common.h 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413
  1. ///////////////////////////////////////////////////////////////////////////////
  2. //
  3. /// \file common.h
  4. /// \brief Definitions common to the whole liblzma library
  5. //
  6. // Author: Lasse Collin
  7. //
  8. // This file has been put into the public domain.
  9. // You can do whatever you want with this file.
  10. //
  11. ///////////////////////////////////////////////////////////////////////////////
  12. #ifndef LZMA_COMMON_H
  13. #define LZMA_COMMON_H
  14. #include "sysdefs.h"
  15. #include "mythread.h"
  16. #include "tuklib_integer.h"
  17. // LZMA_API_EXPORT is used to mark the exported API functions.
  18. // It's used to define the LZMA_API macro.
  19. //
  20. // lzma_attr_visibility_hidden is used for marking *declarations* of extern
  21. // variables that are internal to liblzma (-fvisibility=hidden alone is
  22. // enough to hide the *definitions*). Such markings allow slightly more
  23. // efficient code to accesses those variables in ELF shared libraries.
  24. #if defined(_WIN32) || defined(__CYGWIN__)
  25. # ifdef DLL_EXPORT
  26. # define LZMA_API_EXPORT __declspec(dllexport)
  27. # else
  28. # define LZMA_API_EXPORT
  29. # endif
  30. # define lzma_attr_visibility_hidden
  31. // Don't use ifdef or defined() below.
  32. #elif HAVE_VISIBILITY
  33. # define LZMA_API_EXPORT __attribute__((__visibility__("default")))
  34. # define lzma_attr_visibility_hidden \
  35. __attribute__((__visibility__("hidden")))
  36. #else
  37. # define LZMA_API_EXPORT
  38. # define lzma_attr_visibility_hidden
  39. #endif
  40. #define LZMA_API(type) LZMA_API_EXPORT type LZMA_API_CALL
  41. #include "lzma.h"
  42. // This is for detecting modern GCC and Clang attributes
  43. // like __symver__ in GCC >= 10.
  44. #ifdef __has_attribute
  45. # define lzma_has_attribute(attr) __has_attribute(attr)
  46. #else
  47. # define lzma_has_attribute(attr) 0
  48. #endif
  49. // The extra symbol versioning in the C files may only be used when
  50. // building a shared library. If HAVE_SYMBOL_VERSIONS_LINUX is defined
  51. // to 2 then symbol versioning is done only if also PIC is defined.
  52. // By default Libtool defines PIC when building a shared library and
  53. // doesn't define it when building a static library but it can be
  54. // overridden with --with-pic and --without-pic. configure let's rely
  55. // on PIC if neither --with-pic or --without-pic was used.
  56. #if defined(HAVE_SYMBOL_VERSIONS_LINUX) \
  57. && (HAVE_SYMBOL_VERSIONS_LINUX == 2 && !defined(PIC))
  58. # undef HAVE_SYMBOL_VERSIONS_LINUX
  59. #endif
  60. #ifdef HAVE_SYMBOL_VERSIONS_LINUX
  61. // To keep link-time optimization (LTO, -flto) working with GCC,
  62. // the __symver__ attribute must be used instead of __asm__(".symver ...").
  63. // Otherwise the symbol versions may be lost, resulting in broken liblzma
  64. // that has wrong default versions in the exported symbol list!
  65. // The attribute was added in GCC 10; LTO with older GCC is not supported.
  66. //
  67. // To keep -Wmissing-prototypes happy, use LZMA_SYMVER_API only with function
  68. // declarations (including those with __alias__ attribute) and LZMA_API with
  69. // the function definitions. This means a little bit of silly copy-and-paste
  70. // between declarations and definitions though.
  71. //
  72. // As of GCC 12.2, the __symver__ attribute supports only @ and @@ but the
  73. // very convenient @@@ isn't supported (it's supported by GNU assembler
  74. // since 2000). When using @@ instead of @@@, the internal name must not be
  75. // the same as the external name to avoid problems in some situations. This
  76. // is why "#define foo_52 foo" is needed for the default symbol versions.
  77. //
  78. // __has_attribute is supported before GCC 10 and it is supported in Clang 14
  79. // too (which doesn't support __symver__) so use it to detect if __symver__
  80. // is available. This should be far more reliable than looking at compiler
  81. // version macros as nowadays especially __GNUC__ is defined by many compilers.
  82. # if lzma_has_attribute(__symver__)
  83. # define LZMA_SYMVER_API(extnamever, type, intname) \
  84. extern __attribute__((__symver__(extnamever))) \
  85. LZMA_API(type) intname
  86. # else
  87. # define LZMA_SYMVER_API(extnamever, type, intname) \
  88. __asm__(".symver " #intname "," extnamever); \
  89. extern LZMA_API(type) intname
  90. # endif
  91. #endif
  92. // MSVC has __forceinline which shouldn't be combined with the inline keyword
  93. // (results in a warning).
  94. //
  95. // GCC 3.1 added always_inline attribute so we don't need to check
  96. // for __GNUC__ version. Similarly, all relevant Clang versions
  97. // support it (at least Clang 3.0.0 does already).
  98. // Other compilers might support too which also support __has_attribute
  99. // (Solaris Studio) so do that check too.
  100. #if defined(_MSC_VER)
  101. # define lzma_always_inline __forceinline
  102. #elif defined(__GNUC__) || defined(__clang__) || defined(__INTEL_COMPILER) \
  103. || lzma_has_attribute(__always_inline__)
  104. # define lzma_always_inline inline __attribute__((__always_inline__))
  105. #else
  106. # define lzma_always_inline inline
  107. #endif
  108. // These allow helping the compiler in some often-executed branches, whose
  109. // result is almost always the same.
  110. #ifdef __GNUC__
  111. # define likely(expr) __builtin_expect(expr, true)
  112. # define unlikely(expr) __builtin_expect(expr, false)
  113. #else
  114. # define likely(expr) (expr)
  115. # define unlikely(expr) (expr)
  116. #endif
  117. /// Size of temporary buffers needed in some filters
  118. #define LZMA_BUFFER_SIZE 4096
  119. /// Maximum number of worker threads within one multithreaded component.
  120. /// The limit exists solely to make it simpler to prevent integer overflows
  121. /// when allocating structures etc. This should be big enough for now...
  122. /// the code won't scale anywhere close to this number anyway.
  123. #define LZMA_THREADS_MAX 16384
  124. /// Starting value for memory usage estimates. Instead of calculating size
  125. /// of _every_ structure and taking into account malloc() overhead etc., we
  126. /// add a base size to all memory usage estimates. It's not very accurate
  127. /// but should be easily good enough.
  128. #define LZMA_MEMUSAGE_BASE (UINT64_C(1) << 15)
  129. /// Start of internal Filter ID space. These IDs must never be used
  130. /// in Streams.
  131. #define LZMA_FILTER_RESERVED_START (LZMA_VLI_C(1) << 62)
  132. /// Supported flags that can be passed to lzma_stream_decoder(),
  133. /// lzma_auto_decoder(), or lzma_stream_decoder_mt().
  134. #define LZMA_SUPPORTED_FLAGS \
  135. ( LZMA_TELL_NO_CHECK \
  136. | LZMA_TELL_UNSUPPORTED_CHECK \
  137. | LZMA_TELL_ANY_CHECK \
  138. | LZMA_IGNORE_CHECK \
  139. | LZMA_CONCATENATED \
  140. | LZMA_FAIL_FAST )
  141. /// Largest valid lzma_action value as unsigned integer.
  142. #define LZMA_ACTION_MAX ((unsigned int)(LZMA_FULL_BARRIER))
  143. /// Special return value (lzma_ret) to indicate that a timeout was reached
  144. /// and lzma_code() must not return LZMA_BUF_ERROR. This is converted to
  145. /// LZMA_OK in lzma_code().
  146. #define LZMA_TIMED_OUT LZMA_RET_INTERNAL1
  147. /// Special return value (lzma_ret) for use in stream_decoder_mt.c to
  148. /// indicate Index was detected instead of a Block Header.
  149. #define LZMA_INDEX_DETECTED LZMA_RET_INTERNAL2
  150. typedef struct lzma_next_coder_s lzma_next_coder;
  151. typedef struct lzma_filter_info_s lzma_filter_info;
  152. /// Type of a function used to initialize a filter encoder or decoder
  153. typedef lzma_ret (*lzma_init_function)(
  154. lzma_next_coder *next, const lzma_allocator *allocator,
  155. const lzma_filter_info *filters);
  156. /// Type of a function to do some kind of coding work (filters, Stream,
  157. /// Block encoders/decoders etc.). Some special coders use don't use both
  158. /// input and output buffers, but for simplicity they still use this same
  159. /// function prototype.
  160. typedef lzma_ret (*lzma_code_function)(
  161. void *coder, const lzma_allocator *allocator,
  162. const uint8_t *restrict in, size_t *restrict in_pos,
  163. size_t in_size, uint8_t *restrict out,
  164. size_t *restrict out_pos, size_t out_size,
  165. lzma_action action);
  166. /// Type of a function to free the memory allocated for the coder
  167. typedef void (*lzma_end_function)(
  168. void *coder, const lzma_allocator *allocator);
  169. /// Raw coder validates and converts an array of lzma_filter structures to
  170. /// an array of lzma_filter_info structures. This array is used with
  171. /// lzma_next_filter_init to initialize the filter chain.
  172. struct lzma_filter_info_s {
  173. /// Filter ID. This can be used to share the same initiazation
  174. /// function *and* data structures with different Filter IDs
  175. /// (LZMA_FILTER_LZMA1EXT does it), and also by the encoder
  176. /// with lzma_filters_update() if filter chain is updated
  177. /// in the middle of a raw stream or Block (LZMA_SYNC_FLUSH).
  178. lzma_vli id;
  179. /// Pointer to function used to initialize the filter.
  180. /// This is NULL to indicate end of array.
  181. lzma_init_function init;
  182. /// Pointer to filter's options structure
  183. void *options;
  184. };
  185. /// Hold data and function pointers of the next filter in the chain.
  186. struct lzma_next_coder_s {
  187. /// Pointer to coder-specific data
  188. void *coder;
  189. /// Filter ID. This is LZMA_VLI_UNKNOWN when this structure doesn't
  190. /// point to a filter coder.
  191. lzma_vli id;
  192. /// "Pointer" to init function. This is never called here.
  193. /// We need only to detect if we are initializing a coder
  194. /// that was allocated earlier. See lzma_next_coder_init and
  195. /// lzma_next_strm_init macros in this file.
  196. uintptr_t init;
  197. /// Pointer to function to do the actual coding
  198. lzma_code_function code;
  199. /// Pointer to function to free lzma_next_coder.coder. This can
  200. /// be NULL; in that case, lzma_free is called to free
  201. /// lzma_next_coder.coder.
  202. lzma_end_function end;
  203. /// Pointer to a function to get progress information. If this is NULL,
  204. /// lzma_stream.total_in and .total_out are used instead.
  205. void (*get_progress)(void *coder,
  206. uint64_t *progress_in, uint64_t *progress_out);
  207. /// Pointer to function to return the type of the integrity check.
  208. /// Most coders won't support this.
  209. lzma_check (*get_check)(const void *coder);
  210. /// Pointer to function to get and/or change the memory usage limit.
  211. /// If new_memlimit == 0, the limit is not changed.
  212. lzma_ret (*memconfig)(void *coder, uint64_t *memusage,
  213. uint64_t *old_memlimit, uint64_t new_memlimit);
  214. /// Update the filter-specific options or the whole filter chain
  215. /// in the encoder.
  216. lzma_ret (*update)(void *coder, const lzma_allocator *allocator,
  217. const lzma_filter *filters,
  218. const lzma_filter *reversed_filters);
  219. /// Set how many bytes of output this coder may produce at maximum.
  220. /// On success LZMA_OK must be returned.
  221. /// If the filter chain as a whole cannot support this feature,
  222. /// this must return LZMA_OPTIONS_ERROR.
  223. /// If no input has been given to the coder and the requested limit
  224. /// is too small, this must return LZMA_BUF_ERROR. If input has been
  225. /// seen, LZMA_OK is allowed too.
  226. lzma_ret (*set_out_limit)(void *coder, uint64_t *uncomp_size,
  227. uint64_t out_limit);
  228. };
  229. /// Macro to initialize lzma_next_coder structure
  230. #define LZMA_NEXT_CODER_INIT \
  231. (lzma_next_coder){ \
  232. .coder = NULL, \
  233. .init = (uintptr_t)(NULL), \
  234. .id = LZMA_VLI_UNKNOWN, \
  235. .code = NULL, \
  236. .end = NULL, \
  237. .get_progress = NULL, \
  238. .get_check = NULL, \
  239. .memconfig = NULL, \
  240. .update = NULL, \
  241. .set_out_limit = NULL, \
  242. }
  243. /// Internal data for lzma_strm_init, lzma_code, and lzma_end. A pointer to
  244. /// this is stored in lzma_stream.
  245. struct lzma_internal_s {
  246. /// The actual coder that should do something useful
  247. lzma_next_coder next;
  248. /// Track the state of the coder. This is used to validate arguments
  249. /// so that the actual coders can rely on e.g. that LZMA_SYNC_FLUSH
  250. /// is used on every call to lzma_code until next.code has returned
  251. /// LZMA_STREAM_END.
  252. enum {
  253. ISEQ_RUN,
  254. ISEQ_SYNC_FLUSH,
  255. ISEQ_FULL_FLUSH,
  256. ISEQ_FINISH,
  257. ISEQ_FULL_BARRIER,
  258. ISEQ_END,
  259. ISEQ_ERROR,
  260. } sequence;
  261. /// A copy of lzma_stream avail_in. This is used to verify that the
  262. /// amount of input doesn't change once e.g. LZMA_FINISH has been
  263. /// used.
  264. size_t avail_in;
  265. /// Indicates which lzma_action values are allowed by next.code.
  266. bool supported_actions[LZMA_ACTION_MAX + 1];
  267. /// If true, lzma_code will return LZMA_BUF_ERROR if no progress was
  268. /// made (no input consumed and no output produced by next.code).
  269. bool allow_buf_error;
  270. };
  271. /// Allocates memory
  272. lzma_attr_alloc_size(1)
  273. extern void *lzma_alloc(size_t size, const lzma_allocator *allocator);
  274. /// Allocates memory and zeroes it (like calloc()). This can be faster
  275. /// than lzma_alloc() + memzero() while being backward compatible with
  276. /// custom allocators.
  277. lzma_attr_alloc_size(1)
  278. extern void *lzma_alloc_zero(size_t size, const lzma_allocator *allocator);
  279. /// Frees memory
  280. extern void lzma_free(void *ptr, const lzma_allocator *allocator);
  281. /// Allocates strm->internal if it is NULL, and initializes *strm and
  282. /// strm->internal. This function is only called via lzma_next_strm_init macro.
  283. extern lzma_ret lzma_strm_init(lzma_stream *strm);
  284. /// Initializes the next filter in the chain, if any. This takes care of
  285. /// freeing the memory of previously initialized filter if it is different
  286. /// than the filter being initialized now. This way the actual filter
  287. /// initialization functions don't need to use lzma_next_coder_init macro.
  288. extern lzma_ret lzma_next_filter_init(lzma_next_coder *next,
  289. const lzma_allocator *allocator,
  290. const lzma_filter_info *filters);
  291. /// Update the next filter in the chain, if any. This checks that
  292. /// the application is not trying to change the Filter IDs.
  293. extern lzma_ret lzma_next_filter_update(
  294. lzma_next_coder *next, const lzma_allocator *allocator,
  295. const lzma_filter *reversed_filters);
  296. /// Frees the memory allocated for next->coder either using next->end or,
  297. /// if next->end is NULL, using lzma_free.
  298. extern void lzma_next_end(lzma_next_coder *next,
  299. const lzma_allocator *allocator);
  300. /// Copy as much data as possible from in[] to out[] and update *in_pos
  301. /// and *out_pos accordingly. Returns the number of bytes copied.
  302. extern size_t lzma_bufcpy(const uint8_t *restrict in, size_t *restrict in_pos,
  303. size_t in_size, uint8_t *restrict out,
  304. size_t *restrict out_pos, size_t out_size);
  305. /// \brief Return if expression doesn't evaluate to LZMA_OK
  306. ///
  307. /// There are several situations where we want to return immediately
  308. /// with the value of expr if it isn't LZMA_OK. This macro shortens
  309. /// the code a little.
  310. #define return_if_error(expr) \
  311. do { \
  312. const lzma_ret ret_ = (expr); \
  313. if (ret_ != LZMA_OK) \
  314. return ret_; \
  315. } while (0)
  316. /// If next isn't already initialized, free the previous coder. Then mark
  317. /// that next is _possibly_ initialized for the coder using this macro.
  318. /// "Possibly" means that if e.g. allocation of next->coder fails, the
  319. /// structure isn't actually initialized for this coder, but leaving
  320. /// next->init to func is still OK.
  321. #define lzma_next_coder_init(func, next, allocator) \
  322. do { \
  323. if ((uintptr_t)(func) != (next)->init) \
  324. lzma_next_end(next, allocator); \
  325. (next)->init = (uintptr_t)(func); \
  326. } while (0)
  327. /// Initializes lzma_strm and calls func() to initialize strm->internal->next.
  328. /// (The function being called will use lzma_next_coder_init()). If
  329. /// initialization fails, memory that wasn't freed by func() is freed
  330. /// along strm->internal.
  331. #define lzma_next_strm_init(func, strm, ...) \
  332. do { \
  333. return_if_error(lzma_strm_init(strm)); \
  334. const lzma_ret ret_ = func(&(strm)->internal->next, \
  335. (strm)->allocator, __VA_ARGS__); \
  336. if (ret_ != LZMA_OK) { \
  337. lzma_end(strm); \
  338. return ret_; \
  339. } \
  340. } while (0)
  341. #endif