fd=open(path, O_RDONLY);
423
            if(fd==-1) {
424
                return FALSE;
425
            }
426
427
            /* get a view of the mapping */
428
            data=mmap(0, length, PROT_READ, MAP_PRIVATE, fd, 0);
429
            close(fd); /* no longer needed */
430
            if(data==MAP_FAILED) {
431
                return FALSE;
432
            }
433
            pData->map = (char *)data + length;
434
            pData->pHeader=(const DataHeader *)data;
435
            pData->mapAddr = data;
436
            return TRUE;
437
        }
438
439
#       ifdef OS390BATCH
440
            /* ### hack: we still need to get u_getDataDirectory() fixed
441
            for OS/390 (batch mode - always return "//"? )
442
            and this here straightened out with LIB_PREFIX and LIB_SUFFIX (both empty?!)
443
            This is probably due to the strange file system on OS/390.  It's more like
444
            a database with short entry names than a typical file system. */
445
            /* U_ICUDATA_NAME should always have the correct name */
446
            /* BUT FOR BATCH MODE IT IS AN EXCEPTION BECAUSE */
447
            /* THE FIRST THREE LETTERS ARE PREASSIGNED TO THE */
448
            /* PROJECT!!!!! */
449
            uprv_strcpy(pathBuffer, "IXMI" U_ICU_VERSION_SHORT "DA");
450
#       else
451
            /* set up the library name */
452
            uprv_strcpy(basename, LIB_PREFIX U_LIBICUDATA_NAME U_ICU_VERSION_SHORT LIB_SUFFIX);
453
#       endif
454
455
#       ifdef UDATA_DEBUG
456
             fprintf(stderr, "dllload: %s ", pathBuffer);
457
#       endif
458
459
        handle=dllload(pathBuffer);
460
461
#       ifdef UDATA_DEBUG
462
               fprintf(stderr, " -> %08X\n", handle );
463
#       endif
464
465
        if(handle != NULL) {
466
               /* we have a data DLL - what kind of lookup do we need here? */
467
               /* try to find the Table of Contents */
468
               UDataMemory_init(pData); /* Clear the output struct.        */
469
               val=dllqueryvar((dllhandle*)handle, U_ICUDATA_ENTRY_NAME);
470
               if(val == 0) {
471
                    /* failed... so keep looking */
472
                    return FALSE;
473
               }
474
#              ifdef UDATA_DEBUG
475
                    fprintf(stderr, "dllqueryvar(%08X, %s) -> %08X\n", handle, U_ICUDATA_ENTRY_NAME, val);
476
#              endif
477
478
               pData->pHeader=(const DataHeader *)val;
479
               return TRUE;
480
         } else {
481
               return FALSE; /* no handle */
482
         }
483
    }
484
485
    U_CFUNC void uprv_unmapFile(UDataMemory *pData) {
486
        if(pData!=NULL && pData->map!=NULL) {
487
            uprv_free(pData->map);
488
            pData->map     = NULL;
489
            pData->mapAddr = NULL;
490
            pData->pHeader = NULL;
491
        }   
492
    }
493
494
#else
495
#   error MAP_IMPLEMENTATION is set incorrectly
496
#endif