ghc/rts/Linker.c

   1 /* -----------------------------------------------------------------------------
   2  *
   3  * (c) The GHC Team, 2000-2004
   4  *
   5  * RTS Object Linker
   6  *
   7  * ---------------------------------------------------------------------------*/
   8
   9 #if 0
  10 #include "PosixSource.h"
  11 #endif
  12
  13 /* Linux needs _GNU_SOURCE to get RTLD_DEFAULT from <dlfcn.h> and
  14    MREMAP_MAYMOVE from <sys/mman.h>.
  15  */
  16 #ifdef __linux__
  17 #define _GNU_SOURCE
  18 #endif
  19
  20 #include "Rts.h"
  21 #include "RtsFlags.h"
  22 #include "HsFFI.h"
  23 #include "Hash.h"
  24 #include "Linker.h"
  25 #include "LinkerInternals.h"
  26 #include "RtsUtils.h"
  27 #include "Schedule.h"
  28 #include "Storage.h"
  29
  30 #ifdef HAVE_SYS_TYPES_H
  31 #include <sys/types.h>
  32 #endif
  33
  34 #include <stdlib.h>
  35 #include <string.h>
  36
  37 #ifdef HAVE_SYS_STAT_H
  38 #include <sys/stat.h>
  39 #endif
  40
  41 #if defined(HAVE_DLFCN_H)
  42 #include <dlfcn.h>
  43 #endif
  44
  45 #if defined(cygwin32_HOST_OS)
  46 #ifdef HAVE_DIRENT_H
  47 #include <dirent.h>
  48 #endif
  49
  50 #ifdef HAVE_SYS_TIME_H
  51 #include <sys/time.h>
  52 #endif
  53 #include <regex.h>
  54 #include <sys/fcntl.h>
  55 #include <sys/termios.h>
  56 #include <sys/utime.h>
  57 #include <sys/utsname.h>
  58 #include <sys/wait.h>
  59 #endif
  60
  61 #if defined(ia64_HOST_ARCH) || defined(openbsd_HOST_OS) || defined(linux_HOST_OS) || defined(freebsd_HOST_OS)
  62 #define USE_MMAP
  63 #include <fcntl.h>
  64 #include <sys/mman.h>
  65
  66 #if defined(openbsd_HOST_OS) || defined(linux_HOST_OS) || defined(freebsd_HOST_OS)
  67 #ifdef HAVE_UNISTD_H
  68 #include <unistd.h>
  69 #endif
  70 #endif
  71
  72 #endif
  73
  74 #if defined(linux_HOST_OS) || defined(solaris2_HOST_OS) || defined(freebsd_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  75 #  define OBJFORMAT_ELF
  76 #elif defined(cygwin32_HOST_OS) || defined (mingw32_HOST_OS)
  77 #  define OBJFORMAT_PEi386
  78 #  include <windows.h>
  79 #  include <math.h>
  80 #elif defined(darwin_HOST_OS)
  81 #  include <mach-o/ppc/reloc.h>
  82 #  define OBJFORMAT_MACHO
  83 #  include <mach-o/loader.h>
  84 #  include <mach-o/nlist.h>
  85 #  include <mach-o/reloc.h>
  86 #  include <mach-o/dyld.h>
  87 #endif
  88
  89 /* Hash table mapping symbol names to Symbol */
  90 static /*Str*/HashTable *symhash;
  91
  92 /* List of currently loaded objects */
  93 ObjectCode *objects = NULL;     /* initially empty */
  94
  95 #if defined(OBJFORMAT_ELF)
  96 static int ocVerifyImage_ELF    ( ObjectCode* oc );
  97 static int ocGetNames_ELF       ( ObjectCode* oc );
  98 static int ocResolve_ELF        ( ObjectCode* oc );
  99 #if defined(powerpc_HOST_ARCH)
 100 static int ocAllocateJumpIslands_ELF ( ObjectCode* oc );
 101 #endif
 102 #elif defined(OBJFORMAT_PEi386)
 103 static int ocVerifyImage_PEi386 ( ObjectCode* oc );
 104 static int ocGetNames_PEi386    ( ObjectCode* oc );
 105 static int ocResolve_PEi386     ( ObjectCode* oc );
 106 #elif defined(OBJFORMAT_MACHO)
 107 static int ocAllocateJumpIslands_MachO ( ObjectCode* oc );
 108 static int ocVerifyImage_MachO    ( ObjectCode* oc );
 109 static int ocGetNames_MachO       ( ObjectCode* oc );
 110 static int ocResolve_MachO        ( ObjectCode* oc );
 111
 112 static void machoInitSymbolsWithoutUnderscore( void );
 113 #endif
 114
 115 /* -----------------------------------------------------------------------------
 116  * Built-in symbols from the RTS
 117  */
 118
 119 typedef struct _RtsSymbolVal {
 120     char   *lbl;
 121     void   *addr;
 122 } RtsSymbolVal;
 123
 124
 125 #if !defined(PAR)
 126 #define Maybe_ForeignObj        SymX(mkForeignObjzh_fast)
 127
 128 #define Maybe_Stable_Names      SymX(mkWeakzh_fast)                     \
 129                                 SymX(makeStableNamezh_fast)             \
 130                                 SymX(finalizzeWeakzh_fast)
 131 #else
 132 /* These are not available in GUM!!! -- HWL */
 133 #define Maybe_ForeignObj
 134 #define Maybe_Stable_Names
 135 #endif
 136
 137 #if !defined (mingw32_HOST_OS)
 138 #define RTS_POSIX_ONLY_SYMBOLS                  \
 139       SymX(stg_sig_install)                     \
 140       Sym(nocldstop)
 141 #endif
 142
 143 #if defined (cygwin32_HOST_OS)
 144 #define RTS_MINGW_ONLY_SYMBOLS /**/
 145 /* Don't have the ability to read import libs / archives, so
 146  * we have to stupidly list a lot of what libcygwin.a
 147  * exports; sigh.
 148  */
 149 #define RTS_CYGWIN_ONLY_SYMBOLS                 \
 150       SymX(regfree)                             \
 151       SymX(regexec)                             \
 152       SymX(regerror)                            \
 153       SymX(regcomp)                             \
 154       SymX(__errno)                             \
 155       SymX(access)                              \
 156       SymX(chmod)                               \
 157       SymX(chdir)                               \
 158       SymX(close)                               \
 159       SymX(creat)                               \
 160       SymX(dup)                                 \
 161       SymX(dup2)                                \
 162       SymX(fstat)                               \
 163       SymX(fcntl)                               \
 164       SymX(getcwd)                              \
 165       SymX(getenv)                              \
 166       SymX(lseek)                               \
 167       SymX(open)                                \
 168       SymX(fpathconf)                           \
 169       SymX(pathconf)                            \
 170       SymX(stat)                                \
 171       SymX(pow)                                 \
 172       SymX(tanh)                                \
 173       SymX(cosh)                                \
 174       SymX(sinh)                                \
 175       SymX(atan)                                \
 176       SymX(acos)                                \
 177       SymX(asin)                                \
 178       SymX(tan)                                 \
 179       SymX(cos)                                 \
 180       SymX(sin)                                 \
 181       SymX(exp)                                 \
 182       SymX(log)                                 \
 183       SymX(sqrt)                                \
 184       SymX(localtime_r)                         \
 185       SymX(gmtime_r)                            \
 186       SymX(mktime)                              \
 187       Sym(_imp___tzname)                        \
 188       SymX(gettimeofday)                        \
 189       SymX(timezone)                            \
 190       SymX(tcgetattr)                           \
 191       SymX(tcsetattr)                           \
 192       SymX(memcpy)                              \
 193       SymX(memmove)                             \
 194       SymX(realloc)                             \
 195       SymX(malloc)                              \
 196       SymX(free)                                \
 197       SymX(fork)                                \
 198       SymX(lstat)                               \
 199       SymX(isatty)                              \
 200       SymX(mkdir)                               \
 201       SymX(opendir)                             \
 202       SymX(readdir)                             \
 203       SymX(rewinddir)                           \
 204       SymX(closedir)                            \
 205       SymX(link)                                \
 206       SymX(mkfifo)                              \
 207       SymX(pipe)                                \
 208       SymX(read)                                \
 209       SymX(rename)                              \
 210       SymX(rmdir)                               \
 211       SymX(select)                              \
 212       SymX(system)                              \
 213       SymX(write)                               \
 214       SymX(strcmp)                              \
 215       SymX(strcpy)                              \
 216       SymX(strncpy)                             \
 217       SymX(strerror)                            \
 218       SymX(sigaddset)                           \
 219       SymX(sigemptyset)                         \
 220       SymX(sigprocmask)                         \
 221       SymX(umask)                               \
 222       SymX(uname)                               \
 223       SymX(unlink)                              \
 224       SymX(utime)                               \
 225       SymX(waitpid)
 226
 227 #elif !defined(mingw32_HOST_OS)
 228 #define RTS_MINGW_ONLY_SYMBOLS /**/
 229 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 230 #else /* defined(mingw32_HOST_OS) */
 231 #define RTS_POSIX_ONLY_SYMBOLS  /**/
 232 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 233
 234 /* Extra syms gen'ed by mingw-2's gcc-3.2: */
 235 #if __GNUC__>=3
 236 #define RTS_MINGW_EXTRA_SYMS                    \
 237       Sym(_imp____mb_cur_max)                   \
 238       Sym(_imp___pctype)
 239 #else
 240 #define RTS_MINGW_EXTRA_SYMS
 241 #endif
 242
 243 /* These are statically linked from the mingw libraries into the ghc
 244    executable, so we have to employ this hack. */
 245 #define RTS_MINGW_ONLY_SYMBOLS                  \
 246       SymX(asyncReadzh_fast)                    \
 247       SymX(asyncWritezh_fast)                   \
 248       SymX(asyncDoProczh_fast)                  \
 249       SymX(memset)                              \
 250       SymX(inet_ntoa)                           \
 251       SymX(inet_addr)                           \
 252       SymX(htonl)                               \
 253       SymX(recvfrom)                            \
 254       SymX(listen)                              \
 255       SymX(bind)                                \
 256       SymX(shutdown)                            \
 257       SymX(connect)                             \
 258       SymX(htons)                               \
 259       SymX(ntohs)                               \
 260       SymX(getservbyname)                       \
 261       SymX(getservbyport)                       \
 262       SymX(getprotobynumber)                    \
 263       SymX(getprotobyname)                      \
 264       SymX(gethostbyname)                       \
 265       SymX(gethostbyaddr)                       \
 266       SymX(gethostname)                         \
 267       SymX(strcpy)                              \
 268       SymX(strncpy)                             \
 269       SymX(abort)                               \
 270       Sym(_alloca)                              \
 271       Sym(isxdigit)                             \
 272       Sym(isupper)                              \
 273       Sym(ispunct)                              \
 274       Sym(islower)                              \
 275       Sym(isspace)                              \
 276       Sym(isprint)                              \
 277       Sym(isdigit)                              \
 278       Sym(iscntrl)                              \
 279       Sym(isalpha)                              \
 280       Sym(isalnum)                              \
 281       SymX(strcmp)                              \
 282       SymX(memmove)                             \
 283       SymX(realloc)                             \
 284       SymX(malloc)                              \
 285       SymX(pow)                                 \
 286       SymX(tanh)                                \
 287       SymX(cosh)                                \
 288       SymX(sinh)                                \
 289       SymX(atan)                                \
 290       SymX(acos)                                \
 291       SymX(asin)                                \
 292       SymX(tan)                                 \
 293       SymX(cos)                                 \
 294       SymX(sin)                                 \
 295       SymX(exp)                                 \
 296       SymX(log)                                 \
 297       SymX(sqrt)                                \
 298       SymX(memcpy)                              \
 299       SymX(rts_InstallConsoleEvent)             \
 300       SymX(rts_ConsoleHandlerDone)              \
 301       Sym(mktime)                               \
 302       Sym(_imp___timezone)                      \
 303       Sym(_imp___tzname)                        \
 304       Sym(_imp___iob)                           \
 305       Sym(_imp___osver)                         \
 306       Sym(localtime)                            \
 307       Sym(gmtime)                               \
 308       Sym(opendir)                              \
 309       Sym(readdir)                              \
 310       Sym(rewinddir)                            \
 311       RTS_MINGW_EXTRA_SYMS                      \
 312       Sym(closedir)
 313 #endif
 314
 315 #ifndef SMP
 316 # define MAIN_CAP_SYM SymX(MainCapability)
 317 #else
 318 # define MAIN_CAP_SYM
 319 #endif
 320
 321 #if !defined(mingw32_HOST_OS)
 322 #define RTS_USER_SIGNALS_SYMBOLS \
 323    SymX(startSignalHandler) \
 324    SymX(setIOManagerPipe)
 325 #else
 326 #define RTS_USER_SIGNALS_SYMBOLS /* nothing */
 327 #endif
 328
 329 #ifdef TABLES_NEXT_TO_CODE
 330 #define RTS_RET_SYMBOLS /* nothing */
 331 #else
 332 #define RTS_RET_SYMBOLS                         \
 333       SymX(stg_enter_ret)                       \
 334       SymX(stg_gc_fun_ret)                      \
 335       SymX(stg_ap_0_ret)                        \
 336       SymX(stg_ap_v_ret)                        \
 337       SymX(stg_ap_f_ret)                        \
 338       SymX(stg_ap_d_ret)                        \
 339       SymX(stg_ap_l_ret)                        \
 340       SymX(stg_ap_n_ret)                        \
 341       SymX(stg_ap_p_ret)                        \
 342       SymX(stg_ap_pv_ret)                       \
 343       SymX(stg_ap_pp_ret)                       \
 344       SymX(stg_ap_ppv_ret)                      \
 345       SymX(stg_ap_ppp_ret)                      \
 346       SymX(stg_ap_pppv_ret)                     \
 347       SymX(stg_ap_pppp_ret)                     \
 348       SymX(stg_ap_ppppp_ret)                    \
 349       SymX(stg_ap_pppppp_ret)
 350 #endif
 351
 352 #define RTS_SYMBOLS                             \
 353       Maybe_ForeignObj                          \
 354       Maybe_Stable_Names                        \
 355       Sym(StgReturn)                            \
 356       SymX(stg_enter_info)                      \
 357       SymX(stg_gc_void_info)                    \
 358       SymX(__stg_gc_enter_1)                    \
 359       SymX(stg_gc_noregs)                       \
 360       SymX(stg_gc_unpt_r1_info)                 \
 361       SymX(stg_gc_unpt_r1)                      \
 362       SymX(stg_gc_unbx_r1_info)                 \
 363       SymX(stg_gc_unbx_r1)                      \
 364       SymX(stg_gc_f1_info)                      \
 365       SymX(stg_gc_f1)                           \
 366       SymX(stg_gc_d1_info)                      \
 367       SymX(stg_gc_d1)                           \
 368       SymX(stg_gc_l1_info)                      \
 369       SymX(stg_gc_l1)                           \
 370       SymX(__stg_gc_fun)                        \
 371       SymX(stg_gc_fun_info)                     \
 372       SymX(stg_gc_gen)                          \
 373       SymX(stg_gc_gen_info)                     \
 374       SymX(stg_gc_gen_hp)                       \
 375       SymX(stg_gc_ut)                           \
 376       SymX(stg_gen_yield)                       \
 377       SymX(stg_yield_noregs)                    \
 378       SymX(stg_yield_to_interpreter)            \
 379       SymX(stg_gen_block)                       \
 380       SymX(stg_block_noregs)                    \
 381       SymX(stg_block_1)                         \
 382       SymX(stg_block_takemvar)                  \
 383       SymX(stg_block_putmvar)                   \
 384       SymX(stg_seq_frame_info)                  \
 385       MAIN_CAP_SYM                              \
 386       SymX(MallocFailHook)                      \
 387       SymX(OnExitHook)                          \
 388       SymX(OutOfHeapHook)                       \
 389       SymX(StackOverflowHook)                   \
 390       SymX(__encodeDouble)                      \
 391       SymX(__encodeFloat)                       \
 392       SymX(addDLL)                              \
 393       SymX(__gmpn_gcd_1)                        \
 394       SymX(__gmpz_cmp)                          \
 395       SymX(__gmpz_cmp_si)                       \
 396       SymX(__gmpz_cmp_ui)                       \
 397       SymX(__gmpz_get_si)                       \
 398       SymX(__gmpz_get_ui)                       \
 399       SymX(__int_encodeDouble)                  \
 400       SymX(__int_encodeFloat)                   \
 401       SymX(andIntegerzh_fast)                   \
 402       SymX(atomicallyzh_fast)                   \
 403       SymX(barf)                                \
 404       SymX(debugBelch)                          \
 405       SymX(errorBelch)                          \
 406       SymX(blockAsyncExceptionszh_fast)         \
 407       SymX(catchzh_fast)                        \
 408       SymX(catchRetryzh_fast)                   \
 409       SymX(catchSTMzh_fast)                     \
 410       SymX(closure_flags)                       \
 411       SymX(cmp_thread)                          \
 412       SymX(cmpIntegerzh_fast)                   \
 413       SymX(cmpIntegerIntzh_fast)                \
 414       SymX(complementIntegerzh_fast)            \
 415       SymX(createAdjustor)                      \
 416       SymX(decodeDoublezh_fast)                 \
 417       SymX(decodeFloatzh_fast)                  \
 418       SymX(defaultsHook)                        \
 419       SymX(delayzh_fast)                        \
 420       SymX(deRefWeakzh_fast)                    \
 421       SymX(deRefStablePtrzh_fast)               \
 422       SymX(divExactIntegerzh_fast)              \
 423       SymX(divModIntegerzh_fast)                \
 424       SymX(forkzh_fast)                         \
 425       SymX(forkProcess)                         \
 426       SymX(forkOS_createThread)                 \
 427       SymX(freeHaskellFunctionPtr)              \
 428       SymX(freeStablePtr)                       \
 429       SymX(gcdIntegerzh_fast)                   \
 430       SymX(gcdIntegerIntzh_fast)                \
 431       SymX(gcdIntzh_fast)                       \
 432       SymX(genSymZh)                            \
 433       SymX(genericRaise)                        \
 434       SymX(getProgArgv)                         \
 435       SymX(getStablePtr)                        \
 436       SymX(hs_init)                             \
 437       SymX(hs_exit)                             \
 438       SymX(hs_set_argv)                         \
 439       SymX(hs_add_root)                         \
 440       SymX(hs_perform_gc)                       \
 441       SymX(hs_free_stable_ptr)                  \
 442       SymX(hs_free_fun_ptr)                     \
 443       SymX(initLinker)                          \
 444       SymX(int2Integerzh_fast)                  \
 445       SymX(integer2Intzh_fast)                  \
 446       SymX(integer2Wordzh_fast)                 \
 447       SymX(isCurrentThreadBoundzh_fast)         \
 448       SymX(isDoubleDenormalized)                \
 449       SymX(isDoubleInfinite)                    \
 450       SymX(isDoubleNaN)                         \
 451       SymX(isDoubleNegativeZero)                \
 452       SymX(isEmptyMVarzh_fast)                  \
 453       SymX(isFloatDenormalized)                 \
 454       SymX(isFloatInfinite)                     \
 455       SymX(isFloatNaN)                          \
 456       SymX(isFloatNegativeZero)                 \
 457       SymX(killThreadzh_fast)                   \
 458       SymX(loadObj)                             \
 459       SymX(lookupSymbol)                        \
 460       SymX(makeStablePtrzh_fast)                \
 461       SymX(minusIntegerzh_fast)                 \
 462       SymX(mkApUpd0zh_fast)                     \
 463       SymX(myThreadIdzh_fast)                   \
 464       SymX(labelThreadzh_fast)                  \
 465       SymX(newArrayzh_fast)                     \
 466       SymX(newBCOzh_fast)                       \
 467       SymX(newByteArrayzh_fast)                 \
 468       SymX_redirect(newCAF, newDynCAF)          \
 469       SymX(newMVarzh_fast)                      \
 470       SymX(newMutVarzh_fast)                    \
 471       SymX(newTVarzh_fast)                      \
 472       SymX(atomicModifyMutVarzh_fast)           \
 473       SymX(newPinnedByteArrayzh_fast)           \
 474       SymX(orIntegerzh_fast)                    \
 475       SymX(performGC)                           \
 476       SymX(performMajorGC)                      \
 477       SymX(plusIntegerzh_fast)                  \
 478       SymX(prog_argc)                           \
 479       SymX(prog_argv)                           \
 480       SymX(putMVarzh_fast)                      \
 481       SymX(quotIntegerzh_fast)                  \
 482       SymX(quotRemIntegerzh_fast)               \
 483       SymX(raisezh_fast)                        \
 484       SymX(raiseIOzh_fast)                      \
 485       SymX(readTVarzh_fast)                     \
 486       SymX(remIntegerzh_fast)                   \
 487       SymX(resetNonBlockingFd)                  \
 488       SymX(resumeThread)                        \
 489       SymX(resolveObjs)                         \
 490       SymX(retryzh_fast)                        \
 491       SymX(rts_apply)                           \
 492       SymX(rts_checkSchedStatus)                \
 493       SymX(rts_eval)                            \
 494       SymX(rts_evalIO)                          \
 495       SymX(rts_evalLazyIO)                      \
 496       SymX(rts_evalStableIO)                    \
 497       SymX(rts_eval_)                           \
 498       SymX(rts_getBool)                         \
 499       SymX(rts_getChar)                         \
 500       SymX(rts_getDouble)                       \
 501       SymX(rts_getFloat)                        \
 502       SymX(rts_getInt)                          \
 503       SymX(rts_getInt32)                        \
 504       SymX(rts_getPtr)                          \
 505       SymX(rts_getFunPtr)                       \
 506       SymX(rts_getStablePtr)                    \
 507       SymX(rts_getThreadId)                     \
 508       SymX(rts_getWord)                         \
 509       SymX(rts_getWord32)                       \
 510       SymX(rts_lock)                            \
 511       SymX(rts_mkBool)                          \
 512       SymX(rts_mkChar)                          \
 513       SymX(rts_mkDouble)                        \
 514       SymX(rts_mkFloat)                         \
 515       SymX(rts_mkInt)                           \
 516       SymX(rts_mkInt16)                         \
 517       SymX(rts_mkInt32)                         \
 518       SymX(rts_mkInt64)                         \
 519       SymX(rts_mkInt8)                          \
 520       SymX(rts_mkPtr)                           \
 521       SymX(rts_mkFunPtr)                        \
 522       SymX(rts_mkStablePtr)                     \
 523       SymX(rts_mkString)                        \
 524       SymX(rts_mkWord)                          \
 525       SymX(rts_mkWord16)                        \
 526       SymX(rts_mkWord32)                        \
 527       SymX(rts_mkWord64)                        \
 528       SymX(rts_mkWord8)                         \
 529       SymX(rts_unlock)                          \
 530       SymX(rtsSupportsBoundThreads)             \
 531       SymX(run_queue_hd)                        \
 532       SymX(__hscore_get_saved_termios)          \
 533       SymX(__hscore_set_saved_termios)          \
 534       SymX(setProgArgv)                         \
 535       SymX(startupHaskell)                      \
 536       SymX(shutdownHaskell)                     \
 537       SymX(shutdownHaskellAndExit)              \
 538       SymX(stable_ptr_table)                    \
 539       SymX(stackOverflow)                       \
 540       SymX(stg_CAF_BLACKHOLE_info)              \
 541       SymX(awakenBlockedQueue)                  \
 542       SymX(stg_CHARLIKE_closure)                \
 543       SymX(stg_EMPTY_MVAR_info)                 \
 544       SymX(stg_IND_STATIC_info)                 \
 545       SymX(stg_INTLIKE_closure)                 \
 546       SymX(stg_MUT_ARR_PTRS_FROZEN_info)        \
 547       SymX(stg_MUT_ARR_PTRS_FROZEN0_info)       \
 548       SymX(stg_WEAK_info)                       \
 549       SymX(stg_ap_0_info)                       \
 550       SymX(stg_ap_v_info)                       \
 551       SymX(stg_ap_f_info)                       \
 552       SymX(stg_ap_d_info)                       \
 553       SymX(stg_ap_l_info)                       \
 554       SymX(stg_ap_n_info)                       \
 555       SymX(stg_ap_p_info)                       \
 556       SymX(stg_ap_pv_info)                      \
 557       SymX(stg_ap_pp_info)                      \
 558       SymX(stg_ap_ppv_info)                     \
 559       SymX(stg_ap_ppp_info)                     \
 560       SymX(stg_ap_pppv_info)                    \
 561       SymX(stg_ap_pppp_info)                    \
 562       SymX(stg_ap_ppppp_info)                   \
 563       SymX(stg_ap_pppppp_info)                  \
 564       SymX(stg_ap_1_upd_info)                   \
 565       SymX(stg_ap_2_upd_info)                   \
 566       SymX(stg_ap_3_upd_info)                   \
 567       SymX(stg_ap_4_upd_info)                   \
 568       SymX(stg_ap_5_upd_info)                   \
 569       SymX(stg_ap_6_upd_info)                   \
 570       SymX(stg_ap_7_upd_info)                   \
 571       SymX(stg_exit)                            \
 572       SymX(stg_sel_0_upd_info)                  \
 573       SymX(stg_sel_10_upd_info)                 \
 574       SymX(stg_sel_11_upd_info)                 \
 575       SymX(stg_sel_12_upd_info)                 \
 576       SymX(stg_sel_13_upd_info)                 \
 577       SymX(stg_sel_14_upd_info)                 \
 578       SymX(stg_sel_15_upd_info)                 \
 579       SymX(stg_sel_1_upd_info)                  \
 580       SymX(stg_sel_2_upd_info)                  \
 581       SymX(stg_sel_3_upd_info)                  \
 582       SymX(stg_sel_4_upd_info)                  \
 583       SymX(stg_sel_5_upd_info)                  \
 584       SymX(stg_sel_6_upd_info)                  \
 585       SymX(stg_sel_7_upd_info)                  \
 586       SymX(stg_sel_8_upd_info)                  \
 587       SymX(stg_sel_9_upd_info)                  \
 588       SymX(stg_upd_frame_info)                  \
 589       SymX(suspendThread)                       \
 590       SymX(takeMVarzh_fast)                     \
 591       SymX(timesIntegerzh_fast)                 \
 592       SymX(tryPutMVarzh_fast)                   \
 593       SymX(tryTakeMVarzh_fast)                  \
 594       SymX(unblockAsyncExceptionszh_fast)       \
 595       SymX(unloadObj)                           \
 596       SymX(unsafeThawArrayzh_fast)              \
 597       SymX(waitReadzh_fast)                     \
 598       SymX(waitWritezh_fast)                    \
 599       SymX(word2Integerzh_fast)                 \
 600       SymX(writeTVarzh_fast)                    \
 601       SymX(xorIntegerzh_fast)                   \
 602       SymX(yieldzh_fast)                        \
 603       RTS_USER_SIGNALS_SYMBOLS
 604
 605 #ifdef SUPPORT_LONG_LONGS
 606 #define RTS_LONG_LONG_SYMS                      \
 607       SymX(int64ToIntegerzh_fast)               \
 608       SymX(word64ToIntegerzh_fast)
 609 #else
 610 #define RTS_LONG_LONG_SYMS /* nothing */
 611 #endif
 612
 613 // 64-bit support functions in libgcc.a
 614 #if defined(__GNUC__) && SIZEOF_VOID_P <= 4
 615 #define RTS_LIBGCC_SYMBOLS                      \
 616       Sym(__divdi3)                             \
 617       Sym(__udivdi3)                            \
 618       Sym(__moddi3)                             \
 619       Sym(__umoddi3)                            \
 620       Sym(__muldi3)                             \
 621       Sym(__ashldi3)                            \
 622       Sym(__ashrdi3)                            \
 623       Sym(__lshrdi3)                            \
 624       Sym(__eprintf)
 625 #elif defined(ia64_HOST_ARCH)
 626 #define RTS_LIBGCC_SYMBOLS                      \
 627       Sym(__divdi3)                             \
 628       Sym(__udivdi3)                            \
 629       Sym(__moddi3)                             \
 630       Sym(__umoddi3)                            \
 631       Sym(__divsf3)                             \
 632       Sym(__divdf3)
 633 #else
 634 #define RTS_LIBGCC_SYMBOLS
 635 #endif
 636
 637 #ifdef darwin_HOST_OS
 638       // Symbols that don't have a leading underscore
 639       // on Mac OS X. They have to receive special treatment,
 640       // see machoInitSymbolsWithoutUnderscore()
 641 #define RTS_MACHO_NOUNDERLINE_SYMBOLS           \
 642       Sym(saveFP)                               \
 643       Sym(restFP)
 644 #endif
 645
 646 /* entirely bogus claims about types of these symbols */
 647 #define Sym(vvv)  extern void vvv(void);
 648 #define SymX(vvv) /**/
 649 #define SymX_redirect(vvv,xxx) /**/
 650 RTS_SYMBOLS
 651 RTS_RET_SYMBOLS
 652 RTS_LONG_LONG_SYMS
 653 RTS_POSIX_ONLY_SYMBOLS
 654 RTS_MINGW_ONLY_SYMBOLS
 655 RTS_CYGWIN_ONLY_SYMBOLS
 656 RTS_LIBGCC_SYMBOLS
 657 #undef Sym
 658 #undef SymX
 659 #undef SymX_redirect
 660
 661 #ifdef LEADING_UNDERSCORE
 662 #define MAYBE_LEADING_UNDERSCORE_STR(s) ("_" s)
 663 #else
 664 #define MAYBE_LEADING_UNDERSCORE_STR(s) (s)
 665 #endif
 666
 667 #define Sym(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
 668                     (void*)(&(vvv)) },
 669 #define SymX(vvv) Sym(vvv)
 670
 671 // SymX_redirect allows us to redirect references to one symbol to
 672 // another symbol.  See newCAF/newDynCAF for an example.
 673 #define SymX_redirect(vvv,xxx) \
 674     { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
 675       (void*)(&(xxx)) },
 676
 677 static RtsSymbolVal rtsSyms[] = {
 678       RTS_SYMBOLS
 679       RTS_RET_SYMBOLS
 680       RTS_LONG_LONG_SYMS
 681       RTS_POSIX_ONLY_SYMBOLS
 682       RTS_MINGW_ONLY_SYMBOLS
 683       RTS_CYGWIN_ONLY_SYMBOLS
 684       RTS_LIBGCC_SYMBOLS
 685       { 0, 0 } /* sentinel */
 686 };
 687
 688 /* -----------------------------------------------------------------------------
 689  * Insert symbols into hash tables, checking for duplicates.
 690  */
 691 static void ghciInsertStrHashTable ( char* obj_name,
 692                                      HashTable *table,
 693                                      char* key,
 694                                      void *data
 695                                    )
 696 {
 697    if (lookupHashTable(table, (StgWord)key) == NULL)
 698    {
 699       insertStrHashTable(table, (StgWord)key, data);
 700       return;
 701    }
 702    debugBelch(
 703       "\n\n"
 704       "GHCi runtime linker: fatal error: I found a duplicate definition for symbol\n"
 705       "   %s\n"
 706       "whilst processing object file\n"
 707       "   %s\n"
 708       "This could be caused by:\n"
 709       "   * Loading two different object files which export the same symbol\n"
 710       "   * Specifying the same object file twice on the GHCi command line\n"
 711       "   * An incorrect `package.conf' entry, causing some object to be\n"
 712       "     loaded twice.\n"
 713       "GHCi cannot safely continue in this situation.  Exiting now.  Sorry.\n"
 714       "\n",
 715       (char*)key,
 716       obj_name
 717    );
 718    exit(1);
 719 }
 720
 721
 722 /* -----------------------------------------------------------------------------
 723  * initialize the object linker
 724  */
 725
 726
 727 static int linker_init_done = 0 ;
 728
 729 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
 730 static void *dl_prog_handle;
 731 #endif
 732
 733 /* dlopen(NULL,..) doesn't work so we grab libc explicitly */
 734 #if defined(openbsd_HOST_OS)
 735 static void *dl_libc_handle;
 736 #endif
 737
 738 void
 739 initLinker( void )
 740 {
 741     RtsSymbolVal *sym;
 742
 743     /* Make initLinker idempotent, so we can call it
 744        before evey relevant operation; that means we
 745        don't need to initialise the linker separately */
 746     if (linker_init_done == 1) { return; } else {
 747       linker_init_done = 1;
 748     }
 749
 750     symhash = allocStrHashTable();
 751
 752     /* populate the symbol table with stuff from the RTS */
 753     for (sym = rtsSyms; sym->lbl != NULL; sym++) {
 754         ghciInsertStrHashTable("(GHCi built-in symbols)",
 755                                symhash, sym->lbl, sym->addr);
 756     }
 757 #   if defined(OBJFORMAT_MACHO)
 758     machoInitSymbolsWithoutUnderscore();
 759 #   endif
 760
 761 #   if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
 762 #   if defined(RTLD_DEFAULT)
 763     dl_prog_handle = RTLD_DEFAULT;
 764 #   else
 765     dl_prog_handle = dlopen(NULL, RTLD_LAZY);
 766 #   if defined(openbsd_HOST_OS)
 767     dl_libc_handle = dlopen("libc.so", RTLD_LAZY);
 768 #   endif
 769 #   endif /* RTLD_DEFAULT */
 770 #   endif
 771 }
 772
 773 /* -----------------------------------------------------------------------------
 774  *                  Loading DLL or .so dynamic libraries
 775  * -----------------------------------------------------------------------------
 776  *
 777  * Add a DLL from which symbols may be found.  In the ELF case, just
 778  * do RTLD_GLOBAL-style add, so no further messing around needs to
 779  * happen in order that symbols in the loaded .so are findable --
 780  * lookupSymbol() will subsequently see them by dlsym on the program's
 781  * dl-handle.  Returns NULL if success, otherwise ptr to an err msg.
 782  *
 783  * In the PEi386 case, open the DLLs and put handles to them in a
 784  * linked list.  When looking for a symbol, try all handles in the
 785  * list.  This means that we need to load even DLLs that are guaranteed
 786  * to be in the ghc.exe image already, just so we can get a handle
 787  * to give to loadSymbol, so that we can find the symbols.  For such
 788  * libraries, the LoadLibrary call should be a no-op except for returning
 789  * the handle.
 790  *
 791  */
 792
 793 #if defined(OBJFORMAT_PEi386)
 794 /* A record for storing handles into DLLs. */
 795
 796 typedef
 797    struct _OpenedDLL {
 798       char*              name;
 799       struct _OpenedDLL* next;
 800       HINSTANCE instance;
 801    }
 802    OpenedDLL;
 803
 804 /* A list thereof. */
 805 static OpenedDLL* opened_dlls = NULL;
 806 #endif
 807
 808 char *
 809 addDLL( char *dll_name )
 810 {
 811 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
 812    /* ------------------- ELF DLL loader ------------------- */
 813    void *hdl;
 814    char *errmsg;
 815
 816    initLinker();
 817
 818    hdl= dlopen(dll_name, RTLD_NOW | RTLD_GLOBAL);
 819
 820    if (hdl == NULL) {
 821       /* dlopen failed; return a ptr to the error msg. */
 822       errmsg = dlerror();
 823       if (errmsg == NULL) errmsg = "addDLL: unknown error";
 824       return errmsg;
 825    } else {
 826       return NULL;
 827    }
 828    /*NOTREACHED*/
 829
 830 #  elif defined(OBJFORMAT_PEi386)
 831    /* ------------------- Win32 DLL loader ------------------- */
 832
 833    char*      buf;
 834    OpenedDLL* o_dll;
 835    HINSTANCE  instance;
 836
 837    initLinker();
 838
 839    /* debugBelch("\naddDLL; dll_name = `%s'\n", dll_name); */
 840
 841    /* See if we've already got it, and ignore if so. */
 842    for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
 843       if (0 == strcmp(o_dll->name, dll_name))
 844          return NULL;
 845    }
 846
 847    /* The file name has no suffix (yet) so that we can try
 848       both foo.dll and foo.drv
 849
 850       The documentation for LoadLibrary says:
 851         If no file name extension is specified in the lpFileName
 852         parameter, the default library extension .dll is
 853         appended. However, the file name string can include a trailing
 854         point character (.) to indicate that the module name has no
 855         extension. */
 856
 857    buf = stgMallocBytes(strlen(dll_name) + 10, "addDLL");
 858    sprintf(buf, "%s.DLL", dll_name);
 859    instance = LoadLibrary(buf);
 860    if (instance == NULL) {
 861          sprintf(buf, "%s.DRV", dll_name);      // KAA: allow loading of drivers (like winspool.drv)
 862          instance = LoadLibrary(buf);
 863          if (instance == NULL) {
 864                 stgFree(buf);
 865
 866             /* LoadLibrary failed; return a ptr to the error msg. */
 867             return "addDLL: unknown error";
 868          }
 869    }
 870    stgFree(buf);
 871
 872    /* Add this DLL to the list of DLLs in which to search for symbols. */
 873    o_dll = stgMallocBytes( sizeof(OpenedDLL), "addDLL" );
 874    o_dll->name     = stgMallocBytes(1+strlen(dll_name), "addDLL");
 875    strcpy(o_dll->name, dll_name);
 876    o_dll->instance = instance;
 877    o_dll->next     = opened_dlls;
 878    opened_dlls     = o_dll;
 879
 880    return NULL;
 881 #  else
 882    barf("addDLL: not implemented on this platform");
 883 #  endif
 884 }
 885
 886 /* -----------------------------------------------------------------------------
 887  * lookup a symbol in the hash table
 888  */
 889 void *
 890 lookupSymbol( char *lbl )
 891 {
 892     void *val;
 893     initLinker() ;
 894     ASSERT(symhash != NULL);
 895     val = lookupStrHashTable(symhash, lbl);
 896
 897     if (val == NULL) {
 898 #       if defined(OBJFORMAT_ELF)
 899 #       if defined(openbsd_HOST_OS)
 900         val = dlsym(dl_prog_handle, lbl);
 901         return (val != NULL) ? val : dlsym(dl_libc_handle,lbl);
 902 #       else /* not openbsd */
 903         return dlsym(dl_prog_handle, lbl);
 904 #       endif
 905 #       elif defined(OBJFORMAT_MACHO)
 906         if(NSIsSymbolNameDefined(lbl)) {
 907             NSSymbol symbol = NSLookupAndBindSymbol(lbl);
 908             return NSAddressOfSymbol(symbol);
 909         } else {
 910             return NULL;
 911         }
 912 #       elif defined(OBJFORMAT_PEi386)
 913         OpenedDLL* o_dll;
 914         void* sym;
 915         for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
 916           /* debugBelch("look in %s for %s\n", o_dll->name, lbl); */
 917            if (lbl[0] == '_') {
 918               /* HACK: if the name has an initial underscore, try stripping
 919                  it off & look that up first. I've yet to verify whether there's
 920                  a Rule that governs whether an initial '_' *should always* be
 921                  stripped off when mapping from import lib name to the DLL name.
 922               */
 923               sym = GetProcAddress(o_dll->instance, (lbl+1));
 924               if (sym != NULL) {
 925                 /*debugBelch("found %s in %s\n", lbl+1,o_dll->name);*/
 926                 return sym;
 927               }
 928            }
 929            sym = GetProcAddress(o_dll->instance, lbl);
 930            if (sym != NULL) {
 931              /*debugBelch("found %s in %s\n", lbl,o_dll->name);*/
 932              return sym;
 933            }
 934         }
 935         return NULL;
 936 #       else
 937         ASSERT(2+2 == 5);
 938         return NULL;
 939 #       endif
 940     } else {
 941         return val;
 942     }
 943 }
 944
 945 static
 946 __attribute((unused))
 947 void *
 948 lookupLocalSymbol( ObjectCode* oc, char *lbl )
 949 {
 950     void *val;
 951     initLinker() ;
 952     val = lookupStrHashTable(oc->lochash, lbl);
 953
 954     if (val == NULL) {
 955         return NULL;
 956     } else {
 957         return val;
 958     }
 959 }
 960
 961
 962 /* -----------------------------------------------------------------------------
 963  * Debugging aid: look in GHCi's object symbol tables for symbols
 964  * within DELTA bytes of the specified address, and show their names.
 965  */
 966 #ifdef DEBUG
 967 void ghci_enquire ( char* addr );
 968
 969 void ghci_enquire ( char* addr )
 970 {
 971    int   i;
 972    char* sym;
 973    char* a;
 974    const int DELTA = 64;
 975    ObjectCode* oc;
 976
 977    initLinker();
 978
 979    for (oc = objects; oc; oc = oc->next) {
 980       for (i = 0; i < oc->n_symbols; i++) {
 981          sym = oc->symbols[i];
 982          if (sym == NULL) continue;
 983          // debugBelch("enquire %p %p\n", sym, oc->lochash);
 984          a = NULL;
 985          if (oc->lochash != NULL) {
 986             a = lookupStrHashTable(oc->lochash, sym);
 987          }
 988          if (a == NULL) {
 989             a = lookupStrHashTable(symhash, sym);
 990          }
 991          if (a == NULL) {
 992              // debugBelch("ghci_enquire: can't find %s\n", sym);
 993          }
 994          else if (addr-DELTA <= a && a <= addr+DELTA) {
 995             debugBelch("%p + %3d  ==  `%s'\n", addr, a - addr, sym);
 996          }
 997       }
 998    }
 999 }
1000 #endif
1001
1002 #ifdef ia64_HOST_ARCH
1003 static unsigned int PLTSize(void);
1004 #endif
1005
1006 /* -----------------------------------------------------------------------------
1007  * Load an obj (populate the global symbol table, but don't resolve yet)
1008  *
1009  * Returns: 1 if ok, 0 on error.
1010  */
1011 HsInt
1012 loadObj( char *path )
1013 {
1014    ObjectCode* oc;
1015    struct stat st;
1016    int r, n;
1017 #ifdef USE_MMAP
1018    int fd, pagesize;
1019    void *map_addr = NULL;
1020 #else
1021    FILE *f;
1022 #endif
1023
1024    initLinker();
1025
1026    /* debugBelch("loadObj %s\n", path ); */
1027
1028    /* Check that we haven't already loaded this object.
1029       Ignore requests to load multiple times */
1030    {
1031        ObjectCode *o;
1032        int is_dup = 0;
1033        for (o = objects; o; o = o->next) {
1034           if (0 == strcmp(o->fileName, path)) {
1035              is_dup = 1;
1036              break; /* don't need to search further */
1037           }
1038        }
1039        if (is_dup) {
1040           IF_DEBUG(linker, debugBelch(
1041             "GHCi runtime linker: warning: looks like you're trying to load the\n"
1042             "same object file twice:\n"
1043             "   %s\n"
1044             "GHCi will ignore this, but be warned.\n"
1045             , path));
1046           return 1; /* success */
1047        }
1048    }
1049
1050    oc = stgMallocBytes(sizeof(ObjectCode), "loadObj(oc)");
1051
1052 #  if defined(OBJFORMAT_ELF)
1053    oc->formatName = "ELF";
1054 #  elif defined(OBJFORMAT_PEi386)
1055    oc->formatName = "PEi386";
1056 #  elif defined(OBJFORMAT_MACHO)
1057    oc->formatName = "Mach-O";
1058 #  else
1059    stgFree(oc);
1060    barf("loadObj: not implemented on this platform");
1061 #  endif
1062
1063    r = stat(path, &st);
1064    if (r == -1) { return 0; }
1065
1066    /* sigh, strdup() isn't a POSIX function, so do it the long way */
1067    oc->fileName = stgMallocBytes( strlen(path)+1, "loadObj" );
1068    strcpy(oc->fileName, path);
1069
1070    oc->fileSize          = st.st_size;
1071    oc->symbols           = NULL;
1072    oc->sections          = NULL;
1073    oc->lochash           = allocStrHashTable();
1074    oc->proddables        = NULL;
1075
1076    /* chain it onto the list of objects */
1077    oc->next              = objects;
1078    objects               = oc;
1079
1080 #ifdef USE_MMAP
1081 #define ROUND_UP(x,size) ((x + size - 1) & ~(size - 1))
1082
1083    /* On many architectures malloc'd memory isn't executable, so we need to use mmap. */
1084
1085 #if defined(openbsd_HOST_OS)
1086    fd = open(path, O_RDONLY, S_IRUSR);
1087 #else
1088    fd = open(path, O_RDONLY);
1089 #endif
1090    if (fd == -1)
1091       barf("loadObj: can't open `%s'", path);
1092
1093    pagesize = getpagesize();
1094
1095 #ifdef ia64_HOST_ARCH
1096    /* The PLT needs to be right before the object */
1097    n = ROUND_UP(PLTSize(), pagesize);
1098    oc->plt = mmap(NULL, n, PROT_EXEC|PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
1099    if (oc->plt == MAP_FAILED)
1100       barf("loadObj: can't allocate PLT");
1101
1102    oc->pltIndex = 0;
1103    map_addr = oc->plt + n;
1104 #endif
1105
1106    n = ROUND_UP(oc->fileSize, pagesize);
1107
1108    /* Link objects into the lower 2Gb on x86_64.  GHC assumes the
1109     * small memory model on this architecture (see gcc docs,
1110     * -mcmodel=small).
1111     */
1112 #ifdef x86_64_HOST_ARCH
1113 #define EXTRA_MAP_FLAGS MAP_32BIT
1114 #else
1115 #define EXTRA_MAP_FLAGS 0
1116 #endif
1117
1118    oc->image = mmap(map_addr, n, PROT_EXEC|PROT_READ|PROT_WRITE,
1119                     MAP_PRIVATE|EXTRA_MAP_FLAGS, fd, 0);
1120    if (oc->image == MAP_FAILED)
1121       barf("loadObj: can't map `%s'", path);
1122
1123    close(fd);
1124
1125 #else /* !USE_MMAP */
1126
1127    oc->image = stgMallocBytes(oc->fileSize, "loadObj(image)");
1128
1129    /* load the image into memory */
1130    f = fopen(path, "rb");
1131    if (!f)
1132        barf("loadObj: can't read `%s'", path);
1133
1134    n = fread ( oc->image, 1, oc->fileSize, f );
1135    if (n != oc->fileSize)
1136       barf("loadObj: error whilst reading `%s'", path);
1137
1138    fclose(f);
1139
1140 #endif /* USE_MMAP */
1141
1142 #  if defined(OBJFORMAT_MACHO)
1143    r = ocAllocateJumpIslands_MachO ( oc );
1144    if (!r) { return r; }
1145 #  elif defined(OBJFORMAT_ELF) && defined(powerpc_HOST_ARCH)
1146    r = ocAllocateJumpIslands_ELF ( oc );
1147    if (!r) { return r; }
1148 #endif
1149
1150    /* verify the in-memory image */
1151 #  if defined(OBJFORMAT_ELF)
1152    r = ocVerifyImage_ELF ( oc );
1153 #  elif defined(OBJFORMAT_PEi386)
1154    r = ocVerifyImage_PEi386 ( oc );
1155 #  elif defined(OBJFORMAT_MACHO)
1156    r = ocVerifyImage_MachO ( oc );
1157 #  else
1158    barf("loadObj: no verify method");
1159 #  endif
1160    if (!r) { return r; }
1161
1162    /* build the symbol list for this image */
1163 #  if defined(OBJFORMAT_ELF)
1164    r = ocGetNames_ELF ( oc );
1165 #  elif defined(OBJFORMAT_PEi386)
1166    r = ocGetNames_PEi386 ( oc );
1167 #  elif defined(OBJFORMAT_MACHO)
1168    r = ocGetNames_MachO ( oc );
1169 #  else
1170    barf("loadObj: no getNames method");
1171 #  endif
1172    if (!r) { return r; }
1173
1174    /* loaded, but not resolved yet */
1175    oc->status = OBJECT_LOADED;
1176
1177    return 1;
1178 }
1179
1180 /* -----------------------------------------------------------------------------
1181  * resolve all the currently unlinked objects in memory
1182  *
1183  * Returns: 1 if ok, 0 on error.
1184  */
1185 HsInt
1186 resolveObjs( void )
1187 {
1188     ObjectCode *oc;
1189     int r;
1190
1191     initLinker();
1192
1193     for (oc = objects; oc; oc = oc->next) {
1194         if (oc->status != OBJECT_RESOLVED) {
1195 #           if defined(OBJFORMAT_ELF)
1196             r = ocResolve_ELF ( oc );
1197 #           elif defined(OBJFORMAT_PEi386)
1198             r = ocResolve_PEi386 ( oc );
1199 #           elif defined(OBJFORMAT_MACHO)
1200             r = ocResolve_MachO ( oc );
1201 #           else
1202             barf("resolveObjs: not implemented on this platform");
1203 #           endif
1204             if (!r) { return r; }
1205             oc->status = OBJECT_RESOLVED;
1206         }
1207     }
1208     return 1;
1209 }
1210
1211 /* -----------------------------------------------------------------------------
1212  * delete an object from the pool
1213  */
1214 HsInt
1215 unloadObj( char *path )
1216 {
1217     ObjectCode *oc, *prev;
1218
1219     ASSERT(symhash != NULL);
1220     ASSERT(objects != NULL);
1221
1222     initLinker();
1223
1224     prev = NULL;
1225     for (oc = objects; oc; prev = oc, oc = oc->next) {
1226         if (!strcmp(oc->fileName,path)) {
1227
1228             /* Remove all the mappings for the symbols within this
1229              * object..
1230              */
1231             {
1232                 int i;
1233                 for (i = 0; i < oc->n_symbols; i++) {
1234                    if (oc->symbols[i] != NULL) {
1235                        removeStrHashTable(symhash, oc->symbols[i], NULL);
1236                    }
1237                 }
1238             }
1239
1240             if (prev == NULL) {
1241                 objects = oc->next;
1242             } else {
1243                 prev->next = oc->next;
1244             }
1245
1246             /* We're going to leave this in place, in case there are
1247                any pointers from the heap into it: */
1248             /* stgFree(oc->image); */
1249             stgFree(oc->fileName);
1250             stgFree(oc->symbols);
1251             stgFree(oc->sections);
1252             /* The local hash table should have been freed at the end
1253                of the ocResolve_ call on it. */
1254             ASSERT(oc->lochash == NULL);
1255             stgFree(oc);
1256             return 1;
1257         }
1258     }
1259
1260     errorBelch("unloadObj: can't find `%s' to unload", path);
1261     return 0;
1262 }
1263
1264 /* -----------------------------------------------------------------------------
1265  * Sanity checking.  For each ObjectCode, maintain a list of address ranges
1266  * which may be prodded during relocation, and abort if we try and write
1267  * outside any of these.
1268  */
1269 static void addProddableBlock ( ObjectCode* oc, void* start, int size )
1270 {
1271    ProddableBlock* pb
1272       = stgMallocBytes(sizeof(ProddableBlock), "addProddableBlock");
1273    /* debugBelch("aPB %p %p %d\n", oc, start, size); */
1274    ASSERT(size > 0);
1275    pb->start      = start;
1276    pb->size       = size;
1277    pb->next       = oc->proddables;
1278    oc->proddables = pb;
1279 }
1280
1281 static void checkProddableBlock ( ObjectCode* oc, void* addr )
1282 {
1283    ProddableBlock* pb;
1284    for (pb = oc->proddables; pb != NULL; pb = pb->next) {
1285       char* s = (char*)(pb->start);
1286       char* e = s + pb->size - 1;
1287       char* a = (char*)addr;
1288       /* Assumes that the biggest fixup involves a 4-byte write.  This
1289          probably needs to be changed to 8 (ie, +7) on 64-bit
1290          plats. */
1291       if (a >= s && (a+3) <= e) return;
1292    }
1293    barf("checkProddableBlock: invalid fixup in runtime linker");
1294 }
1295
1296 /* -----------------------------------------------------------------------------
1297  * Section management.
1298  */
1299 static void addSection ( ObjectCode* oc, SectionKind kind,
1300                          void* start, void* end )
1301 {
1302    Section* s   = stgMallocBytes(sizeof(Section), "addSection");
1303    s->start     = start;
1304    s->end       = end;
1305    s->kind      = kind;
1306    s->next      = oc->sections;
1307    oc->sections = s;
1308    /*
1309    debugBelch("addSection: %p-%p (size %d), kind %d\n",
1310                    start, ((char*)end)-1, end - start + 1, kind );
1311    */
1312 }
1313
1314
1315 /* --------------------------------------------------------------------------
1316  * PowerPC specifics (jump islands)
1317  * ------------------------------------------------------------------------*/
1318
1319 #if defined(powerpc_HOST_ARCH)
1320
1321 /*
1322   ocAllocateJumpIslands
1323
1324   Allocate additional space at the end of the object file image to make room
1325   for jump islands.
1326
1327   PowerPC relative branch instructions have a 24 bit displacement field.
1328   As PPC code is always 4-byte-aligned, this yields a +-32MB range.
1329   If a particular imported symbol is outside this range, we have to redirect
1330   the jump to a short piece of new code that just loads the 32bit absolute
1331   address and jumps there.
1332   This function just allocates space for one 16 byte ppcJumpIsland for every
1333   undefined symbol in the object file. The code for the islands is filled in by
1334   makeJumpIsland below.
1335 */
1336
1337 static int ocAllocateJumpIslands( ObjectCode* oc, int count, int first )
1338 {
1339 #ifdef USE_MMAP
1340   int pagesize, n, m;
1341 #endif
1342   int aligned;
1343
1344   if( count > 0 )
1345   {
1346     // round up to the nearest 4
1347     aligned = (oc->fileSize + 3) & ~3;
1348
1349 #ifdef USE_MMAP
1350     #ifndef linux_HOST_OS /* mremap is a linux extension */
1351         #error ocAllocateJumpIslands doesnt want USE_MMAP to be defined
1352     #endif
1353
1354     pagesize = getpagesize();
1355     n = ROUND_UP( oc->fileSize, pagesize );
1356     m = ROUND_UP( aligned + sizeof (ppcJumpIsland) * count, pagesize );
1357
1358     /* The effect of this mremap() call is only the ensure that we have
1359      * a sufficient number of virtually contiguous pages.  As returned from
1360      * mremap, the pages past the end of the file are not backed.  We give
1361      * them a backing by using MAP_FIXED to map in anonymous pages.
1362      */
1363     if( (oc->image = mremap( oc->image, n, m, MREMAP_MAYMOVE )) == MAP_FAILED )
1364     {
1365       errorBelch( "Unable to mremap for Jump Islands\n" );
1366       return 0;
1367     }
1368
1369     if( mmap( oc->image + n, m - n, PROT_READ | PROT_WRITE | PROT_EXEC,
1370               MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED, 0, 0 ) == MAP_FAILED )
1371     {
1372       errorBelch( "Unable to mmap( MAP_FIXED ) for Jump Islands\n" );
1373       return 0;
1374     }
1375
1376 #else
1377     oc->image = stgReallocBytes( oc->image,
1378                                  aligned + sizeof (ppcJumpIsland) * count,
1379                                  "ocAllocateJumpIslands" );
1380 #endif /* USE_MMAP */
1381
1382     oc->jump_islands = (ppcJumpIsland *) (oc->image + aligned);
1383     memset( oc->jump_islands, 0, sizeof (ppcJumpIsland) * count );
1384   }
1385   else
1386     oc->jump_islands = NULL;
1387
1388   oc->island_start_symbol = first;
1389   oc->n_islands = count;
1390
1391   return 1;
1392 }
1393
1394 static unsigned long makeJumpIsland( ObjectCode* oc,
1395                                      unsigned long symbolNumber,
1396                                      unsigned long target )
1397 {
1398   ppcJumpIsland *island;
1399
1400   if( symbolNumber < oc->island_start_symbol ||
1401       symbolNumber - oc->island_start_symbol > oc->n_islands)
1402     return 0;
1403
1404   island = &oc->jump_islands[symbolNumber - oc->island_start_symbol];
1405
1406   // lis r12, hi16(target)
1407   island->lis_r12     = 0x3d80;
1408   island->hi_addr     = target >> 16;
1409
1410   // ori r12, r12, lo16(target)
1411   island->ori_r12_r12 = 0x618c;
1412   island->lo_addr     = target & 0xffff;
1413
1414   // mtctr r12
1415   island->mtctr_r12   = 0x7d8903a6;
1416
1417   // bctr
1418   island->bctr        = 0x4e800420;
1419
1420   return (unsigned long) island;
1421 }
1422
1423 /*
1424    ocFlushInstructionCache
1425
1426    Flush the data & instruction caches.
1427    Because the PPC has split data/instruction caches, we have to
1428    do that whenever we modify code at runtime.
1429  */
1430
1431 static void ocFlushInstructionCache( ObjectCode *oc )
1432 {
1433     int n = (oc->fileSize + sizeof( ppcJumpIsland ) * oc->n_islands + 3) / 4;
1434     unsigned long *p = (unsigned long *) oc->image;
1435
1436     while( n-- )
1437     {
1438         __asm__ volatile ( "dcbf 0,%0\n\t"
1439                            "sync\n\t"
1440                            "icbi 0,%0"
1441                            :
1442                            : "r" (p)
1443                          );
1444         p++;
1445     }
1446     __asm__ volatile ( "sync\n\t"
1447                        "isync"
1448                      );
1449 }
1450 #endif
1451
1452 /* --------------------------------------------------------------------------
1453  * PEi386 specifics (Win32 targets)
1454  * ------------------------------------------------------------------------*/
1455
1456 /* The information for this linker comes from
1457       Microsoft Portable Executable
1458       and Common Object File Format Specification
1459       revision 5.1 January 1998
1460    which SimonM says comes from the MS Developer Network CDs.
1461
1462    It can be found there (on older CDs), but can also be found
1463    online at:
1464
1465       http://www.microsoft.com/hwdev/hardware/PECOFF.asp
1466
1467    (this is Rev 6.0 from February 1999).
1468
1469    Things move, so if that fails, try searching for it via
1470
1471       http://www.google.com/search?q=PE+COFF+specification
1472
1473    The ultimate reference for the PE format is the Winnt.h
1474    header file that comes with the Platform SDKs; as always,
1475    implementations will drift wrt their documentation.
1476
1477    A good background article on the PE format is Matt Pietrek's
1478    March 1994 article in Microsoft System Journal (MSJ)
1479    (Vol.9, No. 3): "Peering Inside the PE: A Tour of the
1480    Win32 Portable Executable File Format." The info in there
1481    has recently been updated in a two part article in
1482    MSDN magazine, issues Feb and March 2002,
1483    "Inside Windows: An In-Depth Look into the Win32 Portable
1484    Executable File Format"
1485
1486    John Levine's book "Linkers and Loaders" contains useful
1487    info on PE too.
1488 */
1489
1490
1491 #if defined(OBJFORMAT_PEi386)
1492
1493
1494
1495 typedef unsigned char  UChar;
1496 typedef unsigned short UInt16;
1497 typedef unsigned int   UInt32;
1498 typedef          int   Int32;
1499
1500
1501 typedef
1502    struct {
1503       UInt16 Machine;
1504       UInt16 NumberOfSections;
1505       UInt32 TimeDateStamp;
1506       UInt32 PointerToSymbolTable;
1507       UInt32 NumberOfSymbols;
1508       UInt16 SizeOfOptionalHeader;
1509       UInt16 Characteristics;
1510    }
1511    COFF_header;
1512
1513 #define sizeof_COFF_header 20
1514
1515
1516 typedef
1517    struct {
1518       UChar  Name[8];
1519       UInt32 VirtualSize;
1520       UInt32 VirtualAddress;
1521       UInt32 SizeOfRawData;
1522       UInt32 PointerToRawData;
1523       UInt32 PointerToRelocations;
1524       UInt32 PointerToLinenumbers;
1525       UInt16 NumberOfRelocations;
1526       UInt16 NumberOfLineNumbers;
1527       UInt32 Characteristics;
1528    }
1529    COFF_section;
1530
1531 #define sizeof_COFF_section 40
1532
1533
1534 typedef
1535    struct {
1536       UChar  Name[8];
1537       UInt32 Value;
1538       UInt16 SectionNumber;
1539       UInt16 Type;
1540       UChar  StorageClass;
1541       UChar  NumberOfAuxSymbols;
1542    }
1543    COFF_symbol;
1544
1545 #define sizeof_COFF_symbol 18
1546
1547
1548 typedef
1549    struct {
1550       UInt32 VirtualAddress;
1551       UInt32 SymbolTableIndex;
1552       UInt16 Type;
1553    }
1554    COFF_reloc;
1555
1556 #define sizeof_COFF_reloc 10
1557
1558
1559 /* From PE spec doc, section 3.3.2 */
1560 /* Note use of MYIMAGE_* since IMAGE_* are already defined in
1561    windows.h -- for the same purpose, but I want to know what I'm
1562    getting, here. */
1563 #define MYIMAGE_FILE_RELOCS_STRIPPED     0x0001
1564 #define MYIMAGE_FILE_EXECUTABLE_IMAGE    0x0002
1565 #define MYIMAGE_FILE_DLL                 0x2000
1566 #define MYIMAGE_FILE_SYSTEM              0x1000
1567 #define MYIMAGE_FILE_BYTES_REVERSED_HI   0x8000
1568 #define MYIMAGE_FILE_BYTES_REVERSED_LO   0x0080
1569 #define MYIMAGE_FILE_32BIT_MACHINE       0x0100
1570
1571 /* From PE spec doc, section 5.4.2 and 5.4.4 */
1572 #define MYIMAGE_SYM_CLASS_EXTERNAL       2
1573 #define MYIMAGE_SYM_CLASS_STATIC         3
1574 #define MYIMAGE_SYM_UNDEFINED            0
1575
1576 /* From PE spec doc, section 4.1 */
1577 #define MYIMAGE_SCN_CNT_CODE             0x00000020
1578 #define MYIMAGE_SCN_CNT_INITIALIZED_DATA 0x00000040
1579 #define MYIMAGE_SCN_LNK_NRELOC_OVFL      0x01000000
1580
1581 /* From PE spec doc, section 5.2.1 */
1582 #define MYIMAGE_REL_I386_DIR32           0x0006
1583 #define MYIMAGE_REL_I386_REL32           0x0014
1584
1585
1586 /* We use myindex to calculate array addresses, rather than
1587    simply doing the normal subscript thing.  That's because
1588    some of the above structs have sizes which are not
1589    a whole number of words.  GCC rounds their sizes up to a
1590    whole number of words, which means that the address calcs
1591    arising from using normal C indexing or pointer arithmetic
1592    are just plain wrong.  Sigh.
1593 */
1594 static UChar *
1595 myindex ( int scale, void* base, int index )
1596 {
1597    return
1598       ((UChar*)base) + scale * index;
1599 }
1600
1601
1602 static void
1603 printName ( UChar* name, UChar* strtab )
1604 {
1605    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
1606       UInt32 strtab_offset = * (UInt32*)(name+4);
1607       debugBelch("%s", strtab + strtab_offset );
1608    } else {
1609       int i;
1610       for (i = 0; i < 8; i++) {
1611          if (name[i] == 0) break;
1612          debugBelch("%c", name[i] );
1613       }
1614    }
1615 }
1616
1617
1618 static void
1619 copyName ( UChar* name, UChar* strtab, UChar* dst, int dstSize )
1620 {
1621    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
1622       UInt32 strtab_offset = * (UInt32*)(name+4);
1623       strncpy ( dst, strtab+strtab_offset, dstSize );
1624       dst[dstSize-1] = 0;
1625    } else {
1626       int i = 0;
1627       while (1) {
1628          if (i >= 8) break;
1629          if (name[i] == 0) break;
1630          dst[i] = name[i];
1631          i++;
1632       }
1633       dst[i] = 0;
1634    }
1635 }
1636
1637
1638 static UChar *
1639 cstring_from_COFF_symbol_name ( UChar* name, UChar* strtab )
1640 {
1641    UChar* newstr;
1642    /* If the string is longer than 8 bytes, look in the
1643       string table for it -- this will be correctly zero terminated.
1644    */
1645    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
1646       UInt32 strtab_offset = * (UInt32*)(name+4);
1647       return ((UChar*)strtab) + strtab_offset;
1648    }
1649    /* Otherwise, if shorter than 8 bytes, return the original,
1650       which by defn is correctly terminated.
1651    */
1652    if (name[7]==0) return name;
1653    /* The annoying case: 8 bytes.  Copy into a temporary
1654       (which is never freed ...)
1655    */
1656    newstr = stgMallocBytes(9, "cstring_from_COFF_symbol_name");
1657    ASSERT(newstr);
1658    strncpy(newstr,name,8);
1659    newstr[8] = 0;
1660    return newstr;
1661 }
1662
1663
1664 /* Just compares the short names (first 8 chars) */
1665 static COFF_section *
1666 findPEi386SectionCalled ( ObjectCode* oc,  char* name )
1667 {
1668    int i;
1669    COFF_header* hdr
1670       = (COFF_header*)(oc->image);
1671    COFF_section* sectab
1672       = (COFF_section*) (
1673            ((UChar*)(oc->image))
1674            + sizeof_COFF_header + hdr->SizeOfOptionalHeader
1675         );
1676    for (i = 0; i < hdr->NumberOfSections; i++) {
1677       UChar* n1;
1678       UChar* n2;
1679       COFF_section* section_i
1680          = (COFF_section*)
1681            myindex ( sizeof_COFF_section, sectab, i );
1682       n1 = (UChar*) &(section_i->Name);
1683       n2 = name;
1684       if (n1[0]==n2[0] && n1[1]==n2[1] && n1[2]==n2[2] &&
1685           n1[3]==n2[3] && n1[4]==n2[4] && n1[5]==n2[5] &&
1686           n1[6]==n2[6] && n1[7]==n2[7])
1687          return section_i;
1688    }
1689
1690    return NULL;
1691 }
1692
1693
1694 static void
1695 zapTrailingAtSign ( UChar* sym )
1696 {
1697 #  define my_isdigit(c) ((c) >= '0' && (c) <= '9')
1698    int i, j;
1699    if (sym[0] == 0) return;
1700    i = 0;
1701    while (sym[i] != 0) i++;
1702    i--;
1703    j = i;
1704    while (j > 0 && my_isdigit(sym[j])) j--;
1705    if (j > 0 && sym[j] == '@' && j != i) sym[j] = 0;
1706 #  undef my_isdigit
1707 }
1708
1709
1710 static int
1711 ocVerifyImage_PEi386 ( ObjectCode* oc )
1712 {
1713    int i;
1714    UInt32 j, noRelocs;
1715    COFF_header*  hdr;
1716    COFF_section* sectab;
1717    COFF_symbol*  symtab;
1718    UChar*        strtab;
1719    /* debugBelch("\nLOADING %s\n", oc->fileName); */
1720    hdr = (COFF_header*)(oc->image);
1721    sectab = (COFF_section*) (
1722                ((UChar*)(oc->image))
1723                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
1724             );
1725    symtab = (COFF_symbol*) (
1726                ((UChar*)(oc->image))
1727                + hdr->PointerToSymbolTable
1728             );
1729    strtab = ((UChar*)symtab)
1730             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
1731
1732    if (hdr->Machine != 0x14c) {
1733       errorBelch("Not x86 PEi386");
1734       return 0;
1735    }
1736    if (hdr->SizeOfOptionalHeader != 0) {
1737       errorBelch("PEi386 with nonempty optional header");
1738       return 0;
1739    }
1740    if ( /* (hdr->Characteristics & MYIMAGE_FILE_RELOCS_STRIPPED) || */
1741         (hdr->Characteristics & MYIMAGE_FILE_EXECUTABLE_IMAGE) ||
1742         (hdr->Characteristics & MYIMAGE_FILE_DLL) ||
1743         (hdr->Characteristics & MYIMAGE_FILE_SYSTEM) ) {
1744       errorBelch("Not a PEi386 object file");
1745       return 0;
1746    }
1747    if ( (hdr->Characteristics & MYIMAGE_FILE_BYTES_REVERSED_HI)
1748         /* || !(hdr->Characteristics & MYIMAGE_FILE_32BIT_MACHINE) */ ) {
1749       errorBelch("Invalid PEi386 word size or endiannness: %d",
1750             (int)(hdr->Characteristics));
1751       return 0;
1752    }
1753    /* If the string table size is way crazy, this might indicate that
1754       there are more than 64k relocations, despite claims to the
1755       contrary.  Hence this test. */
1756    /* debugBelch("strtab size %d\n", * (UInt32*)strtab); */
1757 #if 0
1758    if ( (*(UInt32*)strtab) > 600000 ) {
1759       /* Note that 600k has no special significance other than being
1760          big enough to handle the almost-2MB-sized lumps that
1761          constitute HSwin32*.o. */
1762       debugBelch("PEi386 object has suspiciously large string table; > 64k relocs?");
1763       return 0;
1764    }
1765 #endif
1766
1767    /* No further verification after this point; only debug printing. */
1768    i = 0;
1769    IF_DEBUG(linker, i=1);
1770    if (i == 0) return 1;
1771
1772    debugBelch( "sectab offset = %d\n", ((UChar*)sectab) - ((UChar*)hdr) );
1773    debugBelch( "symtab offset = %d\n", ((UChar*)symtab) - ((UChar*)hdr) );
1774    debugBelch( "strtab offset = %d\n", ((UChar*)strtab) - ((UChar*)hdr) );
1775
1776    debugBelch("\n" );
1777    debugBelch( "Machine:           0x%x\n", (UInt32)(hdr->Machine) );
1778    debugBelch( "# sections:        %d\n",   (UInt32)(hdr->NumberOfSections) );
1779    debugBelch( "time/date:         0x%x\n", (UInt32)(hdr->TimeDateStamp) );
1780    debugBelch( "symtab offset:     %d\n",   (UInt32)(hdr->PointerToSymbolTable) );
1781    debugBelch( "# symbols:         %d\n",   (UInt32)(hdr->NumberOfSymbols) );
1782    debugBelch( "sz of opt hdr:     %d\n",   (UInt32)(hdr->SizeOfOptionalHeader) );
1783    debugBelch( "characteristics:   0x%x\n", (UInt32)(hdr->Characteristics) );
1784
1785    /* Print the section table. */
1786    debugBelch("\n" );
1787    for (i = 0; i < hdr->NumberOfSections; i++) {
1788       COFF_reloc* reltab;
1789       COFF_section* sectab_i
1790          = (COFF_section*)
1791            myindex ( sizeof_COFF_section, sectab, i );
1792       debugBelch(
1793                 "\n"
1794                 "section %d\n"
1795                 "     name `",
1796                 i
1797               );
1798       printName ( sectab_i->Name, strtab );
1799       debugBelch(
1800                 "'\n"
1801                 "    vsize %d\n"
1802                 "    vaddr %d\n"
1803                 "  data sz %d\n"
1804                 " data off %d\n"
1805                 "  num rel %d\n"
1806                 "  off rel %d\n"
1807                 "  ptr raw 0x%x\n",
1808                 sectab_i->VirtualSize,
1809                 sectab_i->VirtualAddress,
1810                 sectab_i->SizeOfRawData,
1811                 sectab_i->PointerToRawData,
1812                 sectab_i->NumberOfRelocations,
1813                 sectab_i->PointerToRelocations,
1814                 sectab_i->PointerToRawData
1815               );
1816       reltab = (COFF_reloc*) (
1817                   ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
1818                );
1819
1820       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
1821         /* If the relocation field (a short) has overflowed, the
1822          * real count can be found in the first reloc entry.
1823          *
1824          * See Section 4.1 (last para) of the PE spec (rev6.0).
1825          */
1826         COFF_reloc* rel = (COFF_reloc*)
1827                            myindex ( sizeof_COFF_reloc, reltab, 0 );
1828         noRelocs = rel->VirtualAddress;
1829         j = 1;
1830       } else {
1831         noRelocs = sectab_i->NumberOfRelocations;
1832         j = 0;
1833       }
1834
1835       for (; j < noRelocs; j++) {
1836          COFF_symbol* sym;
1837          COFF_reloc* rel = (COFF_reloc*)
1838                            myindex ( sizeof_COFF_reloc, reltab, j );
1839          debugBelch(
1840                    "        type 0x%-4x   vaddr 0x%-8x   name `",
1841                    (UInt32)rel->Type,
1842                    rel->VirtualAddress );
1843          sym = (COFF_symbol*)
1844                myindex ( sizeof_COFF_symbol, symtab, rel->SymbolTableIndex );
1845          /* Hmm..mysterious looking offset - what's it for? SOF */
1846          printName ( sym->Name, strtab -10 );
1847          debugBelch("'\n" );
1848       }
1849
1850       debugBelch("\n" );
1851    }
1852    debugBelch("\n" );
1853    debugBelch("string table has size 0x%x\n", * (UInt32*)strtab );
1854    debugBelch("---START of string table---\n");
1855    for (i = 4; i < *(Int32*)strtab; i++) {
1856       if (strtab[i] == 0)
1857          debugBelch("\n"); else
1858          debugBelch("%c", strtab[i] );
1859    }
1860    debugBelch("--- END  of string table---\n");
1861
1862    debugBelch("\n" );
1863    i = 0;
1864    while (1) {
1865       COFF_symbol* symtab_i;
1866       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
1867       symtab_i = (COFF_symbol*)
1868                  myindex ( sizeof_COFF_symbol, symtab, i );
1869       debugBelch(
1870                 "symbol %d\n"
1871                 "     name `",
1872                 i
1873               );
1874       printName ( symtab_i->Name, strtab );
1875       debugBelch(
1876                 "'\n"
1877                 "    value 0x%x\n"
1878                 "   1+sec# %d\n"
1879                 "     type 0x%x\n"
1880                 "   sclass 0x%x\n"
1881                 "     nAux %d\n",
1882                 symtab_i->Value,
1883                 (Int32)(symtab_i->SectionNumber),
1884                 (UInt32)symtab_i->Type,
1885                 (UInt32)symtab_i->StorageClass,
1886                 (UInt32)symtab_i->NumberOfAuxSymbols
1887               );
1888       i += symtab_i->NumberOfAuxSymbols;
1889       i++;
1890    }
1891
1892    debugBelch("\n" );
1893    return 1;
1894 }
1895
1896
1897 static int
1898 ocGetNames_PEi386 ( ObjectCode* oc )
1899 {
1900    COFF_header*  hdr;
1901    COFF_section* sectab;
1902    COFF_symbol*  symtab;
1903    UChar*        strtab;
1904
1905    UChar* sname;
1906    void*  addr;
1907    int    i;
1908
1909    hdr = (COFF_header*)(oc->image);
1910    sectab = (COFF_section*) (
1911                ((UChar*)(oc->image))
1912                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
1913             );
1914    symtab = (COFF_symbol*) (
1915                ((UChar*)(oc->image))
1916                + hdr->PointerToSymbolTable
1917             );
1918    strtab = ((UChar*)(oc->image))
1919             + hdr->PointerToSymbolTable
1920             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
1921
1922    /* Allocate space for any (local, anonymous) .bss sections. */
1923
1924    for (i = 0; i < hdr->NumberOfSections; i++) {
1925       UChar* zspace;
1926       COFF_section* sectab_i
1927          = (COFF_section*)
1928            myindex ( sizeof_COFF_section, sectab, i );
1929       if (0 != strcmp(sectab_i->Name, ".bss")) continue;
1930       if (sectab_i->VirtualSize == 0) continue;
1931       /* This is a non-empty .bss section.  Allocate zeroed space for
1932          it, and set its PointerToRawData field such that oc->image +
1933          PointerToRawData == addr_of_zeroed_space.  */
1934       zspace = stgCallocBytes(1, sectab_i->VirtualSize,
1935                               "ocGetNames_PEi386(anonymous bss)");
1936       sectab_i->PointerToRawData = ((UChar*)zspace) - ((UChar*)(oc->image));
1937       addProddableBlock(oc, zspace, sectab_i->VirtualSize);
1938       /* debugBelch("BSS anon section at 0x%x\n", zspace); */
1939    }
1940
1941    /* Copy section information into the ObjectCode. */
1942
1943    for (i = 0; i < hdr->NumberOfSections; i++) {
1944       UChar* start;
1945       UChar* end;
1946       UInt32 sz;
1947
1948       SectionKind kind
1949          = SECTIONKIND_OTHER;
1950       COFF_section* sectab_i
1951          = (COFF_section*)
1952            myindex ( sizeof_COFF_section, sectab, i );
1953       IF_DEBUG(linker, debugBelch("section name = %s\n", sectab_i->Name ));
1954
1955 #     if 0
1956       /* I'm sure this is the Right Way to do it.  However, the
1957          alternative of testing the sectab_i->Name field seems to
1958          work ok with Cygwin.
1959       */
1960       if (sectab_i->Characteristics & MYIMAGE_SCN_CNT_CODE ||
1961           sectab_i->Characteristics & MYIMAGE_SCN_CNT_INITIALIZED_DATA)
1962          kind = SECTIONKIND_CODE_OR_RODATA;
1963 #     endif
1964
1965       if (0==strcmp(".text",sectab_i->Name) ||
1966           0==strcmp(".rodata",sectab_i->Name))
1967          kind = SECTIONKIND_CODE_OR_RODATA;
1968       if (0==strcmp(".data",sectab_i->Name) ||
1969           0==strcmp(".bss",sectab_i->Name))
1970          kind = SECTIONKIND_RWDATA;
1971
1972       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
1973       sz = sectab_i->SizeOfRawData;
1974       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
1975
1976       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
1977       end   = start + sz - 1;
1978
1979       if (kind == SECTIONKIND_OTHER
1980           /* Ignore sections called which contain stabs debugging
1981              information. */
1982           && 0 != strcmp(".stab", sectab_i->Name)
1983           && 0 != strcmp(".stabstr", sectab_i->Name)
1984          ) {
1985          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", sectab_i->Name, oc->fileName);
1986          return 0;
1987       }
1988
1989       if (kind != SECTIONKIND_OTHER && end >= start) {
1990          addSection(oc, kind, start, end);
1991          addProddableBlock(oc, start, end - start + 1);
1992       }
1993    }
1994
1995    /* Copy exported symbols into the ObjectCode. */
1996
1997    oc->n_symbols = hdr->NumberOfSymbols;
1998    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
1999                                   "ocGetNames_PEi386(oc->symbols)");
2000    /* Call me paranoid; I don't care. */
2001    for (i = 0; i < oc->n_symbols; i++)
2002       oc->symbols[i] = NULL;
2003
2004    i = 0;
2005    while (1) {
2006       COFF_symbol* symtab_i;
2007       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2008       symtab_i = (COFF_symbol*)
2009                  myindex ( sizeof_COFF_symbol, symtab, i );
2010
2011       addr  = NULL;
2012
2013       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
2014           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
2015          /* This symbol is global and defined, viz, exported */
2016          /* for MYIMAGE_SYMCLASS_EXTERNAL
2017                 && !MYIMAGE_SYM_UNDEFINED,
2018             the address of the symbol is:
2019                 address of relevant section + offset in section
2020          */
2021          COFF_section* sectabent
2022             = (COFF_section*) myindex ( sizeof_COFF_section,
2023                                         sectab,
2024                                         symtab_i->SectionNumber-1 );
2025          addr = ((UChar*)(oc->image))
2026                 + (sectabent->PointerToRawData
2027                    + symtab_i->Value);
2028       }
2029       else
2030       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
2031           && symtab_i->Value > 0) {
2032          /* This symbol isn't in any section at all, ie, global bss.
2033             Allocate zeroed space for it. */
2034          addr = stgCallocBytes(1, symtab_i->Value,
2035                                "ocGetNames_PEi386(non-anonymous bss)");
2036          addSection(oc, SECTIONKIND_RWDATA, addr,
2037                         ((UChar*)addr) + symtab_i->Value - 1);
2038          addProddableBlock(oc, addr, symtab_i->Value);
2039          /* debugBelch("BSS      section at 0x%x\n", addr); */
2040       }
2041
2042       if (addr != NULL ) {
2043          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
2044          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
2045          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
2046          ASSERT(i >= 0 && i < oc->n_symbols);
2047          /* cstring_from_COFF_symbol_name always succeeds. */
2048          oc->symbols[i] = sname;
2049          ghciInsertStrHashTable(oc->fileName, symhash, sname, addr);
2050       } else {
2051 #        if 0
2052          debugBelch(
2053                    "IGNORING symbol %d\n"
2054                    "     name `",
2055                    i
2056                  );
2057          printName ( symtab_i->Name, strtab );
2058          debugBelch(
2059                    "'\n"
2060                    "    value 0x%x\n"
2061                    "   1+sec# %d\n"
2062                    "     type 0x%x\n"
2063                    "   sclass 0x%x\n"
2064                    "     nAux %d\n",
2065                    symtab_i->Value,
2066                    (Int32)(symtab_i->SectionNumber),
2067                    (UInt32)symtab_i->Type,
2068                    (UInt32)symtab_i->StorageClass,
2069                    (UInt32)symtab_i->NumberOfAuxSymbols
2070                  );
2071 #        endif
2072       }
2073
2074       i += symtab_i->NumberOfAuxSymbols;
2075       i++;
2076    }
2077
2078    return 1;
2079 }
2080
2081
2082 static int
2083 ocResolve_PEi386 ( ObjectCode* oc )
2084 {
2085    COFF_header*  hdr;
2086    COFF_section* sectab;
2087    COFF_symbol*  symtab;
2088    UChar*        strtab;
2089
2090    UInt32        A;
2091    UInt32        S;
2092    UInt32*       pP;
2093
2094    int i;
2095    UInt32 j, noRelocs;
2096
2097    /* ToDo: should be variable-sized?  But is at least safe in the
2098       sense of buffer-overrun-proof. */
2099    char symbol[1000];
2100    /* debugBelch("resolving for %s\n", oc->fileName); */
2101
2102    hdr = (COFF_header*)(oc->image);
2103    sectab = (COFF_section*) (
2104                ((UChar*)(oc->image))
2105                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2106             );
2107    symtab = (COFF_symbol*) (
2108                ((UChar*)(oc->image))
2109                + hdr->PointerToSymbolTable
2110             );
2111    strtab = ((UChar*)(oc->image))
2112             + hdr->PointerToSymbolTable
2113             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2114
2115    for (i = 0; i < hdr->NumberOfSections; i++) {
2116       COFF_section* sectab_i
2117          = (COFF_section*)
2118            myindex ( sizeof_COFF_section, sectab, i );
2119       COFF_reloc* reltab
2120          = (COFF_reloc*) (
2121               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2122            );
2123
2124       /* Ignore sections called which contain stabs debugging
2125          information. */
2126       if (0 == strcmp(".stab", sectab_i->Name)
2127           || 0 == strcmp(".stabstr", sectab_i->Name))
2128          continue;
2129
2130       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2131         /* If the relocation field (a short) has overflowed, the
2132          * real count can be found in the first reloc entry.
2133          *
2134          * See Section 4.1 (last para) of the PE spec (rev6.0).
2135          *
2136          * Nov2003 update: the GNU linker still doesn't correctly
2137          * handle the generation of relocatable object files with
2138          * overflown relocations. Hence the output to warn of potential
2139          * troubles.
2140          */
2141         COFF_reloc* rel = (COFF_reloc*)
2142                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2143         noRelocs = rel->VirtualAddress;
2144         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
2145                    noRelocs);
2146         j = 1;
2147       } else {
2148         noRelocs = sectab_i->NumberOfRelocations;
2149         j = 0;
2150       }
2151
2152
2153       for (; j < noRelocs; j++) {
2154          COFF_symbol* sym;
2155          COFF_reloc* reltab_j
2156             = (COFF_reloc*)
2157               myindex ( sizeof_COFF_reloc, reltab, j );
2158
2159          /* the location to patch */
2160          pP = (UInt32*)(
2161                  ((UChar*)(oc->image))
2162                  + (sectab_i->PointerToRawData
2163                     + reltab_j->VirtualAddress
2164                     - sectab_i->VirtualAddress )
2165               );
2166          /* the existing contents of pP */
2167          A = *pP;
2168          /* the symbol to connect to */
2169          sym = (COFF_symbol*)
2170                myindex ( sizeof_COFF_symbol,
2171                          symtab, reltab_j->SymbolTableIndex );
2172          IF_DEBUG(linker,
2173                   debugBelch(
2174                             "reloc sec %2d num %3d:  type 0x%-4x   "
2175                             "vaddr 0x%-8x   name `",
2176                             i, j,
2177                             (UInt32)reltab_j->Type,
2178                             reltab_j->VirtualAddress );
2179                             printName ( sym->Name, strtab );
2180                             debugBelch("'\n" ));
2181
2182          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
2183             COFF_section* section_sym
2184                = findPEi386SectionCalled ( oc, sym->Name );
2185             if (!section_sym) {
2186                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
2187                return 0;
2188             }
2189             S = ((UInt32)(oc->image))
2190                 + (section_sym->PointerToRawData
2191                    + sym->Value);
2192          } else {
2193             copyName ( sym->Name, strtab, symbol, 1000-1 );
2194             (void*)S = lookupLocalSymbol( oc, symbol );
2195             if ((void*)S != NULL) goto foundit;
2196             (void*)S = lookupSymbol( symbol );
2197             if ((void*)S != NULL) goto foundit;
2198             zapTrailingAtSign ( symbol );
2199             (void*)S = lookupLocalSymbol( oc, symbol );
2200             if ((void*)S != NULL) goto foundit;
2201             (void*)S = lookupSymbol( symbol );
2202             if ((void*)S != NULL) goto foundit;
2203             /* Newline first because the interactive linker has printed "linking..." */
2204             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, symbol);
2205             return 0;
2206            foundit:;
2207          }
2208          checkProddableBlock(oc, pP);
2209          switch (reltab_j->Type) {
2210             case MYIMAGE_REL_I386_DIR32:
2211                *pP = A + S;
2212                break;
2213             case MYIMAGE_REL_I386_REL32:
2214                /* Tricky.  We have to insert a displacement at
2215                   pP which, when added to the PC for the _next_
2216                   insn, gives the address of the target (S).
2217                   Problem is to know the address of the next insn
2218                   when we only know pP.  We assume that this
2219                   literal field is always the last in the insn,
2220                   so that the address of the next insn is pP+4
2221                   -- hence the constant 4.
2222                   Also I don't know if A should be added, but so
2223                   far it has always been zero.
2224                */
2225                ASSERT(A==0);
2226                *pP = S - ((UInt32)pP) - 4;
2227                break;
2228             default:
2229                debugBelch("%s: unhandled PEi386 relocation type %d",
2230                      oc->fileName, reltab_j->Type);
2231                return 0;
2232          }
2233
2234       }
2235    }
2236
2237    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
2238    return 1;
2239 }
2240
2241 #endif /* defined(OBJFORMAT_PEi386) */
2242
2243
2244 /* --------------------------------------------------------------------------
2245  * ELF specifics
2246  * ------------------------------------------------------------------------*/
2247
2248 #if defined(OBJFORMAT_ELF)
2249
2250 #define FALSE 0
2251 #define TRUE  1
2252
2253 #if defined(sparc_HOST_ARCH)
2254 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
2255 #elif defined(i386_HOST_ARCH)
2256 #  define ELF_TARGET_386    /* Used inside <elf.h> */
2257 #elif defined(x86_64_HOST_ARCH)
2258 #  define ELF_TARGET_X64_64
2259 #  define ELF_64BIT
2260 #elif defined (ia64_HOST_ARCH)
2261 #  define ELF_TARGET_IA64   /* Used inside <elf.h> */
2262 #  define ELF_64BIT
2263 #  define ELF_FUNCTION_DESC /* calling convention uses function descriptors */
2264 #  define ELF_NEED_GOT      /* needs Global Offset Table */
2265 #  define ELF_NEED_PLT      /* needs Procedure Linkage Tables */
2266 #endif
2267
2268 #if !defined(openbsd_HOST_OS)
2269 #include <elf.h>
2270 #else
2271 /* openbsd elf has things in different places, with diff names */
2272 #include <elf_abi.h>
2273 #include <machine/reloc.h>
2274 #define R_386_32    RELOC_32
2275 #define R_386_PC32  RELOC_PC32
2276 #endif
2277
2278 /*
2279  * Define a set of types which can be used for both ELF32 and ELF64
2280  */
2281
2282 #ifdef ELF_64BIT
2283 #define ELFCLASS    ELFCLASS64
2284 #define Elf_Addr    Elf64_Addr
2285 #define Elf_Word    Elf64_Word
2286 #define Elf_Sword   Elf64_Sword
2287 #define Elf_Ehdr    Elf64_Ehdr
2288 #define Elf_Phdr    Elf64_Phdr
2289 #define Elf_Shdr    Elf64_Shdr
2290 #define Elf_Sym     Elf64_Sym
2291 #define Elf_Rel     Elf64_Rel
2292 #define Elf_Rela    Elf64_Rela
2293 #define ELF_ST_TYPE ELF64_ST_TYPE
2294 #define ELF_ST_BIND ELF64_ST_BIND
2295 #define ELF_R_TYPE  ELF64_R_TYPE
2296 #define ELF_R_SYM   ELF64_R_SYM
2297 #else
2298 #define ELFCLASS    ELFCLASS32
2299 #define Elf_Addr    Elf32_Addr
2300 #define Elf_Word    Elf32_Word
2301 #define Elf_Sword   Elf32_Sword
2302 #define Elf_Ehdr    Elf32_Ehdr
2303 #define Elf_Phdr    Elf32_Phdr
2304 #define Elf_Shdr    Elf32_Shdr
2305 #define Elf_Sym     Elf32_Sym
2306 #define Elf_Rel     Elf32_Rel
2307 #define Elf_Rela    Elf32_Rela
2308 #ifndef ELF_ST_TYPE
2309 #define ELF_ST_TYPE ELF32_ST_TYPE
2310 #endif
2311 #ifndef ELF_ST_BIND
2312 #define ELF_ST_BIND ELF32_ST_BIND
2313 #endif
2314 #ifndef ELF_R_TYPE
2315 #define ELF_R_TYPE  ELF32_R_TYPE
2316 #endif
2317 #ifndef ELF_R_SYM
2318 #define ELF_R_SYM   ELF32_R_SYM
2319 #endif
2320 #endif
2321
2322
2323 /*
2324  * Functions to allocate entries in dynamic sections.  Currently we simply
2325  * preallocate a large number, and we don't check if a entry for the given
2326  * target already exists (a linear search is too slow).  Ideally these
2327  * entries would be associated with symbols.
2328  */
2329
2330 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
2331 #define GOT_SIZE            0x20000
2332 #define FUNCTION_TABLE_SIZE 0x10000
2333 #define PLT_SIZE            0x08000
2334
2335 #ifdef ELF_NEED_GOT
2336 static Elf_Addr got[GOT_SIZE];
2337 static unsigned int gotIndex;
2338 static Elf_Addr gp_val = (Elf_Addr)got;
2339
2340 static Elf_Addr
2341 allocateGOTEntry(Elf_Addr target)
2342 {
2343    Elf_Addr *entry;
2344
2345    if (gotIndex >= GOT_SIZE)
2346       barf("Global offset table overflow");
2347
2348    entry = &got[gotIndex++];
2349    *entry = target;
2350    return (Elf_Addr)entry;
2351 }
2352 #endif
2353
2354 #ifdef ELF_FUNCTION_DESC
2355 typedef struct {
2356    Elf_Addr ip;
2357    Elf_Addr gp;
2358 } FunctionDesc;
2359
2360 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
2361 static unsigned int functionTableIndex;
2362
2363 static Elf_Addr
2364 allocateFunctionDesc(Elf_Addr target)
2365 {
2366    FunctionDesc *entry;
2367
2368    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
2369       barf("Function table overflow");
2370
2371    entry = &functionTable[functionTableIndex++];
2372    entry->ip = target;
2373    entry->gp = (Elf_Addr)gp_val;
2374    return (Elf_Addr)entry;
2375 }
2376
2377 static Elf_Addr
2378 copyFunctionDesc(Elf_Addr target)
2379 {
2380    FunctionDesc *olddesc = (FunctionDesc *)target;
2381    FunctionDesc *newdesc;
2382
2383    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
2384    newdesc->gp = olddesc->gp;
2385    return (Elf_Addr)newdesc;
2386 }
2387 #endif
2388
2389 #ifdef ELF_NEED_PLT
2390 #ifdef ia64_HOST_ARCH
2391 static void ia64_reloc_gprel22(Elf_Addr target, Elf_Addr value);
2392 static void ia64_reloc_pcrel21(Elf_Addr target, Elf_Addr value, ObjectCode *oc);
2393
2394 static unsigned char plt_code[] =
2395 {
2396    /* taken from binutils bfd/elfxx-ia64.c */
2397    0x0b, 0x78, 0x00, 0x02, 0x00, 0x24,  /*   [MMI]       addl r15=0,r1;;    */
2398    0x00, 0x41, 0x3c, 0x30, 0x28, 0xc0,  /*               ld8 r16=[r15],8    */
2399    0x01, 0x08, 0x00, 0x84,              /*               mov r14=r1;;       */
2400    0x11, 0x08, 0x00, 0x1e, 0x18, 0x10,  /*   [MIB]       ld8 r1=[r15]       */
2401    0x60, 0x80, 0x04, 0x80, 0x03, 0x00,  /*               mov b6=r16         */
2402    0x60, 0x00, 0x80, 0x00               /*               br.few b6;;        */
2403 };
2404
2405 /* If we can't get to the function descriptor via gp, take a local copy of it */
2406 #define PLT_RELOC(code, target) { \
2407    Elf64_Sxword rel_value = target - gp_val; \
2408    if ((rel_value > 0x1fffff) || (rel_value < -0x1fffff)) \
2409       ia64_reloc_gprel22((Elf_Addr)code, copyFunctionDesc(target)); \
2410    else \
2411       ia64_reloc_gprel22((Elf_Addr)code, target); \
2412    }
2413 #endif
2414
2415 typedef struct {
2416    unsigned char code[sizeof(plt_code)];
2417 } PLTEntry;
2418
2419 static Elf_Addr
2420 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
2421 {
2422    PLTEntry *plt = (PLTEntry *)oc->plt;
2423    PLTEntry *entry;
2424
2425    if (oc->pltIndex >= PLT_SIZE)
2426       barf("Procedure table overflow");
2427
2428    entry = &plt[oc->pltIndex++];
2429    memcpy(entry->code, plt_code, sizeof(entry->code));
2430    PLT_RELOC(entry->code, target);
2431    return (Elf_Addr)entry;
2432 }
2433
2434 static unsigned int
2435 PLTSize(void)
2436 {
2437    return (PLT_SIZE * sizeof(PLTEntry));
2438 }
2439 #endif
2440
2441
2442 /*
2443  * Generic ELF functions
2444  */
2445
2446 static char *
2447 findElfSection ( void* objImage, Elf_Word sh_type )
2448 {
2449    char* ehdrC = (char*)objImage;
2450    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
2451    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
2452    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
2453    char* ptr = NULL;
2454    int i;
2455
2456    for (i = 0; i < ehdr->e_shnum; i++) {
2457       if (shdr[i].sh_type == sh_type
2458           /* Ignore the section header's string table. */
2459           && i != ehdr->e_shstrndx
2460           /* Ignore string tables named .stabstr, as they contain
2461              debugging info. */
2462           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
2463          ) {
2464          ptr = ehdrC + shdr[i].sh_offset;
2465          break;
2466       }
2467    }
2468    return ptr;
2469 }
2470
2471 #if defined(ia64_HOST_ARCH)
2472 static Elf_Addr
2473 findElfSegment ( void* objImage, Elf_Addr vaddr )
2474 {
2475    char* ehdrC = (char*)objImage;
2476    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
2477    Elf_Phdr* phdr = (Elf_Phdr*)(ehdrC + ehdr->e_phoff);
2478    Elf_Addr segaddr = 0;
2479    int i;
2480
2481    for (i = 0; i < ehdr->e_phnum; i++) {
2482       segaddr = phdr[i].p_vaddr;
2483       if ((vaddr >= segaddr) && (vaddr < segaddr + phdr[i].p_memsz))
2484               break;
2485    }
2486    return segaddr;
2487 }
2488 #endif
2489
2490 static int
2491 ocVerifyImage_ELF ( ObjectCode* oc )
2492 {
2493    Elf_Shdr* shdr;
2494    Elf_Sym*  stab;
2495    int i, j, nent, nstrtab, nsymtabs;
2496    char* sh_strtab;
2497    char* strtab;
2498
2499    char*     ehdrC = (char*)(oc->image);
2500    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
2501
2502    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
2503        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
2504        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
2505        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
2506       errorBelch("%s: not an ELF object", oc->fileName);
2507       return 0;
2508    }
2509
2510    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
2511       errorBelch("%s: unsupported ELF format", oc->fileName);
2512       return 0;
2513    }
2514
2515    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
2516        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
2517    } else
2518    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
2519        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
2520    } else {
2521        errorBelch("%s: unknown endiannness", oc->fileName);
2522        return 0;
2523    }
2524
2525    if (ehdr->e_type != ET_REL) {
2526       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
2527       return 0;
2528    }
2529    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
2530
2531    IF_DEBUG(linker,debugBelch( "Architecture is " ));
2532    switch (ehdr->e_machine) {
2533       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
2534       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
2535 #ifdef EM_IA_64
2536       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
2537 #endif
2538       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
2539 #ifdef EM_X86_64
2540       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
2541 #endif
2542       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
2543                      errorBelch("%s: unknown architecture", oc->fileName);
2544                      return 0;
2545    }
2546
2547    IF_DEBUG(linker,debugBelch(
2548              "\nSection header table: start %d, n_entries %d, ent_size %d\n",
2549              ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
2550
2551    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
2552
2553    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
2554
2555    if (ehdr->e_shstrndx == SHN_UNDEF) {
2556       errorBelch("%s: no section header string table", oc->fileName);
2557       return 0;
2558    } else {
2559       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
2560                           ehdr->e_shstrndx));
2561       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
2562    }
2563
2564    for (i = 0; i < ehdr->e_shnum; i++) {
2565       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
2566       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
2567       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
2568       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
2569       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
2570                ehdrC + shdr[i].sh_offset,
2571                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
2572
2573       if (shdr[i].sh_type == SHT_REL) {
2574           IF_DEBUG(linker,debugBelch("Rel  " ));
2575       } else if (shdr[i].sh_type == SHT_RELA) {
2576           IF_DEBUG(linker,debugBelch("RelA " ));
2577       } else {
2578           IF_DEBUG(linker,debugBelch("     "));
2579       }
2580       if (sh_strtab) {
2581           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
2582       }
2583    }
2584
2585    IF_DEBUG(linker,debugBelch( "\nString tables" ));
2586    strtab = NULL;
2587    nstrtab = 0;
2588    for (i = 0; i < ehdr->e_shnum; i++) {
2589       if (shdr[i].sh_type == SHT_STRTAB
2590           /* Ignore the section header's string table. */
2591           && i != ehdr->e_shstrndx
2592           /* Ignore string tables named .stabstr, as they contain
2593              debugging info. */
2594           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
2595          ) {
2596          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
2597          strtab = ehdrC + shdr[i].sh_offset;
2598          nstrtab++;
2599       }
2600    }
2601    if (nstrtab != 1) {
2602       errorBelch("%s: no string tables, or too many", oc->fileName);
2603       return 0;
2604    }
2605
2606    nsymtabs = 0;
2607    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
2608    for (i = 0; i < ehdr->e_shnum; i++) {
2609       if (shdr[i].sh_type != SHT_SYMTAB) continue;
2610       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
2611       nsymtabs++;
2612       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
2613       nent = shdr[i].sh_size / sizeof(Elf_Sym);
2614       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%d rem)\n",
2615                nent,
2616                shdr[i].sh_size % sizeof(Elf_Sym)
2617              ));
2618       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
2619          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
2620          return 0;
2621       }
2622       for (j = 0; j < nent; j++) {
2623          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
2624          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
2625                              (int)stab[j].st_shndx,
2626                              (int)stab[j].st_size,
2627                              (char*)stab[j].st_value ));
2628
2629          IF_DEBUG(linker,debugBelch("type=" ));
2630          switch (ELF_ST_TYPE(stab[j].st_info)) {
2631             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
2632             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
2633             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
2634             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
2635             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
2636             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
2637          }
2638          IF_DEBUG(linker,debugBelch("  " ));
2639
2640          IF_DEBUG(linker,debugBelch("bind=" ));
2641          switch (ELF_ST_BIND(stab[j].st_info)) {
2642             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
2643             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
2644             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
2645             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
2646          }
2647          IF_DEBUG(linker,debugBelch("  " ));
2648
2649          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
2650       }
2651    }
2652
2653    if (nsymtabs == 0) {
2654       errorBelch("%s: didn't find any symbol tables", oc->fileName);
2655       return 0;
2656    }
2657
2658    return 1;
2659 }
2660
2661 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
2662 {
2663     *is_bss = FALSE;
2664
2665     if (hdr->sh_type == SHT_PROGBITS
2666         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
2667         /* .text-style section */
2668         return SECTIONKIND_CODE_OR_RODATA;
2669     }
2670
2671     if (hdr->sh_type == SHT_PROGBITS
2672             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
2673             /* .data-style section */
2674             return SECTIONKIND_RWDATA;
2675     }
2676
2677     if (hdr->sh_type == SHT_PROGBITS
2678         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
2679         /* .rodata-style section */
2680         return SECTIONKIND_CODE_OR_RODATA;
2681     }
2682
2683     if (hdr->sh_type == SHT_NOBITS
2684         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
2685         /* .bss-style section */
2686         *is_bss = TRUE;
2687         return SECTIONKIND_RWDATA;
2688     }
2689
2690     return SECTIONKIND_OTHER;
2691 }
2692
2693
2694 static int
2695 ocGetNames_ELF ( ObjectCode* oc )
2696 {
2697    int i, j, k, nent;
2698    Elf_Sym* stab;
2699
2700    char*     ehdrC    = (char*)(oc->image);
2701    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
2702    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
2703    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
2704
2705    ASSERT(symhash != NULL);
2706
2707    if (!strtab) {
2708       errorBelch("%s: no strtab", oc->fileName);
2709       return 0;
2710    }
2711
2712    k = 0;
2713    for (i = 0; i < ehdr->e_shnum; i++) {
2714       /* Figure out what kind of section it is.  Logic derived from
2715          Figure 1.14 ("Special Sections") of the ELF document
2716          ("Portable Formats Specification, Version 1.1"). */
2717       int         is_bss = FALSE;
2718       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
2719
2720       if (is_bss && shdr[i].sh_size > 0) {
2721          /* This is a non-empty .bss section.  Allocate zeroed space for
2722             it, and set its .sh_offset field such that
2723             ehdrC + .sh_offset == addr_of_zeroed_space.  */
2724          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
2725                                        "ocGetNames_ELF(BSS)");
2726          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
2727          /*
2728          debugBelch("BSS section at 0x%x, size %d\n",
2729                          zspace, shdr[i].sh_size);
2730          */
2731       }
2732
2733       /* fill in the section info */
2734       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
2735          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
2736          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
2737                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
2738       }
2739
2740       if (shdr[i].sh_type != SHT_SYMTAB) continue;
2741
2742       /* copy stuff into this module's object symbol table */
2743       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
2744       nent = shdr[i].sh_size / sizeof(Elf_Sym);
2745
2746       oc->n_symbols = nent;
2747       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
2748                                    "ocGetNames_ELF(oc->symbols)");
2749
2750       for (j = 0; j < nent; j++) {
2751
2752          char  isLocal = FALSE; /* avoids uninit-var warning */
2753          char* ad      = NULL;
2754          char* nm      = strtab + stab[j].st_name;
2755          int   secno   = stab[j].st_shndx;
2756
2757          /* Figure out if we want to add it; if so, set ad to its
2758             address.  Otherwise leave ad == NULL. */
2759
2760          if (secno == SHN_COMMON) {
2761             isLocal = FALSE;
2762             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
2763             /*
2764             debugBelch("COMMON symbol, size %d name %s\n",
2765                             stab[j].st_size, nm);
2766             */
2767             /* Pointless to do addProddableBlock() for this area,
2768                since the linker should never poke around in it. */
2769          }
2770          else
2771          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
2772                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
2773               )
2774               /* and not an undefined symbol */
2775               && stab[j].st_shndx != SHN_UNDEF
2776               /* and not in a "special section" */
2777               && stab[j].st_shndx < SHN_LORESERVE
2778               &&
2779               /* and it's a not a section or string table or anything silly */
2780               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
2781                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
2782                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
2783               )
2784             ) {
2785             /* Section 0 is the undefined section, hence > and not >=. */
2786             ASSERT(secno > 0 && secno < ehdr->e_shnum);
2787             /*
2788             if (shdr[secno].sh_type == SHT_NOBITS) {
2789                debugBelch("   BSS symbol, size %d off %d name %s\n",
2790                                stab[j].st_size, stab[j].st_value, nm);
2791             }
2792             */
2793             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
2794             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
2795                isLocal = TRUE;
2796             } else {
2797 #ifdef ELF_FUNCTION_DESC
2798                /* dlsym() and the initialisation table both give us function
2799                 * descriptors, so to be consistent we store function descriptors
2800                 * in the symbol table */
2801                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
2802                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
2803 #endif
2804                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s",
2805                                       ad, oc->fileName, nm ));
2806                isLocal = FALSE;
2807             }
2808          }
2809
2810          /* And the decision is ... */
2811
2812          if (ad != NULL) {
2813             ASSERT(nm != NULL);
2814             oc->symbols[j] = nm;
2815             /* Acquire! */
2816             if (isLocal) {
2817                /* Ignore entirely. */
2818             } else {
2819                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
2820             }
2821          } else {
2822             /* Skip. */
2823             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
2824                                    strtab + stab[j].st_name ));
2825             /*
2826             debugBelch(
2827                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
2828                     (int)ELF_ST_BIND(stab[j].st_info),
2829                     (int)ELF_ST_TYPE(stab[j].st_info),
2830                     (int)stab[j].st_shndx,
2831                     strtab + stab[j].st_name
2832                    );
2833             */
2834             oc->symbols[j] = NULL;
2835          }
2836
2837       }
2838    }
2839
2840    return 1;
2841 }
2842
2843 /* Do ELF relocations which lack an explicit addend.  All x86-linux
2844    relocations appear to be of this form. */
2845 static int
2846 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
2847                          Elf_Shdr* shdr, int shnum,
2848                          Elf_Sym*  stab, char* strtab )
2849 {
2850    int j;
2851    char *symbol;
2852    Elf_Word* targ;
2853    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
2854    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
2855    int target_shndx = shdr[shnum].sh_info;
2856    int symtab_shndx = shdr[shnum].sh_link;
2857
2858    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
2859    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
2860    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
2861                           target_shndx, symtab_shndx ));
2862
2863    /* Skip sections that we're not interested in. */
2864    {
2865        int is_bss;
2866        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
2867        if (kind == SECTIONKIND_OTHER) {
2868            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
2869            return 1;
2870        }
2871    }
2872
2873    for (j = 0; j < nent; j++) {
2874       Elf_Addr offset = rtab[j].r_offset;
2875       Elf_Addr info   = rtab[j].r_info;
2876
2877       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
2878       Elf_Word* pP = (Elf_Word*)P;
2879       Elf_Addr  A  = *pP;
2880       Elf_Addr  S;
2881       void*     S_tmp;
2882       Elf_Addr  value;
2883
2884       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
2885                              j, (void*)offset, (void*)info ));
2886       if (!info) {
2887          IF_DEBUG(linker,debugBelch( " ZERO" ));
2888          S = 0;
2889       } else {
2890          Elf_Sym sym = stab[ELF_R_SYM(info)];
2891          /* First see if it is a local symbol. */
2892          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
2893             /* Yes, so we can get the address directly from the ELF symbol
2894                table. */
2895             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
2896             S = (Elf_Addr)
2897                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
2898                        + stab[ELF_R_SYM(info)].st_value);
2899
2900          } else {
2901             /* No, so look up the name in our global table. */
2902             symbol = strtab + sym.st_name;
2903             S_tmp = lookupSymbol( symbol );
2904             S = (Elf_Addr)S_tmp;
2905          }
2906          if (!S) {
2907             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
2908             return 0;
2909          }
2910          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
2911       }
2912
2913       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
2914                              (void*)P, (void*)S, (void*)A ));
2915       checkProddableBlock ( oc, pP );
2916
2917       value = S + A;
2918
2919       switch (ELF_R_TYPE(info)) {
2920 #        ifdef i386_HOST_ARCH
2921          case R_386_32:   *pP = value;     break;
2922          case R_386_PC32: *pP = value - P; break;
2923 #        endif
2924          default:
2925             errorBelch("%s: unhandled ELF relocation(Rel) type %d\n",
2926                   oc->fileName, ELF_R_TYPE(info));
2927             return 0;
2928       }
2929
2930    }
2931    return 1;
2932 }
2933
2934 /* Do ELF relocations for which explicit addends are supplied.
2935    sparc-solaris relocations appear to be of this form. */
2936 static int
2937 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
2938                           Elf_Shdr* shdr, int shnum,
2939                           Elf_Sym*  stab, char* strtab )
2940 {
2941    int j;
2942    char *symbol;
2943    Elf_Addr targ;
2944    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
2945    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
2946    int target_shndx = shdr[shnum].sh_info;
2947    int symtab_shndx = shdr[shnum].sh_link;
2948
2949    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
2950    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
2951    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
2952                           target_shndx, symtab_shndx ));
2953
2954    for (j = 0; j < nent; j++) {
2955 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
2956       /* This #ifdef only serves to avoid unused-var warnings. */
2957       Elf_Addr  offset = rtab[j].r_offset;
2958       Elf_Addr  P      = targ + offset;
2959 #endif
2960       Elf_Addr  info   = rtab[j].r_info;
2961       Elf_Addr  A      = rtab[j].r_addend;
2962       Elf_Addr  S;
2963       void*     S_tmp;
2964       Elf_Addr  value;
2965 #     if defined(sparc_HOST_ARCH)
2966       Elf_Word* pP = (Elf_Word*)P;
2967       Elf_Word  w1, w2;
2968 #     elif defined(ia64_HOST_ARCH)
2969       Elf64_Xword *pP = (Elf64_Xword *)P;
2970       Elf_Addr addr;
2971 #     elif defined(powerpc_HOST_ARCH)
2972       Elf_Sword delta;
2973 #     endif
2974
2975       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
2976                              j, (void*)offset, (void*)info,
2977                                 (void*)A ));
2978       if (!info) {
2979          IF_DEBUG(linker,debugBelch( " ZERO" ));
2980          S = 0;
2981       } else {
2982          Elf_Sym sym = stab[ELF_R_SYM(info)];
2983          /* First see if it is a local symbol. */
2984          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
2985             /* Yes, so we can get the address directly from the ELF symbol
2986                table. */
2987             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
2988             S = (Elf_Addr)
2989                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
2990                        + stab[ELF_R_SYM(info)].st_value);
2991 #ifdef ELF_FUNCTION_DESC
2992             /* Make a function descriptor for this function */
2993             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
2994                S = allocateFunctionDesc(S + A);
2995                A = 0;
2996             }
2997 #endif
2998          } else {
2999             /* No, so look up the name in our global table. */
3000             symbol = strtab + sym.st_name;
3001             S_tmp = lookupSymbol( symbol );
3002             S = (Elf_Addr)S_tmp;
3003
3004 #ifdef ELF_FUNCTION_DESC
3005             /* If a function, already a function descriptor - we would
3006                have to copy it to add an offset. */
3007             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
3008                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
3009 #endif
3010          }
3011          if (!S) {
3012            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3013            return 0;
3014          }
3015          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
3016       }
3017
3018       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
3019                                         (void*)P, (void*)S, (void*)A ));
3020       /* checkProddableBlock ( oc, (void*)P ); */
3021
3022       value = S + A;
3023
3024       switch (ELF_R_TYPE(info)) {
3025 #        if defined(sparc_HOST_ARCH)
3026          case R_SPARC_WDISP30:
3027             w1 = *pP & 0xC0000000;
3028             w2 = (Elf_Word)((value - P) >> 2);
3029             ASSERT((w2 & 0xC0000000) == 0);
3030             w1 |= w2;
3031             *pP = w1;
3032             break;
3033          case R_SPARC_HI22:
3034             w1 = *pP & 0xFFC00000;
3035             w2 = (Elf_Word)(value >> 10);
3036             ASSERT((w2 & 0xFFC00000) == 0);
3037             w1 |= w2;
3038             *pP = w1;
3039             break;
3040          case R_SPARC_LO10:
3041             w1 = *pP & ~0x3FF;
3042             w2 = (Elf_Word)(value & 0x3FF);
3043             ASSERT((w2 & ~0x3FF) == 0);
3044             w1 |= w2;
3045             *pP = w1;
3046             break;
3047          /* According to the Sun documentation:
3048             R_SPARC_UA32
3049             This relocation type resembles R_SPARC_32, except it refers to an
3050             unaligned word. That is, the word to be relocated must be treated
3051             as four separate bytes with arbitrary alignment, not as a word
3052             aligned according to the architecture requirements.
3053
3054             (JRS: which means that freeloading on the R_SPARC_32 case
3055             is probably wrong, but hey ...)
3056          */
3057          case R_SPARC_UA32:
3058          case R_SPARC_32:
3059             w2 = (Elf_Word)value;
3060             *pP = w2;
3061             break;
3062 #        elif defined(ia64_HOST_ARCH)
3063          case R_IA64_DIR64LSB:
3064          case R_IA64_FPTR64LSB:
3065             *pP = value;
3066             break;
3067          case R_IA64_PCREL64LSB:
3068             *pP = value - P;
3069             break;
3070          case R_IA64_SEGREL64LSB:
3071             addr = findElfSegment(ehdrC, value);
3072             *pP = value - addr;
3073             break;
3074          case R_IA64_GPREL22:
3075             ia64_reloc_gprel22(P, value);
3076             break;
3077          case R_IA64_LTOFF22:
3078          case R_IA64_LTOFF22X:
3079          case R_IA64_LTOFF_FPTR22:
3080             addr = allocateGOTEntry(value);
3081             ia64_reloc_gprel22(P, addr);
3082             break;
3083          case R_IA64_PCREL21B:
3084             ia64_reloc_pcrel21(P, S, oc);
3085             break;
3086          case R_IA64_LDXMOV:
3087             /* This goes with R_IA64_LTOFF22X and points to the load to
3088              * convert into a move.  We don't implement relaxation. */
3089             break;
3090 #        elif defined(powerpc_HOST_ARCH)
3091          case R_PPC_ADDR16_LO:
3092             *(Elf32_Half*) P = value;
3093             break;
3094
3095          case R_PPC_ADDR16_HI:
3096             *(Elf32_Half*) P = value >> 16;
3097             break;
3098
3099          case R_PPC_ADDR16_HA:
3100             *(Elf32_Half*) P = (value + 0x8000) >> 16;
3101             break;
3102
3103          case R_PPC_ADDR32:
3104             *(Elf32_Word *) P = value;
3105             break;
3106
3107          case R_PPC_REL32:
3108             *(Elf32_Word *) P = value - P;
3109             break;
3110
3111          case R_PPC_REL24:
3112             delta = value - P;
3113
3114             if( delta << 6 >> 6 != delta )
3115             {
3116                value = makeJumpIsland( oc, ELF_R_SYM(info), value );
3117                delta = value - P;
3118
3119                if( value == 0 || delta << 6 >> 6 != delta )
3120                {
3121                   barf( "Unable to make ppcJumpIsland for #%d",
3122                         ELF_R_SYM(info) );
3123                   return 0;
3124                }
3125             }
3126
3127             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
3128                                           | (delta & 0x3fffffc);
3129             break;
3130 #        endif
3131
3132 #if x86_64_HOST_OS
3133       case R_X86_64_64:
3134           *(Elf64_Xword *)P = value;
3135           break;
3136
3137       case R_X86_64_PC32:
3138           *(Elf64_Word *)P = (Elf64_Word) (value - P);
3139           break;
3140
3141       case R_X86_64_32:
3142           *(Elf64_Word *)P = (Elf64_Word)value;
3143           break;
3144
3145       case R_X86_64_32S:
3146           *(Elf64_Sword *)P = (Elf64_Sword)value;
3147           break;
3148 #endif
3149
3150          default:
3151             errorBelch("%s: unhandled ELF relocation(RelA) type %d\n",
3152                   oc->fileName, ELF_R_TYPE(info));
3153             return 0;
3154       }
3155
3156    }
3157    return 1;
3158 }
3159
3160 static int
3161 ocResolve_ELF ( ObjectCode* oc )
3162 {
3163    char *strtab;
3164    int   shnum, ok;
3165    Elf_Sym*  stab  = NULL;
3166    char*     ehdrC = (char*)(oc->image);
3167    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
3168    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3169
3170    /* first find "the" symbol table */
3171    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
3172
3173    /* also go find the string table */
3174    strtab = findElfSection ( ehdrC, SHT_STRTAB );
3175
3176    if (stab == NULL || strtab == NULL) {
3177       errorBelch("%s: can't find string or symbol table", oc->fileName);
3178       return 0;
3179    }
3180
3181    /* Process the relocation sections. */
3182    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
3183       if (shdr[shnum].sh_type == SHT_REL) {
3184          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
3185                                        shnum, stab, strtab );
3186          if (!ok) return ok;
3187       }
3188       else
3189       if (shdr[shnum].sh_type == SHT_RELA) {
3190          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
3191                                         shnum, stab, strtab );
3192          if (!ok) return ok;
3193       }
3194    }
3195
3196    /* Free the local symbol table; we won't need it again. */
3197    freeHashTable(oc->lochash, NULL);
3198    oc->lochash = NULL;
3199
3200 #if defined(powerpc_HOST_ARCH)
3201    ocFlushInstructionCache( oc );
3202 #endif
3203
3204    return 1;
3205 }
3206
3207 /*
3208  * IA64 specifics
3209  * Instructions are 41 bits long, packed into 128 bit bundles with a 5-bit template
3210  * at the front.  The following utility functions pack and unpack instructions, and
3211  * take care of the most common relocations.
3212  */
3213
3214 #ifdef ia64_HOST_ARCH
3215
3216 static Elf64_Xword
3217 ia64_extract_instruction(Elf64_Xword *target)
3218 {
3219    Elf64_Xword w1, w2;
3220    int slot = (Elf_Addr)target & 3;
3221    (Elf_Addr)target &= ~3;
3222
3223    w1 = *target;
3224    w2 = *(target+1);
3225
3226    switch (slot)
3227    {
3228       case 0:
3229          return ((w1 >> 5) & 0x1ffffffffff);
3230       case 1:
3231          return (w1 >> 46) | ((w2 & 0x7fffff) << 18);
3232       case 2:
3233          return (w2 >> 23);
3234       default:
3235          barf("ia64_extract_instruction: invalid slot %p", target);
3236    }
3237 }
3238
3239 static void
3240 ia64_deposit_instruction(Elf64_Xword *target, Elf64_Xword value)
3241 {
3242    int slot = (Elf_Addr)target & 3;
3243    (Elf_Addr)target &= ~3;
3244
3245    switch (slot)
3246    {
3247       case 0:
3248          *target |= value << 5;
3249          break;
3250       case 1:
3251          *target |= value << 46;
3252          *(target+1) |= value >> 18;
3253          break;
3254       case 2:
3255          *(target+1) |= value << 23;
3256          break;
3257    }
3258 }
3259
3260 static void
3261 ia64_reloc_gprel22(Elf_Addr target, Elf_Addr value)
3262 {
3263    Elf64_Xword instruction;
3264    Elf64_Sxword rel_value;
3265
3266    rel_value = value - gp_val;
3267    if ((rel_value > 0x1fffff) || (rel_value < -0x1fffff))
3268       barf("GP-relative data out of range (address = 0x%lx, gp = 0x%lx)", value, gp_val);
3269
3270    instruction = ia64_extract_instruction((Elf64_Xword *)target);
3271    instruction |= (((rel_value >> 0) & 0x07f) << 13)            /* imm7b */
3272                     | (((rel_value >> 7) & 0x1ff) << 27)        /* imm9d */
3273                     | (((rel_value >> 16) & 0x01f) << 22)       /* imm5c */
3274                     | ((Elf64_Xword)(rel_value < 0) << 36);     /* s */
3275    ia64_deposit_instruction((Elf64_Xword *)target, instruction);
3276 }
3277
3278 static void
3279 ia64_reloc_pcrel21(Elf_Addr target, Elf_Addr value, ObjectCode *oc)
3280 {
3281    Elf64_Xword instruction;
3282    Elf64_Sxword rel_value;
3283    Elf_Addr entry;
3284
3285    entry = allocatePLTEntry(value, oc);
3286
3287    rel_value = (entry >> 4) - (target >> 4);
3288    if ((rel_value > 0xfffff) || (rel_value < -0xfffff))
3289       barf("PLT entry too far away (entry = 0x%lx, target = 0x%lx)", entry, target);
3290
3291    instruction = ia64_extract_instruction((Elf64_Xword *)target);
3292    instruction |= ((rel_value & 0xfffff) << 13)                 /* imm20b */
3293                     | ((Elf64_Xword)(rel_value < 0) << 36);     /* s */
3294    ia64_deposit_instruction((Elf64_Xword *)target, instruction);
3295 }
3296
3297 #endif /* ia64 */
3298
3299 /*
3300  * PowerPC ELF specifics
3301  */
3302
3303 #ifdef powerpc_HOST_ARCH
3304
3305 static int ocAllocateJumpIslands_ELF( ObjectCode *oc )
3306 {
3307   Elf_Ehdr *ehdr;
3308   Elf_Shdr* shdr;
3309   int i;
3310
3311   ehdr = (Elf_Ehdr *) oc->image;
3312   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
3313
3314   for( i = 0; i < ehdr->e_shnum; i++ )
3315     if( shdr[i].sh_type == SHT_SYMTAB )
3316       break;
3317
3318   if( i == ehdr->e_shnum )
3319   {
3320     errorBelch( "This ELF file contains no symtab" );
3321     return 0;
3322   }
3323
3324   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
3325   {
3326     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
3327       shdr[i].sh_entsize, sizeof( Elf_Sym ) );
3328
3329     return 0;
3330   }
3331
3332   return ocAllocateJumpIslands( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
3333 }
3334
3335 #endif /* powerpc */
3336
3337 #endif /* ELF */
3338
3339 /* --------------------------------------------------------------------------
3340  * Mach-O specifics
3341  * ------------------------------------------------------------------------*/
3342
3343 #if defined(OBJFORMAT_MACHO)
3344
3345 /*
3346   Support for MachO linking on Darwin/MacOS X on PowerPC chips
3347   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
3348
3349   I hereby formally apologize for the hackish nature of this code.
3350   Things that need to be done:
3351   *) implement ocVerifyImage_MachO
3352   *) add still more sanity checks.
3353 */
3354
3355 static int ocAllocateJumpIslands_MachO(ObjectCode* oc)
3356 {
3357     struct mach_header *header = (struct mach_header *) oc->image;
3358     struct load_command *lc = (struct load_command *) (header + 1);
3359     unsigned i;
3360
3361     for( i = 0; i < header->ncmds; i++ )
3362     {
3363         if( lc->cmd == LC_SYMTAB )
3364         {
3365                 // Find out the first and last undefined external
3366                 // symbol, so we don't have to allocate too many
3367                 // jump islands.
3368             struct symtab_command *symLC = (struct symtab_command *) lc;
3369             unsigned min = symLC->nsyms, max = 0;
3370             struct nlist *nlist =
3371                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
3372                       : NULL;
3373             for(i=0;i<symLC->nsyms;i++)
3374             {
3375                 if(nlist[i].n_type & N_STAB)
3376                     ;
3377                 else if(nlist[i].n_type & N_EXT)
3378                 {
3379                     if((nlist[i].n_type & N_TYPE) == N_UNDF
3380                         && (nlist[i].n_value == 0))
3381                     {
3382                         if(i < min)
3383                             min = i;
3384                         if(i > max)
3385                             max = i;
3386                     }
3387                 }
3388             }
3389             if(max >= min)
3390                 return ocAllocateJumpIslands(oc, max - min + 1, min);
3391
3392             break;
3393         }
3394
3395         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
3396     }
3397     return ocAllocateJumpIslands(oc,0,0);
3398 }
3399
3400 static int ocVerifyImage_MachO(ObjectCode* oc STG_UNUSED)
3401 {
3402     // FIXME: do some verifying here
3403     return 1;
3404 }
3405
3406 static int resolveImports(
3407     ObjectCode* oc,
3408     char *image,
3409     struct symtab_command *symLC,
3410     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
3411     unsigned long *indirectSyms,
3412     struct nlist *nlist)
3413 {
3414     unsigned i;
3415
3416     for(i=0;i*4<sect->size;i++)
3417     {
3418         // according to otool, reserved1 contains the first index into the indirect symbol table
3419         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
3420         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
3421         void *addr = NULL;
3422
3423         if((symbol->n_type & N_TYPE) == N_UNDF
3424             && (symbol->n_type & N_EXT) && (symbol->n_value != 0))
3425             addr = (void*) (symbol->n_value);
3426         else if((addr = lookupLocalSymbol(oc,nm)) != NULL)
3427             ;
3428         else
3429             addr = lookupSymbol(nm);
3430         if(!addr)
3431         {
3432             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
3433             return 0;
3434         }
3435         ASSERT(addr);
3436         checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
3437         ((void**)(image + sect->offset))[i] = addr;
3438     }
3439
3440     return 1;
3441 }
3442
3443 static unsigned long relocateAddress(
3444     ObjectCode* oc,
3445     int nSections,
3446     struct section* sections,
3447     unsigned long address)
3448 {
3449     int i;
3450     for(i = 0; i < nSections; i++)
3451     {
3452         if(sections[i].addr <= address
3453             && address < sections[i].addr + sections[i].size)
3454         {
3455             return (unsigned long)oc->image
3456                     + sections[i].offset + address - sections[i].addr;
3457         }
3458     }
3459     barf("Invalid Mach-O file:"
3460          "Address out of bounds while relocating object file");
3461     return 0;
3462 }
3463
3464 static int relocateSection(
3465     ObjectCode* oc,
3466     char *image,
3467     struct symtab_command *symLC, struct nlist *nlist,
3468     int nSections, struct section* sections, struct section *sect)
3469 {
3470     struct relocation_info *relocs;
3471     int i,n;
3472
3473     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
3474         return 1;
3475     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
3476         return 1;
3477
3478     n = sect->nreloc;
3479     relocs = (struct relocation_info*) (image + sect->reloff);
3480
3481     for(i=0;i<n;i++)
3482     {
3483         if(relocs[i].r_address & R_SCATTERED)
3484         {
3485             struct scattered_relocation_info *scat =
3486                 (struct scattered_relocation_info*) &relocs[i];
3487
3488             if(!scat->r_pcrel)
3489             {
3490                 if(scat->r_length == 2)
3491                 {
3492                     unsigned long word = 0;
3493                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
3494                     checkProddableBlock(oc,wordPtr);
3495
3496                     // Step 1: Figure out what the relocated value should be
3497                     if(scat->r_type == GENERIC_RELOC_VANILLA)
3498                     {
3499                         word = *wordPtr + (unsigned long) relocateAddress(
3500                                                                 oc,
3501                                                                 nSections,
3502                                                                 sections,
3503                                                                 scat->r_value)
3504                                         - scat->r_value;
3505                     }
3506                     else if(scat->r_type == PPC_RELOC_SECTDIFF
3507                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
3508                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
3509                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF)
3510                     {
3511                         struct scattered_relocation_info *pair =
3512                                 (struct scattered_relocation_info*) &relocs[i+1];
3513
3514                         if(!pair->r_scattered || pair->r_type != PPC_RELOC_PAIR)
3515                             barf("Invalid Mach-O file: "
3516                                  "PPC_RELOC_*_SECTDIFF not followed by PPC_RELOC_PAIR");
3517
3518                         word = (unsigned long)
3519                                (relocateAddress(oc, nSections, sections, scat->r_value)
3520                               - relocateAddress(oc, nSections, sections, pair->r_value));
3521                         i++;
3522                     }
3523                     else if(scat->r_type == PPC_RELOC_HI16
3524                          || scat->r_type == PPC_RELOC_LO16
3525                          || scat->r_type == PPC_RELOC_HA16
3526                          || scat->r_type == PPC_RELOC_LO14)
3527                     {   // these are generated by label+offset things
3528                         struct relocation_info *pair = &relocs[i+1];
3529                         if((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR)
3530                             barf("Invalid Mach-O file: "
3531                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
3532
3533                         if(scat->r_type == PPC_RELOC_LO16)
3534                         {
3535                             word = ((unsigned short*) wordPtr)[1];
3536                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
3537                         }
3538                         else if(scat->r_type == PPC_RELOC_LO14)
3539                         {
3540                             barf("Unsupported Relocation: PPC_RELOC_LO14");
3541                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
3542                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
3543                         }
3544                         else if(scat->r_type == PPC_RELOC_HI16)
3545                         {
3546                             word = ((unsigned short*) wordPtr)[1] << 16;
3547                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
3548                         }
3549                         else if(scat->r_type == PPC_RELOC_HA16)
3550                         {
3551                             word = ((unsigned short*) wordPtr)[1] << 16;
3552                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
3553                         }
3554
3555
3556                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
3557                                                 - scat->r_value;
3558
3559                         i++;
3560                     }
3561                     else
3562                         continue;  // ignore the others
3563
3564                     if(scat->r_type == GENERIC_RELOC_VANILLA
3565                         || scat->r_type == PPC_RELOC_SECTDIFF)
3566                     {
3567                         *wordPtr = word;
3568                     }
3569                     else if(scat->r_type == PPC_RELOC_LO16_SECTDIFF || scat->r_type == PPC_RELOC_LO16)
3570                     {
3571                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
3572                     }
3573                     else if(scat->r_type == PPC_RELOC_HI16_SECTDIFF || scat->r_type == PPC_RELOC_HI16)
3574                     {
3575                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
3576                     }
3577                     else if(scat->r_type == PPC_RELOC_HA16_SECTDIFF || scat->r_type == PPC_RELOC_HA16)
3578                     {
3579                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
3580                             + ((word & (1<<15)) ? 1 : 0);
3581                     }
3582                 }
3583             }
3584
3585             continue; // FIXME: I hope it's OK to ignore all the others.
3586         }
3587         else
3588         {
3589             struct relocation_info *reloc = &relocs[i];
3590             if(reloc->r_pcrel && !reloc->r_extern)
3591                 continue;
3592
3593             if(reloc->r_length == 2)
3594             {
3595                 unsigned long word = 0;
3596                 unsigned long jumpIsland = 0;
3597                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
3598                                                       // to avoid warning and to catch
3599                                                       // bugs.
3600
3601                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
3602                 checkProddableBlock(oc,wordPtr);
3603
3604                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
3605                 {
3606                     word = *wordPtr;
3607                 }
3608                 else if(reloc->r_type == PPC_RELOC_LO16)
3609                 {
3610                     word = ((unsigned short*) wordPtr)[1];
3611                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
3612                 }
3613                 else if(reloc->r_type == PPC_RELOC_HI16)
3614                 {
3615                     word = ((unsigned short*) wordPtr)[1] << 16;
3616                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
3617                 }
3618                 else if(reloc->r_type == PPC_RELOC_HA16)
3619                 {
3620                     word = ((unsigned short*) wordPtr)[1] << 16;
3621                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
3622                 }
3623                 else if(reloc->r_type == PPC_RELOC_BR24)
3624                 {
3625                     word = *wordPtr;
3626                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
3627                 }
3628
3629
3630                 if(!reloc->r_extern)
3631                 {
3632                     long delta =
3633                         sections[reloc->r_symbolnum-1].offset
3634                         - sections[reloc->r_symbolnum-1].addr
3635                         + ((long) image);
3636
3637                     word += delta;
3638                 }
3639                 else
3640                 {
3641                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
3642                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
3643                     void *symbolAddress = lookupSymbol(nm);
3644                     if(!symbolAddress)
3645                     {
3646                         errorBelch("\nunknown symbol `%s'", nm);
3647                         return 0;
3648                     }
3649
3650                     if(reloc->r_pcrel)
3651                     {
3652                             // In the .o file, this should be a relative jump to NULL
3653                             // and we'll change it to a jump to a relative jump to the symbol
3654                         ASSERT(-word == reloc->r_address);
3655                         word = (unsigned long) symbolAddress;
3656                         jumpIsland = makeJumpIsland(oc,reloc->r_symbolnum,word);
3657                         word -= ((long)image) + sect->offset + reloc->r_address;
3658                         if(jumpIsland != 0)
3659                         {
3660                             offsetToJumpIsland = jumpIsland
3661                                 - (((long)image) + sect->offset + reloc->r_address);
3662                         }
3663                     }
3664                     else
3665                     {
3666                         word += (unsigned long) symbolAddress;
3667                     }
3668                 }
3669
3670                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
3671                 {
3672                     *wordPtr = word;
3673                     continue;
3674                 }
3675                 else if(reloc->r_type == PPC_RELOC_LO16)
3676                 {
3677                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
3678                     i++; continue;
3679                 }
3680                 else if(reloc->r_type == PPC_RELOC_HI16)
3681                 {
3682                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
3683                     i++; continue;
3684                 }
3685                 else if(reloc->r_type == PPC_RELOC_HA16)
3686                 {
3687                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
3688                         + ((word & (1<<15)) ? 1 : 0);
3689                     i++; continue;
3690                 }
3691                 else if(reloc->r_type == PPC_RELOC_BR24)
3692                 {
3693                     if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
3694                     {
3695                         // The branch offset is too large.
3696                         // Therefore, we try to use a jump island.
3697                         if(jumpIsland == 0)
3698                         {
3699                             barf("unconditional relative branch out of range: "
3700                                  "no jump island available");
3701                         }
3702
3703                         word = offsetToJumpIsland;
3704                         if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
3705                             barf("unconditional relative branch out of range: "
3706                                  "jump island out of range");
3707                     }
3708                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
3709                     continue;
3710                 }
3711             }
3712             barf("\nunknown relocation %d",reloc->r_type);
3713             return 0;
3714         }
3715     }
3716     return 1;
3717 }
3718
3719 static int ocGetNames_MachO(ObjectCode* oc)
3720 {
3721     char *image = (char*) oc->image;
3722     struct mach_header *header = (struct mach_header*) image;
3723     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
3724     unsigned i,curSymbol = 0;
3725     struct segment_command *segLC = NULL;
3726     struct section *sections;
3727     struct symtab_command *symLC = NULL;
3728     struct nlist *nlist;
3729     unsigned long commonSize = 0;
3730     char    *commonStorage = NULL;
3731     unsigned long commonCounter;
3732
3733     for(i=0;i<header->ncmds;i++)
3734     {
3735         if(lc->cmd == LC_SEGMENT)
3736             segLC = (struct segment_command*) lc;
3737         else if(lc->cmd == LC_SYMTAB)
3738             symLC = (struct symtab_command*) lc;
3739         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
3740     }
3741
3742     sections = (struct section*) (segLC+1);
3743     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
3744                   : NULL;
3745
3746     for(i=0;i<segLC->nsects;i++)
3747     {
3748         if(sections[i].size == 0)
3749             continue;
3750
3751         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
3752         {
3753             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
3754                                       "ocGetNames_MachO(common symbols)");
3755             sections[i].offset = zeroFillArea - image;
3756         }
3757
3758         if(!strcmp(sections[i].sectname,"__text"))
3759             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
3760                 (void*) (image + sections[i].offset),
3761                 (void*) (image + sections[i].offset + sections[i].size));
3762         else if(!strcmp(sections[i].sectname,"__const"))
3763             addSection(oc, SECTIONKIND_RWDATA,
3764                 (void*) (image + sections[i].offset),
3765                 (void*) (image + sections[i].offset + sections[i].size));
3766         else if(!strcmp(sections[i].sectname,"__data"))
3767             addSection(oc, SECTIONKIND_RWDATA,
3768                 (void*) (image + sections[i].offset),
3769                 (void*) (image + sections[i].offset + sections[i].size));
3770         else if(!strcmp(sections[i].sectname,"__bss")
3771                 || !strcmp(sections[i].sectname,"__common"))
3772             addSection(oc, SECTIONKIND_RWDATA,
3773                 (void*) (image + sections[i].offset),
3774                 (void*) (image + sections[i].offset + sections[i].size));
3775
3776         addProddableBlock(oc, (void*) (image + sections[i].offset),
3777                                         sections[i].size);
3778     }
3779
3780         // count external symbols defined here
3781     oc->n_symbols = 0;
3782     if(symLC)
3783     {
3784         for(i=0;i<symLC->nsyms;i++)
3785         {
3786             if(nlist[i].n_type & N_STAB)
3787                 ;
3788             else if(nlist[i].n_type & N_EXT)
3789             {
3790                 if((nlist[i].n_type & N_TYPE) == N_UNDF
3791                     && (nlist[i].n_value != 0))
3792                 {
3793                     commonSize += nlist[i].n_value;
3794                     oc->n_symbols++;
3795                 }
3796                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
3797                     oc->n_symbols++;
3798             }
3799         }
3800     }
3801     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3802                                    "ocGetNames_MachO(oc->symbols)");
3803
3804     if(symLC)
3805     {
3806         for(i=0;i<symLC->nsyms;i++)
3807         {
3808             if(nlist[i].n_type & N_STAB)
3809                 ;
3810             else if((nlist[i].n_type & N_TYPE) == N_SECT)
3811             {
3812                 if(nlist[i].n_type & N_EXT)
3813                 {
3814                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
3815                     ghciInsertStrHashTable(oc->fileName, symhash, nm,
3816                                             image
3817                                             + sections[nlist[i].n_sect-1].offset
3818                                             - sections[nlist[i].n_sect-1].addr
3819                                             + nlist[i].n_value);
3820                     oc->symbols[curSymbol++] = nm;
3821                 }
3822                 else
3823                 {
3824                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
3825                     ghciInsertStrHashTable(oc->fileName, oc->lochash, nm,
3826                                             image
3827                                             + sections[nlist[i].n_sect-1].offset
3828                                             - sections[nlist[i].n_sect-1].addr
3829                                             + nlist[i].n_value);
3830                 }
3831             }
3832         }
3833     }
3834
3835     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
3836     commonCounter = (unsigned long)commonStorage;
3837     if(symLC)
3838     {
3839         for(i=0;i<symLC->nsyms;i++)
3840         {
3841             if((nlist[i].n_type & N_TYPE) == N_UNDF
3842                     && (nlist[i].n_type & N_EXT) && (nlist[i].n_value != 0))
3843             {
3844                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
3845                 unsigned long sz = nlist[i].n_value;
3846
3847                 nlist[i].n_value = commonCounter;
3848
3849                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
3850                                        (void*)commonCounter);
3851                 oc->symbols[curSymbol++] = nm;
3852
3853                 commonCounter += sz;
3854             }
3855         }
3856     }
3857     return 1;
3858 }
3859
3860 static int ocResolve_MachO(ObjectCode* oc)
3861 {
3862     char *image = (char*) oc->image;
3863     struct mach_header *header = (struct mach_header*) image;
3864     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
3865     unsigned i;
3866     struct segment_command *segLC = NULL;
3867     struct section *sections, *la_ptrs = NULL, *nl_ptrs = NULL;
3868     struct symtab_command *symLC = NULL;
3869     struct dysymtab_command *dsymLC = NULL;
3870     struct nlist *nlist;
3871
3872     for(i=0;i<header->ncmds;i++)
3873     {
3874         if(lc->cmd == LC_SEGMENT)
3875             segLC = (struct segment_command*) lc;
3876         else if(lc->cmd == LC_SYMTAB)
3877             symLC = (struct symtab_command*) lc;
3878         else if(lc->cmd == LC_DYSYMTAB)
3879             dsymLC = (struct dysymtab_command*) lc;
3880         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
3881     }
3882
3883     sections = (struct section*) (segLC+1);
3884     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
3885                   : NULL;
3886
3887     for(i=0;i<segLC->nsects;i++)
3888     {
3889         if(!strcmp(sections[i].sectname,"__la_symbol_ptr"))
3890             la_ptrs = &sections[i];
3891         else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr"))
3892             nl_ptrs = &sections[i];
3893     }
3894
3895     if(dsymLC)
3896     {
3897         unsigned long *indirectSyms
3898             = (unsigned long*) (image + dsymLC->indirectsymoff);
3899
3900         if(la_ptrs)
3901             if(!resolveImports(oc,image,symLC,la_ptrs,indirectSyms,nlist))
3902                 return 0;
3903         if(nl_ptrs)
3904             if(!resolveImports(oc,image,symLC,nl_ptrs,indirectSyms,nlist))
3905                 return 0;
3906     }
3907
3908     for(i=0;i<segLC->nsects;i++)
3909     {
3910         if(!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
3911             return 0;
3912     }
3913
3914     /* Free the local symbol table; we won't need it again. */
3915     freeHashTable(oc->lochash, NULL);
3916     oc->lochash = NULL;
3917
3918 #if defined (powerpc_HOST_ARCH)
3919     ocFlushInstructionCache( oc );
3920 #endif
3921
3922     return 1;
3923 }
3924
3925 /*
3926  * The Mach-O object format uses leading underscores. But not everywhere.
3927  * There is a small number of runtime support functions defined in
3928  * libcc_dynamic.a whose name does not have a leading underscore.
3929  * As a consequence, we can't get their address from C code.
3930  * We have to use inline assembler just to take the address of a function.
3931  * Yuck.
3932  */
3933
3934 static void machoInitSymbolsWithoutUnderscore()
3935 {
3936     extern void* symbolsWithoutUnderscore[];
3937     void **p = symbolsWithoutUnderscore;
3938     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
3939
3940 #undef Sym
3941 #define Sym(x)  \
3942     __asm__ volatile(".long " # x);
3943
3944     RTS_MACHO_NOUNDERLINE_SYMBOLS
3945
3946     __asm__ volatile(".text");
3947
3948 #undef Sym
3949 #define Sym(x)  \
3950     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
3951
3952     RTS_MACHO_NOUNDERLINE_SYMBOLS
3953
3954 #undef Sym
3955 }
3956 #endif