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(".rdata",sectab_i->Name)||
1967           0==strcmp(".rodata",sectab_i->Name))
1968          kind = SECTIONKIND_CODE_OR_RODATA;
1969       if (0==strcmp(".data",sectab_i->Name) ||
1970           0==strcmp(".bss",sectab_i->Name))
1971          kind = SECTIONKIND_RWDATA;
1972
1973       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
1974       sz = sectab_i->SizeOfRawData;
1975       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
1976
1977       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
1978       end   = start + sz - 1;
1979
1980       if (kind == SECTIONKIND_OTHER
1981           /* Ignore sections called which contain stabs debugging
1982              information. */
1983           && 0 != strcmp(".stab", sectab_i->Name)
1984           && 0 != strcmp(".stabstr", sectab_i->Name)
1985          ) {
1986          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", sectab_i->Name, oc->fileName);
1987          return 0;
1988       }
1989
1990       if (kind != SECTIONKIND_OTHER && end >= start) {
1991          addSection(oc, kind, start, end);
1992          addProddableBlock(oc, start, end - start + 1);
1993       }
1994    }
1995
1996    /* Copy exported symbols into the ObjectCode. */
1997
1998    oc->n_symbols = hdr->NumberOfSymbols;
1999    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
2000                                   "ocGetNames_PEi386(oc->symbols)");
2001    /* Call me paranoid; I don't care. */
2002    for (i = 0; i < oc->n_symbols; i++)
2003       oc->symbols[i] = NULL;
2004
2005    i = 0;
2006    while (1) {
2007       COFF_symbol* symtab_i;
2008       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2009       symtab_i = (COFF_symbol*)
2010                  myindex ( sizeof_COFF_symbol, symtab, i );
2011
2012       addr  = NULL;
2013
2014       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
2015           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
2016          /* This symbol is global and defined, viz, exported */
2017          /* for MYIMAGE_SYMCLASS_EXTERNAL
2018                 && !MYIMAGE_SYM_UNDEFINED,
2019             the address of the symbol is:
2020                 address of relevant section + offset in section
2021          */
2022          COFF_section* sectabent
2023             = (COFF_section*) myindex ( sizeof_COFF_section,
2024                                         sectab,
2025                                         symtab_i->SectionNumber-1 );
2026          addr = ((UChar*)(oc->image))
2027                 + (sectabent->PointerToRawData
2028                    + symtab_i->Value);
2029       }
2030       else
2031       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
2032           && symtab_i->Value > 0) {
2033          /* This symbol isn't in any section at all, ie, global bss.
2034             Allocate zeroed space for it. */
2035          addr = stgCallocBytes(1, symtab_i->Value,
2036                                "ocGetNames_PEi386(non-anonymous bss)");
2037          addSection(oc, SECTIONKIND_RWDATA, addr,
2038                         ((UChar*)addr) + symtab_i->Value - 1);
2039          addProddableBlock(oc, addr, symtab_i->Value);
2040          /* debugBelch("BSS      section at 0x%x\n", addr); */
2041       }
2042
2043       if (addr != NULL ) {
2044          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
2045          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
2046          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
2047          ASSERT(i >= 0 && i < oc->n_symbols);
2048          /* cstring_from_COFF_symbol_name always succeeds. */
2049          oc->symbols[i] = sname;
2050          ghciInsertStrHashTable(oc->fileName, symhash, sname, addr);
2051       } else {
2052 #        if 0
2053          debugBelch(
2054                    "IGNORING symbol %d\n"
2055                    "     name `",
2056                    i
2057                  );
2058          printName ( symtab_i->Name, strtab );
2059          debugBelch(
2060                    "'\n"
2061                    "    value 0x%x\n"
2062                    "   1+sec# %d\n"
2063                    "     type 0x%x\n"
2064                    "   sclass 0x%x\n"
2065                    "     nAux %d\n",
2066                    symtab_i->Value,
2067                    (Int32)(symtab_i->SectionNumber),
2068                    (UInt32)symtab_i->Type,
2069                    (UInt32)symtab_i->StorageClass,
2070                    (UInt32)symtab_i->NumberOfAuxSymbols
2071                  );
2072 #        endif
2073       }
2074
2075       i += symtab_i->NumberOfAuxSymbols;
2076       i++;
2077    }
2078
2079    return 1;
2080 }
2081
2082
2083 static int
2084 ocResolve_PEi386 ( ObjectCode* oc )
2085 {
2086    COFF_header*  hdr;
2087    COFF_section* sectab;
2088    COFF_symbol*  symtab;
2089    UChar*        strtab;
2090
2091    UInt32        A;
2092    UInt32        S;
2093    UInt32*       pP;
2094
2095    int i;
2096    UInt32 j, noRelocs;
2097
2098    /* ToDo: should be variable-sized?  But is at least safe in the
2099       sense of buffer-overrun-proof. */
2100    char symbol[1000];
2101    /* debugBelch("resolving for %s\n", oc->fileName); */
2102
2103    hdr = (COFF_header*)(oc->image);
2104    sectab = (COFF_section*) (
2105                ((UChar*)(oc->image))
2106                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2107             );
2108    symtab = (COFF_symbol*) (
2109                ((UChar*)(oc->image))
2110                + hdr->PointerToSymbolTable
2111             );
2112    strtab = ((UChar*)(oc->image))
2113             + hdr->PointerToSymbolTable
2114             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2115
2116    for (i = 0; i < hdr->NumberOfSections; i++) {
2117       COFF_section* sectab_i
2118          = (COFF_section*)
2119            myindex ( sizeof_COFF_section, sectab, i );
2120       COFF_reloc* reltab
2121          = (COFF_reloc*) (
2122               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2123            );
2124
2125       /* Ignore sections called which contain stabs debugging
2126          information. */
2127       if (0 == strcmp(".stab", sectab_i->Name)
2128           || 0 == strcmp(".stabstr", sectab_i->Name))
2129          continue;
2130
2131       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2132         /* If the relocation field (a short) has overflowed, the
2133          * real count can be found in the first reloc entry.
2134          *
2135          * See Section 4.1 (last para) of the PE spec (rev6.0).
2136          *
2137          * Nov2003 update: the GNU linker still doesn't correctly
2138          * handle the generation of relocatable object files with
2139          * overflown relocations. Hence the output to warn of potential
2140          * troubles.
2141          */
2142         COFF_reloc* rel = (COFF_reloc*)
2143                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2144         noRelocs = rel->VirtualAddress;
2145         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
2146                    noRelocs);
2147         j = 1;
2148       } else {
2149         noRelocs = sectab_i->NumberOfRelocations;
2150         j = 0;
2151       }
2152
2153
2154       for (; j < noRelocs; j++) {
2155          COFF_symbol* sym;
2156          COFF_reloc* reltab_j
2157             = (COFF_reloc*)
2158               myindex ( sizeof_COFF_reloc, reltab, j );
2159
2160          /* the location to patch */
2161          pP = (UInt32*)(
2162                  ((UChar*)(oc->image))
2163                  + (sectab_i->PointerToRawData
2164                     + reltab_j->VirtualAddress
2165                     - sectab_i->VirtualAddress )
2166               );
2167          /* the existing contents of pP */
2168          A = *pP;
2169          /* the symbol to connect to */
2170          sym = (COFF_symbol*)
2171                myindex ( sizeof_COFF_symbol,
2172                          symtab, reltab_j->SymbolTableIndex );
2173          IF_DEBUG(linker,
2174                   debugBelch(
2175                             "reloc sec %2d num %3d:  type 0x%-4x   "
2176                             "vaddr 0x%-8x   name `",
2177                             i, j,
2178                             (UInt32)reltab_j->Type,
2179                             reltab_j->VirtualAddress );
2180                             printName ( sym->Name, strtab );
2181                             debugBelch("'\n" ));
2182
2183          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
2184             COFF_section* section_sym
2185                = findPEi386SectionCalled ( oc, sym->Name );
2186             if (!section_sym) {
2187                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
2188                return 0;
2189             }
2190             S = ((UInt32)(oc->image))
2191                 + (section_sym->PointerToRawData
2192                    + sym->Value);
2193          } else {
2194             copyName ( sym->Name, strtab, symbol, 1000-1 );
2195             (void*)S = lookupLocalSymbol( oc, symbol );
2196             if ((void*)S != NULL) goto foundit;
2197             (void*)S = lookupSymbol( symbol );
2198             if ((void*)S != NULL) goto foundit;
2199             zapTrailingAtSign ( symbol );
2200             (void*)S = lookupLocalSymbol( oc, symbol );
2201             if ((void*)S != NULL) goto foundit;
2202             (void*)S = lookupSymbol( symbol );
2203             if ((void*)S != NULL) goto foundit;
2204             /* Newline first because the interactive linker has printed "linking..." */
2205             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, symbol);
2206             return 0;
2207            foundit:;
2208          }
2209          checkProddableBlock(oc, pP);
2210          switch (reltab_j->Type) {
2211             case MYIMAGE_REL_I386_DIR32:
2212                *pP = A + S;
2213                break;
2214             case MYIMAGE_REL_I386_REL32:
2215                /* Tricky.  We have to insert a displacement at
2216                   pP which, when added to the PC for the _next_
2217                   insn, gives the address of the target (S).
2218                   Problem is to know the address of the next insn
2219                   when we only know pP.  We assume that this
2220                   literal field is always the last in the insn,
2221                   so that the address of the next insn is pP+4
2222                   -- hence the constant 4.
2223                   Also I don't know if A should be added, but so
2224                   far it has always been zero.
2225                */
2226                ASSERT(A==0);
2227                *pP = S - ((UInt32)pP) - 4;
2228                break;
2229             default:
2230                debugBelch("%s: unhandled PEi386 relocation type %d",
2231                      oc->fileName, reltab_j->Type);
2232                return 0;
2233          }
2234
2235       }
2236    }
2237
2238    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
2239    return 1;
2240 }
2241
2242 #endif /* defined(OBJFORMAT_PEi386) */
2243
2244
2245 /* --------------------------------------------------------------------------
2246  * ELF specifics
2247  * ------------------------------------------------------------------------*/
2248
2249 #if defined(OBJFORMAT_ELF)
2250
2251 #define FALSE 0
2252 #define TRUE  1
2253
2254 #if defined(sparc_HOST_ARCH)
2255 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
2256 #elif defined(i386_HOST_ARCH)
2257 #  define ELF_TARGET_386    /* Used inside <elf.h> */
2258 #elif defined(x86_64_HOST_ARCH)
2259 #  define ELF_TARGET_X64_64
2260 #  define ELF_64BIT
2261 #elif defined (ia64_HOST_ARCH)
2262 #  define ELF_TARGET_IA64   /* Used inside <elf.h> */
2263 #  define ELF_64BIT
2264 #  define ELF_FUNCTION_DESC /* calling convention uses function descriptors */
2265 #  define ELF_NEED_GOT      /* needs Global Offset Table */
2266 #  define ELF_NEED_PLT      /* needs Procedure Linkage Tables */
2267 #endif
2268
2269 #if !defined(openbsd_HOST_OS)
2270 #include <elf.h>
2271 #else
2272 /* openbsd elf has things in different places, with diff names */
2273 #include <elf_abi.h>
2274 #include <machine/reloc.h>
2275 #define R_386_32    RELOC_32
2276 #define R_386_PC32  RELOC_PC32
2277 #endif
2278
2279 /*
2280  * Define a set of types which can be used for both ELF32 and ELF64
2281  */
2282
2283 #ifdef ELF_64BIT
2284 #define ELFCLASS    ELFCLASS64
2285 #define Elf_Addr    Elf64_Addr
2286 #define Elf_Word    Elf64_Word
2287 #define Elf_Sword   Elf64_Sword
2288 #define Elf_Ehdr    Elf64_Ehdr
2289 #define Elf_Phdr    Elf64_Phdr
2290 #define Elf_Shdr    Elf64_Shdr
2291 #define Elf_Sym     Elf64_Sym
2292 #define Elf_Rel     Elf64_Rel
2293 #define Elf_Rela    Elf64_Rela
2294 #define ELF_ST_TYPE ELF64_ST_TYPE
2295 #define ELF_ST_BIND ELF64_ST_BIND
2296 #define ELF_R_TYPE  ELF64_R_TYPE
2297 #define ELF_R_SYM   ELF64_R_SYM
2298 #else
2299 #define ELFCLASS    ELFCLASS32
2300 #define Elf_Addr    Elf32_Addr
2301 #define Elf_Word    Elf32_Word
2302 #define Elf_Sword   Elf32_Sword
2303 #define Elf_Ehdr    Elf32_Ehdr
2304 #define Elf_Phdr    Elf32_Phdr
2305 #define Elf_Shdr    Elf32_Shdr
2306 #define Elf_Sym     Elf32_Sym
2307 #define Elf_Rel     Elf32_Rel
2308 #define Elf_Rela    Elf32_Rela
2309 #ifndef ELF_ST_TYPE
2310 #define ELF_ST_TYPE ELF32_ST_TYPE
2311 #endif
2312 #ifndef ELF_ST_BIND
2313 #define ELF_ST_BIND ELF32_ST_BIND
2314 #endif
2315 #ifndef ELF_R_TYPE
2316 #define ELF_R_TYPE  ELF32_R_TYPE
2317 #endif
2318 #ifndef ELF_R_SYM
2319 #define ELF_R_SYM   ELF32_R_SYM
2320 #endif
2321 #endif
2322
2323
2324 /*
2325  * Functions to allocate entries in dynamic sections.  Currently we simply
2326  * preallocate a large number, and we don't check if a entry for the given
2327  * target already exists (a linear search is too slow).  Ideally these
2328  * entries would be associated with symbols.
2329  */
2330
2331 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
2332 #define GOT_SIZE            0x20000
2333 #define FUNCTION_TABLE_SIZE 0x10000
2334 #define PLT_SIZE            0x08000
2335
2336 #ifdef ELF_NEED_GOT
2337 static Elf_Addr got[GOT_SIZE];
2338 static unsigned int gotIndex;
2339 static Elf_Addr gp_val = (Elf_Addr)got;
2340
2341 static Elf_Addr
2342 allocateGOTEntry(Elf_Addr target)
2343 {
2344    Elf_Addr *entry;
2345
2346    if (gotIndex >= GOT_SIZE)
2347       barf("Global offset table overflow");
2348
2349    entry = &got[gotIndex++];
2350    *entry = target;
2351    return (Elf_Addr)entry;
2352 }
2353 #endif
2354
2355 #ifdef ELF_FUNCTION_DESC
2356 typedef struct {
2357    Elf_Addr ip;
2358    Elf_Addr gp;
2359 } FunctionDesc;
2360
2361 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
2362 static unsigned int functionTableIndex;
2363
2364 static Elf_Addr
2365 allocateFunctionDesc(Elf_Addr target)
2366 {
2367    FunctionDesc *entry;
2368
2369    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
2370       barf("Function table overflow");
2371
2372    entry = &functionTable[functionTableIndex++];
2373    entry->ip = target;
2374    entry->gp = (Elf_Addr)gp_val;
2375    return (Elf_Addr)entry;
2376 }
2377
2378 static Elf_Addr
2379 copyFunctionDesc(Elf_Addr target)
2380 {
2381    FunctionDesc *olddesc = (FunctionDesc *)target;
2382    FunctionDesc *newdesc;
2383
2384    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
2385    newdesc->gp = olddesc->gp;
2386    return (Elf_Addr)newdesc;
2387 }
2388 #endif
2389
2390 #ifdef ELF_NEED_PLT
2391 #ifdef ia64_HOST_ARCH
2392 static void ia64_reloc_gprel22(Elf_Addr target, Elf_Addr value);
2393 static void ia64_reloc_pcrel21(Elf_Addr target, Elf_Addr value, ObjectCode *oc);
2394
2395 static unsigned char plt_code[] =
2396 {
2397    /* taken from binutils bfd/elfxx-ia64.c */
2398    0x0b, 0x78, 0x00, 0x02, 0x00, 0x24,  /*   [MMI]       addl r15=0,r1;;    */
2399    0x00, 0x41, 0x3c, 0x30, 0x28, 0xc0,  /*               ld8 r16=[r15],8    */
2400    0x01, 0x08, 0x00, 0x84,              /*               mov r14=r1;;       */
2401    0x11, 0x08, 0x00, 0x1e, 0x18, 0x10,  /*   [MIB]       ld8 r1=[r15]       */
2402    0x60, 0x80, 0x04, 0x80, 0x03, 0x00,  /*               mov b6=r16         */
2403    0x60, 0x00, 0x80, 0x00               /*               br.few b6;;        */
2404 };
2405
2406 /* If we can't get to the function descriptor via gp, take a local copy of it */
2407 #define PLT_RELOC(code, target) { \
2408    Elf64_Sxword rel_value = target - gp_val; \
2409    if ((rel_value > 0x1fffff) || (rel_value < -0x1fffff)) \
2410       ia64_reloc_gprel22((Elf_Addr)code, copyFunctionDesc(target)); \
2411    else \
2412       ia64_reloc_gprel22((Elf_Addr)code, target); \
2413    }
2414 #endif
2415
2416 typedef struct {
2417    unsigned char code[sizeof(plt_code)];
2418 } PLTEntry;
2419
2420 static Elf_Addr
2421 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
2422 {
2423    PLTEntry *plt = (PLTEntry *)oc->plt;
2424    PLTEntry *entry;
2425
2426    if (oc->pltIndex >= PLT_SIZE)
2427       barf("Procedure table overflow");
2428
2429    entry = &plt[oc->pltIndex++];
2430    memcpy(entry->code, plt_code, sizeof(entry->code));
2431    PLT_RELOC(entry->code, target);
2432    return (Elf_Addr)entry;
2433 }
2434
2435 static unsigned int
2436 PLTSize(void)
2437 {
2438    return (PLT_SIZE * sizeof(PLTEntry));
2439 }
2440 #endif
2441
2442
2443 /*
2444  * Generic ELF functions
2445  */
2446
2447 static char *
2448 findElfSection ( void* objImage, Elf_Word sh_type )
2449 {
2450    char* ehdrC = (char*)objImage;
2451    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
2452    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
2453    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
2454    char* ptr = NULL;
2455    int i;
2456
2457    for (i = 0; i < ehdr->e_shnum; i++) {
2458       if (shdr[i].sh_type == sh_type
2459           /* Ignore the section header's string table. */
2460           && i != ehdr->e_shstrndx
2461           /* Ignore string tables named .stabstr, as they contain
2462              debugging info. */
2463           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
2464          ) {
2465          ptr = ehdrC + shdr[i].sh_offset;
2466          break;
2467       }
2468    }
2469    return ptr;
2470 }
2471
2472 #if defined(ia64_HOST_ARCH)
2473 static Elf_Addr
2474 findElfSegment ( void* objImage, Elf_Addr vaddr )
2475 {
2476    char* ehdrC = (char*)objImage;
2477    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
2478    Elf_Phdr* phdr = (Elf_Phdr*)(ehdrC + ehdr->e_phoff);
2479    Elf_Addr segaddr = 0;
2480    int i;
2481
2482    for (i = 0; i < ehdr->e_phnum; i++) {
2483       segaddr = phdr[i].p_vaddr;
2484       if ((vaddr >= segaddr) && (vaddr < segaddr + phdr[i].p_memsz))
2485               break;
2486    }
2487    return segaddr;
2488 }
2489 #endif
2490
2491 static int
2492 ocVerifyImage_ELF ( ObjectCode* oc )
2493 {
2494    Elf_Shdr* shdr;
2495    Elf_Sym*  stab;
2496    int i, j, nent, nstrtab, nsymtabs;
2497    char* sh_strtab;
2498    char* strtab;
2499
2500    char*     ehdrC = (char*)(oc->image);
2501    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
2502
2503    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
2504        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
2505        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
2506        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
2507       errorBelch("%s: not an ELF object", oc->fileName);
2508       return 0;
2509    }
2510
2511    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
2512       errorBelch("%s: unsupported ELF format", oc->fileName);
2513       return 0;
2514    }
2515
2516    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
2517        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
2518    } else
2519    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
2520        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
2521    } else {
2522        errorBelch("%s: unknown endiannness", oc->fileName);
2523        return 0;
2524    }
2525
2526    if (ehdr->e_type != ET_REL) {
2527       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
2528       return 0;
2529    }
2530    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
2531
2532    IF_DEBUG(linker,debugBelch( "Architecture is " ));
2533    switch (ehdr->e_machine) {
2534       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
2535       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
2536 #ifdef EM_IA_64
2537       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
2538 #endif
2539       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
2540 #ifdef EM_X86_64
2541       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
2542 #endif
2543       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
2544                      errorBelch("%s: unknown architecture", oc->fileName);
2545                      return 0;
2546    }
2547
2548    IF_DEBUG(linker,debugBelch(
2549              "\nSection header table: start %d, n_entries %d, ent_size %d\n",
2550              ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
2551
2552    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
2553
2554    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
2555
2556    if (ehdr->e_shstrndx == SHN_UNDEF) {
2557       errorBelch("%s: no section header string table", oc->fileName);
2558       return 0;
2559    } else {
2560       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
2561                           ehdr->e_shstrndx));
2562       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
2563    }
2564
2565    for (i = 0; i < ehdr->e_shnum; i++) {
2566       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
2567       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
2568       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
2569       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
2570       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
2571                ehdrC + shdr[i].sh_offset,
2572                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
2573
2574       if (shdr[i].sh_type == SHT_REL) {
2575           IF_DEBUG(linker,debugBelch("Rel  " ));
2576       } else if (shdr[i].sh_type == SHT_RELA) {
2577           IF_DEBUG(linker,debugBelch("RelA " ));
2578       } else {
2579           IF_DEBUG(linker,debugBelch("     "));
2580       }
2581       if (sh_strtab) {
2582           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
2583       }
2584    }
2585
2586    IF_DEBUG(linker,debugBelch( "\nString tables" ));
2587    strtab = NULL;
2588    nstrtab = 0;
2589    for (i = 0; i < ehdr->e_shnum; i++) {
2590       if (shdr[i].sh_type == SHT_STRTAB
2591           /* Ignore the section header's string table. */
2592           && i != ehdr->e_shstrndx
2593           /* Ignore string tables named .stabstr, as they contain
2594              debugging info. */
2595           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
2596          ) {
2597          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
2598          strtab = ehdrC + shdr[i].sh_offset;
2599          nstrtab++;
2600       }
2601    }
2602    if (nstrtab != 1) {
2603       errorBelch("%s: no string tables, or too many", oc->fileName);
2604       return 0;
2605    }
2606
2607    nsymtabs = 0;
2608    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
2609    for (i = 0; i < ehdr->e_shnum; i++) {
2610       if (shdr[i].sh_type != SHT_SYMTAB) continue;
2611       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
2612       nsymtabs++;
2613       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
2614       nent = shdr[i].sh_size / sizeof(Elf_Sym);
2615       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%d rem)\n",
2616                nent,
2617                shdr[i].sh_size % sizeof(Elf_Sym)
2618              ));
2619       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
2620          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
2621          return 0;
2622       }
2623       for (j = 0; j < nent; j++) {
2624          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
2625          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
2626                              (int)stab[j].st_shndx,
2627                              (int)stab[j].st_size,
2628                              (char*)stab[j].st_value ));
2629
2630          IF_DEBUG(linker,debugBelch("type=" ));
2631          switch (ELF_ST_TYPE(stab[j].st_info)) {
2632             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
2633             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
2634             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
2635             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
2636             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
2637             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
2638          }
2639          IF_DEBUG(linker,debugBelch("  " ));
2640
2641          IF_DEBUG(linker,debugBelch("bind=" ));
2642          switch (ELF_ST_BIND(stab[j].st_info)) {
2643             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
2644             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
2645             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
2646             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
2647          }
2648          IF_DEBUG(linker,debugBelch("  " ));
2649
2650          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
2651       }
2652    }
2653
2654    if (nsymtabs == 0) {
2655       errorBelch("%s: didn't find any symbol tables", oc->fileName);
2656       return 0;
2657    }
2658
2659    return 1;
2660 }
2661
2662 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
2663 {
2664     *is_bss = FALSE;
2665
2666     if (hdr->sh_type == SHT_PROGBITS
2667         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
2668         /* .text-style section */
2669         return SECTIONKIND_CODE_OR_RODATA;
2670     }
2671
2672     if (hdr->sh_type == SHT_PROGBITS
2673             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
2674             /* .data-style section */
2675             return SECTIONKIND_RWDATA;
2676     }
2677
2678     if (hdr->sh_type == SHT_PROGBITS
2679         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
2680         /* .rodata-style section */
2681         return SECTIONKIND_CODE_OR_RODATA;
2682     }
2683
2684     if (hdr->sh_type == SHT_NOBITS
2685         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
2686         /* .bss-style section */
2687         *is_bss = TRUE;
2688         return SECTIONKIND_RWDATA;
2689     }
2690
2691     return SECTIONKIND_OTHER;
2692 }
2693
2694
2695 static int
2696 ocGetNames_ELF ( ObjectCode* oc )
2697 {
2698    int i, j, k, nent;
2699    Elf_Sym* stab;
2700
2701    char*     ehdrC    = (char*)(oc->image);
2702    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
2703    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
2704    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
2705
2706    ASSERT(symhash != NULL);
2707
2708    if (!strtab) {
2709       errorBelch("%s: no strtab", oc->fileName);
2710       return 0;
2711    }
2712
2713    k = 0;
2714    for (i = 0; i < ehdr->e_shnum; i++) {
2715       /* Figure out what kind of section it is.  Logic derived from
2716          Figure 1.14 ("Special Sections") of the ELF document
2717          ("Portable Formats Specification, Version 1.1"). */
2718       int         is_bss = FALSE;
2719       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
2720
2721       if (is_bss && shdr[i].sh_size > 0) {
2722          /* This is a non-empty .bss section.  Allocate zeroed space for
2723             it, and set its .sh_offset field such that
2724             ehdrC + .sh_offset == addr_of_zeroed_space.  */
2725          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
2726                                        "ocGetNames_ELF(BSS)");
2727          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
2728          /*
2729          debugBelch("BSS section at 0x%x, size %d\n",
2730                          zspace, shdr[i].sh_size);
2731          */
2732       }
2733
2734       /* fill in the section info */
2735       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
2736          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
2737          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
2738                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
2739       }
2740
2741       if (shdr[i].sh_type != SHT_SYMTAB) continue;
2742
2743       /* copy stuff into this module's object symbol table */
2744       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
2745       nent = shdr[i].sh_size / sizeof(Elf_Sym);
2746
2747       oc->n_symbols = nent;
2748       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
2749                                    "ocGetNames_ELF(oc->symbols)");
2750
2751       for (j = 0; j < nent; j++) {
2752
2753          char  isLocal = FALSE; /* avoids uninit-var warning */
2754          char* ad      = NULL;
2755          char* nm      = strtab + stab[j].st_name;
2756          int   secno   = stab[j].st_shndx;
2757
2758          /* Figure out if we want to add it; if so, set ad to its
2759             address.  Otherwise leave ad == NULL. */
2760
2761          if (secno == SHN_COMMON) {
2762             isLocal = FALSE;
2763             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
2764             /*
2765             debugBelch("COMMON symbol, size %d name %s\n",
2766                             stab[j].st_size, nm);
2767             */
2768             /* Pointless to do addProddableBlock() for this area,
2769                since the linker should never poke around in it. */
2770          }
2771          else
2772          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
2773                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
2774               )
2775               /* and not an undefined symbol */
2776               && stab[j].st_shndx != SHN_UNDEF
2777               /* and not in a "special section" */
2778               && stab[j].st_shndx < SHN_LORESERVE
2779               &&
2780               /* and it's a not a section or string table or anything silly */
2781               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
2782                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
2783                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
2784               )
2785             ) {
2786             /* Section 0 is the undefined section, hence > and not >=. */
2787             ASSERT(secno > 0 && secno < ehdr->e_shnum);
2788             /*
2789             if (shdr[secno].sh_type == SHT_NOBITS) {
2790                debugBelch("   BSS symbol, size %d off %d name %s\n",
2791                                stab[j].st_size, stab[j].st_value, nm);
2792             }
2793             */
2794             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
2795             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
2796                isLocal = TRUE;
2797             } else {
2798 #ifdef ELF_FUNCTION_DESC
2799                /* dlsym() and the initialisation table both give us function
2800                 * descriptors, so to be consistent we store function descriptors
2801                 * in the symbol table */
2802                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
2803                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
2804 #endif
2805                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s",
2806                                       ad, oc->fileName, nm ));
2807                isLocal = FALSE;
2808             }
2809          }
2810
2811          /* And the decision is ... */
2812
2813          if (ad != NULL) {
2814             ASSERT(nm != NULL);
2815             oc->symbols[j] = nm;
2816             /* Acquire! */
2817             if (isLocal) {
2818                /* Ignore entirely. */
2819             } else {
2820                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
2821             }
2822          } else {
2823             /* Skip. */
2824             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
2825                                    strtab + stab[j].st_name ));
2826             /*
2827             debugBelch(
2828                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
2829                     (int)ELF_ST_BIND(stab[j].st_info),
2830                     (int)ELF_ST_TYPE(stab[j].st_info),
2831                     (int)stab[j].st_shndx,
2832                     strtab + stab[j].st_name
2833                    );
2834             */
2835             oc->symbols[j] = NULL;
2836          }
2837
2838       }
2839    }
2840
2841    return 1;
2842 }
2843
2844 /* Do ELF relocations which lack an explicit addend.  All x86-linux
2845    relocations appear to be of this form. */
2846 static int
2847 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
2848                          Elf_Shdr* shdr, int shnum,
2849                          Elf_Sym*  stab, char* strtab )
2850 {
2851    int j;
2852    char *symbol;
2853    Elf_Word* targ;
2854    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
2855    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
2856    int target_shndx = shdr[shnum].sh_info;
2857    int symtab_shndx = shdr[shnum].sh_link;
2858
2859    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
2860    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
2861    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
2862                           target_shndx, symtab_shndx ));
2863
2864    /* Skip sections that we're not interested in. */
2865    {
2866        int is_bss;
2867        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
2868        if (kind == SECTIONKIND_OTHER) {
2869            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
2870            return 1;
2871        }
2872    }
2873
2874    for (j = 0; j < nent; j++) {
2875       Elf_Addr offset = rtab[j].r_offset;
2876       Elf_Addr info   = rtab[j].r_info;
2877
2878       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
2879       Elf_Word* pP = (Elf_Word*)P;
2880       Elf_Addr  A  = *pP;
2881       Elf_Addr  S;
2882       void*     S_tmp;
2883       Elf_Addr  value;
2884
2885       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
2886                              j, (void*)offset, (void*)info ));
2887       if (!info) {
2888          IF_DEBUG(linker,debugBelch( " ZERO" ));
2889          S = 0;
2890       } else {
2891          Elf_Sym sym = stab[ELF_R_SYM(info)];
2892          /* First see if it is a local symbol. */
2893          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
2894             /* Yes, so we can get the address directly from the ELF symbol
2895                table. */
2896             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
2897             S = (Elf_Addr)
2898                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
2899                        + stab[ELF_R_SYM(info)].st_value);
2900
2901          } else {
2902             /* No, so look up the name in our global table. */
2903             symbol = strtab + sym.st_name;
2904             S_tmp = lookupSymbol( symbol );
2905             S = (Elf_Addr)S_tmp;
2906          }
2907          if (!S) {
2908             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
2909             return 0;
2910          }
2911          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
2912       }
2913
2914       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
2915                              (void*)P, (void*)S, (void*)A ));
2916       checkProddableBlock ( oc, pP );
2917
2918       value = S + A;
2919
2920       switch (ELF_R_TYPE(info)) {
2921 #        ifdef i386_HOST_ARCH
2922          case R_386_32:   *pP = value;     break;
2923          case R_386_PC32: *pP = value - P; break;
2924 #        endif
2925          default:
2926             errorBelch("%s: unhandled ELF relocation(Rel) type %d\n",
2927                   oc->fileName, ELF_R_TYPE(info));
2928             return 0;
2929       }
2930
2931    }
2932    return 1;
2933 }
2934
2935 /* Do ELF relocations for which explicit addends are supplied.
2936    sparc-solaris relocations appear to be of this form. */
2937 static int
2938 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
2939                           Elf_Shdr* shdr, int shnum,
2940                           Elf_Sym*  stab, char* strtab )
2941 {
2942    int j;
2943    char *symbol;
2944    Elf_Addr targ;
2945    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
2946    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
2947    int target_shndx = shdr[shnum].sh_info;
2948    int symtab_shndx = shdr[shnum].sh_link;
2949
2950    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
2951    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
2952    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
2953                           target_shndx, symtab_shndx ));
2954
2955    for (j = 0; j < nent; j++) {
2956 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
2957       /* This #ifdef only serves to avoid unused-var warnings. */
2958       Elf_Addr  offset = rtab[j].r_offset;
2959       Elf_Addr  P      = targ + offset;
2960 #endif
2961       Elf_Addr  info   = rtab[j].r_info;
2962       Elf_Addr  A      = rtab[j].r_addend;
2963       Elf_Addr  S;
2964       void*     S_tmp;
2965       Elf_Addr  value;
2966 #     if defined(sparc_HOST_ARCH)
2967       Elf_Word* pP = (Elf_Word*)P;
2968       Elf_Word  w1, w2;
2969 #     elif defined(ia64_HOST_ARCH)
2970       Elf64_Xword *pP = (Elf64_Xword *)P;
2971       Elf_Addr addr;
2972 #     elif defined(powerpc_HOST_ARCH)
2973       Elf_Sword delta;
2974 #     endif
2975
2976       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
2977                              j, (void*)offset, (void*)info,
2978                                 (void*)A ));
2979       if (!info) {
2980          IF_DEBUG(linker,debugBelch( " ZERO" ));
2981          S = 0;
2982       } else {
2983          Elf_Sym sym = stab[ELF_R_SYM(info)];
2984          /* First see if it is a local symbol. */
2985          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
2986             /* Yes, so we can get the address directly from the ELF symbol
2987                table. */
2988             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
2989             S = (Elf_Addr)
2990                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
2991                        + stab[ELF_R_SYM(info)].st_value);
2992 #ifdef ELF_FUNCTION_DESC
2993             /* Make a function descriptor for this function */
2994             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
2995                S = allocateFunctionDesc(S + A);
2996                A = 0;
2997             }
2998 #endif
2999          } else {
3000             /* No, so look up the name in our global table. */
3001             symbol = strtab + sym.st_name;
3002             S_tmp = lookupSymbol( symbol );
3003             S = (Elf_Addr)S_tmp;
3004
3005 #ifdef ELF_FUNCTION_DESC
3006             /* If a function, already a function descriptor - we would
3007                have to copy it to add an offset. */
3008             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
3009                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
3010 #endif
3011          }
3012          if (!S) {
3013            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3014            return 0;
3015          }
3016          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
3017       }
3018
3019       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
3020                                         (void*)P, (void*)S, (void*)A ));
3021       /* checkProddableBlock ( oc, (void*)P ); */
3022
3023       value = S + A;
3024
3025       switch (ELF_R_TYPE(info)) {
3026 #        if defined(sparc_HOST_ARCH)
3027          case R_SPARC_WDISP30:
3028             w1 = *pP & 0xC0000000;
3029             w2 = (Elf_Word)((value - P) >> 2);
3030             ASSERT((w2 & 0xC0000000) == 0);
3031             w1 |= w2;
3032             *pP = w1;
3033             break;
3034          case R_SPARC_HI22:
3035             w1 = *pP & 0xFFC00000;
3036             w2 = (Elf_Word)(value >> 10);
3037             ASSERT((w2 & 0xFFC00000) == 0);
3038             w1 |= w2;
3039             *pP = w1;
3040             break;
3041          case R_SPARC_LO10:
3042             w1 = *pP & ~0x3FF;
3043             w2 = (Elf_Word)(value & 0x3FF);
3044             ASSERT((w2 & ~0x3FF) == 0);
3045             w1 |= w2;
3046             *pP = w1;
3047             break;
3048          /* According to the Sun documentation:
3049             R_SPARC_UA32
3050             This relocation type resembles R_SPARC_32, except it refers to an
3051             unaligned word. That is, the word to be relocated must be treated
3052             as four separate bytes with arbitrary alignment, not as a word
3053             aligned according to the architecture requirements.
3054
3055             (JRS: which means that freeloading on the R_SPARC_32 case
3056             is probably wrong, but hey ...)
3057          */
3058          case R_SPARC_UA32:
3059          case R_SPARC_32:
3060             w2 = (Elf_Word)value;
3061             *pP = w2;
3062             break;
3063 #        elif defined(ia64_HOST_ARCH)
3064          case R_IA64_DIR64LSB:
3065          case R_IA64_FPTR64LSB:
3066             *pP = value;
3067             break;
3068          case R_IA64_PCREL64LSB:
3069             *pP = value - P;
3070             break;
3071          case R_IA64_SEGREL64LSB:
3072             addr = findElfSegment(ehdrC, value);
3073             *pP = value - addr;
3074             break;
3075          case R_IA64_GPREL22:
3076             ia64_reloc_gprel22(P, value);
3077             break;
3078          case R_IA64_LTOFF22:
3079          case R_IA64_LTOFF22X:
3080          case R_IA64_LTOFF_FPTR22:
3081             addr = allocateGOTEntry(value);
3082             ia64_reloc_gprel22(P, addr);
3083             break;
3084          case R_IA64_PCREL21B:
3085             ia64_reloc_pcrel21(P, S, oc);
3086             break;
3087          case R_IA64_LDXMOV:
3088             /* This goes with R_IA64_LTOFF22X and points to the load to
3089              * convert into a move.  We don't implement relaxation. */
3090             break;
3091 #        elif defined(powerpc_HOST_ARCH)
3092          case R_PPC_ADDR16_LO:
3093             *(Elf32_Half*) P = value;
3094             break;
3095
3096          case R_PPC_ADDR16_HI:
3097             *(Elf32_Half*) P = value >> 16;
3098             break;
3099
3100          case R_PPC_ADDR16_HA:
3101             *(Elf32_Half*) P = (value + 0x8000) >> 16;
3102             break;
3103
3104          case R_PPC_ADDR32:
3105             *(Elf32_Word *) P = value;
3106             break;
3107
3108          case R_PPC_REL32:
3109             *(Elf32_Word *) P = value - P;
3110             break;
3111
3112          case R_PPC_REL24:
3113             delta = value - P;
3114
3115             if( delta << 6 >> 6 != delta )
3116             {
3117                value = makeJumpIsland( oc, ELF_R_SYM(info), value );
3118                delta = value - P;
3119
3120                if( value == 0 || delta << 6 >> 6 != delta )
3121                {
3122                   barf( "Unable to make ppcJumpIsland for #%d",
3123                         ELF_R_SYM(info) );
3124                   return 0;
3125                }
3126             }
3127
3128             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
3129                                           | (delta & 0x3fffffc);
3130             break;
3131 #        endif
3132
3133 #if x86_64_HOST_OS
3134       case R_X86_64_64:
3135           *(Elf64_Xword *)P = value;
3136           break;
3137
3138       case R_X86_64_PC32:
3139           *(Elf64_Word *)P = (Elf64_Word) (value - P);
3140           break;
3141
3142       case R_X86_64_32:
3143           *(Elf64_Word *)P = (Elf64_Word)value;
3144           break;
3145
3146       case R_X86_64_32S:
3147           *(Elf64_Sword *)P = (Elf64_Sword)value;
3148           break;
3149 #endif
3150
3151          default:
3152             errorBelch("%s: unhandled ELF relocation(RelA) type %d\n",
3153                   oc->fileName, ELF_R_TYPE(info));
3154             return 0;
3155       }
3156
3157    }
3158    return 1;
3159 }
3160
3161 static int
3162 ocResolve_ELF ( ObjectCode* oc )
3163 {
3164    char *strtab;
3165    int   shnum, ok;
3166    Elf_Sym*  stab  = NULL;
3167    char*     ehdrC = (char*)(oc->image);
3168    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
3169    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3170
3171    /* first find "the" symbol table */
3172    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
3173
3174    /* also go find the string table */
3175    strtab = findElfSection ( ehdrC, SHT_STRTAB );
3176
3177    if (stab == NULL || strtab == NULL) {
3178       errorBelch("%s: can't find string or symbol table", oc->fileName);
3179       return 0;
3180    }
3181
3182    /* Process the relocation sections. */
3183    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
3184       if (shdr[shnum].sh_type == SHT_REL) {
3185          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
3186                                        shnum, stab, strtab );
3187          if (!ok) return ok;
3188       }
3189       else
3190       if (shdr[shnum].sh_type == SHT_RELA) {
3191          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
3192                                         shnum, stab, strtab );
3193          if (!ok) return ok;
3194       }
3195    }
3196
3197    /* Free the local symbol table; we won't need it again. */
3198    freeHashTable(oc->lochash, NULL);
3199    oc->lochash = NULL;
3200
3201 #if defined(powerpc_HOST_ARCH)
3202    ocFlushInstructionCache( oc );
3203 #endif
3204
3205    return 1;
3206 }
3207
3208 /*
3209  * IA64 specifics
3210  * Instructions are 41 bits long, packed into 128 bit bundles with a 5-bit template
3211  * at the front.  The following utility functions pack and unpack instructions, and
3212  * take care of the most common relocations.
3213  */
3214
3215 #ifdef ia64_HOST_ARCH
3216
3217 static Elf64_Xword
3218 ia64_extract_instruction(Elf64_Xword *target)
3219 {
3220    Elf64_Xword w1, w2;
3221    int slot = (Elf_Addr)target & 3;
3222    (Elf_Addr)target &= ~3;
3223
3224    w1 = *target;
3225    w2 = *(target+1);
3226
3227    switch (slot)
3228    {
3229       case 0:
3230          return ((w1 >> 5) & 0x1ffffffffff);
3231       case 1:
3232          return (w1 >> 46) | ((w2 & 0x7fffff) << 18);
3233       case 2:
3234          return (w2 >> 23);
3235       default:
3236          barf("ia64_extract_instruction: invalid slot %p", target);
3237    }
3238 }
3239
3240 static void
3241 ia64_deposit_instruction(Elf64_Xword *target, Elf64_Xword value)
3242 {
3243    int slot = (Elf_Addr)target & 3;
3244    (Elf_Addr)target &= ~3;
3245
3246    switch (slot)
3247    {
3248       case 0:
3249          *target |= value << 5;
3250          break;
3251       case 1:
3252          *target |= value << 46;
3253          *(target+1) |= value >> 18;
3254          break;
3255       case 2:
3256          *(target+1) |= value << 23;
3257          break;
3258    }
3259 }
3260
3261 static void
3262 ia64_reloc_gprel22(Elf_Addr target, Elf_Addr value)
3263 {
3264    Elf64_Xword instruction;
3265    Elf64_Sxword rel_value;
3266
3267    rel_value = value - gp_val;
3268    if ((rel_value > 0x1fffff) || (rel_value < -0x1fffff))
3269       barf("GP-relative data out of range (address = 0x%lx, gp = 0x%lx)", value, gp_val);
3270
3271    instruction = ia64_extract_instruction((Elf64_Xword *)target);
3272    instruction |= (((rel_value >> 0) & 0x07f) << 13)            /* imm7b */
3273                     | (((rel_value >> 7) & 0x1ff) << 27)        /* imm9d */
3274                     | (((rel_value >> 16) & 0x01f) << 22)       /* imm5c */
3275                     | ((Elf64_Xword)(rel_value < 0) << 36);     /* s */
3276    ia64_deposit_instruction((Elf64_Xword *)target, instruction);
3277 }
3278
3279 static void
3280 ia64_reloc_pcrel21(Elf_Addr target, Elf_Addr value, ObjectCode *oc)
3281 {
3282    Elf64_Xword instruction;
3283    Elf64_Sxword rel_value;
3284    Elf_Addr entry;
3285
3286    entry = allocatePLTEntry(value, oc);
3287
3288    rel_value = (entry >> 4) - (target >> 4);
3289    if ((rel_value > 0xfffff) || (rel_value < -0xfffff))
3290       barf("PLT entry too far away (entry = 0x%lx, target = 0x%lx)", entry, target);
3291
3292    instruction = ia64_extract_instruction((Elf64_Xword *)target);
3293    instruction |= ((rel_value & 0xfffff) << 13)                 /* imm20b */
3294                     | ((Elf64_Xword)(rel_value < 0) << 36);     /* s */
3295    ia64_deposit_instruction((Elf64_Xword *)target, instruction);
3296 }
3297
3298 #endif /* ia64 */
3299
3300 /*
3301  * PowerPC ELF specifics
3302  */
3303
3304 #ifdef powerpc_HOST_ARCH
3305
3306 static int ocAllocateJumpIslands_ELF( ObjectCode *oc )
3307 {
3308   Elf_Ehdr *ehdr;
3309   Elf_Shdr* shdr;
3310   int i;
3311
3312   ehdr = (Elf_Ehdr *) oc->image;
3313   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
3314
3315   for( i = 0; i < ehdr->e_shnum; i++ )
3316     if( shdr[i].sh_type == SHT_SYMTAB )
3317       break;
3318
3319   if( i == ehdr->e_shnum )
3320   {
3321     errorBelch( "This ELF file contains no symtab" );
3322     return 0;
3323   }
3324
3325   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
3326   {
3327     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
3328       shdr[i].sh_entsize, sizeof( Elf_Sym ) );
3329
3330     return 0;
3331   }
3332
3333   return ocAllocateJumpIslands( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
3334 }
3335
3336 #endif /* powerpc */
3337
3338 #endif /* ELF */
3339
3340 /* --------------------------------------------------------------------------
3341  * Mach-O specifics
3342  * ------------------------------------------------------------------------*/
3343
3344 #if defined(OBJFORMAT_MACHO)
3345
3346 /*
3347   Support for MachO linking on Darwin/MacOS X on PowerPC chips
3348   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
3349
3350   I hereby formally apologize for the hackish nature of this code.
3351   Things that need to be done:
3352   *) implement ocVerifyImage_MachO
3353   *) add still more sanity checks.
3354 */
3355
3356 static int ocAllocateJumpIslands_MachO(ObjectCode* oc)
3357 {
3358     struct mach_header *header = (struct mach_header *) oc->image;
3359     struct load_command *lc = (struct load_command *) (header + 1);
3360     unsigned i;
3361
3362     for( i = 0; i < header->ncmds; i++ )
3363     {
3364         if( lc->cmd == LC_SYMTAB )
3365         {
3366                 // Find out the first and last undefined external
3367                 // symbol, so we don't have to allocate too many
3368                 // jump islands.
3369             struct symtab_command *symLC = (struct symtab_command *) lc;
3370             unsigned min = symLC->nsyms, max = 0;
3371             struct nlist *nlist =
3372                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
3373                       : NULL;
3374             for(i=0;i<symLC->nsyms;i++)
3375             {
3376                 if(nlist[i].n_type & N_STAB)
3377                     ;
3378                 else if(nlist[i].n_type & N_EXT)
3379                 {
3380                     if((nlist[i].n_type & N_TYPE) == N_UNDF
3381                         && (nlist[i].n_value == 0))
3382                     {
3383                         if(i < min)
3384                             min = i;
3385                         if(i > max)
3386                             max = i;
3387                     }
3388                 }
3389             }
3390             if(max >= min)
3391                 return ocAllocateJumpIslands(oc, max - min + 1, min);
3392
3393             break;
3394         }
3395
3396         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
3397     }
3398     return ocAllocateJumpIslands(oc,0,0);
3399 }
3400
3401 static int ocVerifyImage_MachO(ObjectCode* oc STG_UNUSED)
3402 {
3403     // FIXME: do some verifying here
3404     return 1;
3405 }
3406
3407 static int resolveImports(
3408     ObjectCode* oc,
3409     char *image,
3410     struct symtab_command *symLC,
3411     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
3412     unsigned long *indirectSyms,
3413     struct nlist *nlist)
3414 {
3415     unsigned i;
3416
3417     for(i=0;i*4<sect->size;i++)
3418     {
3419         // according to otool, reserved1 contains the first index into the indirect symbol table
3420         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
3421         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
3422         void *addr = NULL;
3423
3424         if((symbol->n_type & N_TYPE) == N_UNDF
3425             && (symbol->n_type & N_EXT) && (symbol->n_value != 0))
3426             addr = (void*) (symbol->n_value);
3427         else if((addr = lookupLocalSymbol(oc,nm)) != NULL)
3428             ;
3429         else
3430             addr = lookupSymbol(nm);
3431         if(!addr)
3432         {
3433             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
3434             return 0;
3435         }
3436         ASSERT(addr);
3437         checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
3438         ((void**)(image + sect->offset))[i] = addr;
3439     }
3440
3441     return 1;
3442 }
3443
3444 static unsigned long relocateAddress(
3445     ObjectCode* oc,
3446     int nSections,
3447     struct section* sections,
3448     unsigned long address)
3449 {
3450     int i;
3451     for(i = 0; i < nSections; i++)
3452     {
3453         if(sections[i].addr <= address
3454             && address < sections[i].addr + sections[i].size)
3455         {
3456             return (unsigned long)oc->image
3457                     + sections[i].offset + address - sections[i].addr;
3458         }
3459     }
3460     barf("Invalid Mach-O file:"
3461          "Address out of bounds while relocating object file");
3462     return 0;
3463 }
3464
3465 static int relocateSection(
3466     ObjectCode* oc,
3467     char *image,
3468     struct symtab_command *symLC, struct nlist *nlist,
3469     int nSections, struct section* sections, struct section *sect)
3470 {
3471     struct relocation_info *relocs;
3472     int i,n;
3473
3474     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
3475         return 1;
3476     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
3477         return 1;
3478
3479     n = sect->nreloc;
3480     relocs = (struct relocation_info*) (image + sect->reloff);
3481
3482     for(i=0;i<n;i++)
3483     {
3484         if(relocs[i].r_address & R_SCATTERED)
3485         {
3486             struct scattered_relocation_info *scat =
3487                 (struct scattered_relocation_info*) &relocs[i];
3488
3489             if(!scat->r_pcrel)
3490             {
3491                 if(scat->r_length == 2)
3492                 {
3493                     unsigned long word = 0;
3494                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
3495                     checkProddableBlock(oc,wordPtr);
3496
3497                     // Step 1: Figure out what the relocated value should be
3498                     if(scat->r_type == GENERIC_RELOC_VANILLA)
3499                     {
3500                         word = *wordPtr + (unsigned long) relocateAddress(
3501                                                                 oc,
3502                                                                 nSections,
3503                                                                 sections,
3504                                                                 scat->r_value)
3505                                         - scat->r_value;
3506                     }
3507                     else if(scat->r_type == PPC_RELOC_SECTDIFF
3508                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
3509                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
3510                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF)
3511                     {
3512                         struct scattered_relocation_info *pair =
3513                                 (struct scattered_relocation_info*) &relocs[i+1];
3514
3515                         if(!pair->r_scattered || pair->r_type != PPC_RELOC_PAIR)
3516                             barf("Invalid Mach-O file: "
3517                                  "PPC_RELOC_*_SECTDIFF not followed by PPC_RELOC_PAIR");
3518
3519                         word = (unsigned long)
3520                                (relocateAddress(oc, nSections, sections, scat->r_value)
3521                               - relocateAddress(oc, nSections, sections, pair->r_value));
3522                         i++;
3523                     }
3524                     else if(scat->r_type == PPC_RELOC_HI16
3525                          || scat->r_type == PPC_RELOC_LO16
3526                          || scat->r_type == PPC_RELOC_HA16
3527                          || scat->r_type == PPC_RELOC_LO14)
3528                     {   // these are generated by label+offset things
3529                         struct relocation_info *pair = &relocs[i+1];
3530                         if((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR)
3531                             barf("Invalid Mach-O file: "
3532                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
3533
3534                         if(scat->r_type == PPC_RELOC_LO16)
3535                         {
3536                             word = ((unsigned short*) wordPtr)[1];
3537                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
3538                         }
3539                         else if(scat->r_type == PPC_RELOC_LO14)
3540                         {
3541                             barf("Unsupported Relocation: PPC_RELOC_LO14");
3542                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
3543                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
3544                         }
3545                         else if(scat->r_type == PPC_RELOC_HI16)
3546                         {
3547                             word = ((unsigned short*) wordPtr)[1] << 16;
3548                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
3549                         }
3550                         else if(scat->r_type == PPC_RELOC_HA16)
3551                         {
3552                             word = ((unsigned short*) wordPtr)[1] << 16;
3553                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
3554                         }
3555
3556
3557                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
3558                                                 - scat->r_value;
3559
3560                         i++;
3561                     }
3562                     else
3563                         continue;  // ignore the others
3564
3565                     if(scat->r_type == GENERIC_RELOC_VANILLA
3566                         || scat->r_type == PPC_RELOC_SECTDIFF)
3567                     {
3568                         *wordPtr = word;
3569                     }
3570                     else if(scat->r_type == PPC_RELOC_LO16_SECTDIFF || scat->r_type == PPC_RELOC_LO16)
3571                     {
3572                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
3573                     }
3574                     else if(scat->r_type == PPC_RELOC_HI16_SECTDIFF || scat->r_type == PPC_RELOC_HI16)
3575                     {
3576                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
3577                     }
3578                     else if(scat->r_type == PPC_RELOC_HA16_SECTDIFF || scat->r_type == PPC_RELOC_HA16)
3579                     {
3580                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
3581                             + ((word & (1<<15)) ? 1 : 0);
3582                     }
3583                 }
3584             }
3585
3586             continue; // FIXME: I hope it's OK to ignore all the others.
3587         }
3588         else
3589         {
3590             struct relocation_info *reloc = &relocs[i];
3591             if(reloc->r_pcrel && !reloc->r_extern)
3592                 continue;
3593
3594             if(reloc->r_length == 2)
3595             {
3596                 unsigned long word = 0;
3597                 unsigned long jumpIsland = 0;
3598                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
3599                                                       // to avoid warning and to catch
3600                                                       // bugs.
3601
3602                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
3603                 checkProddableBlock(oc,wordPtr);
3604
3605                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
3606                 {
3607                     word = *wordPtr;
3608                 }
3609                 else if(reloc->r_type == PPC_RELOC_LO16)
3610                 {
3611                     word = ((unsigned short*) wordPtr)[1];
3612                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
3613                 }
3614                 else if(reloc->r_type == PPC_RELOC_HI16)
3615                 {
3616                     word = ((unsigned short*) wordPtr)[1] << 16;
3617                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
3618                 }
3619                 else if(reloc->r_type == PPC_RELOC_HA16)
3620                 {
3621                     word = ((unsigned short*) wordPtr)[1] << 16;
3622                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
3623                 }
3624                 else if(reloc->r_type == PPC_RELOC_BR24)
3625                 {
3626                     word = *wordPtr;
3627                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
3628                 }
3629
3630
3631                 if(!reloc->r_extern)
3632                 {
3633                     long delta =
3634                         sections[reloc->r_symbolnum-1].offset
3635                         - sections[reloc->r_symbolnum-1].addr
3636                         + ((long) image);
3637
3638                     word += delta;
3639                 }
3640                 else
3641                 {
3642                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
3643                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
3644                     void *symbolAddress = lookupSymbol(nm);
3645                     if(!symbolAddress)
3646                     {
3647                         errorBelch("\nunknown symbol `%s'", nm);
3648                         return 0;
3649                     }
3650
3651                     if(reloc->r_pcrel)
3652                     {
3653                             // In the .o file, this should be a relative jump to NULL
3654                             // and we'll change it to a jump to a relative jump to the symbol
3655                         ASSERT(-word == reloc->r_address);
3656                         word = (unsigned long) symbolAddress;
3657                         jumpIsland = makeJumpIsland(oc,reloc->r_symbolnum,word);
3658                         word -= ((long)image) + sect->offset + reloc->r_address;
3659                         if(jumpIsland != 0)
3660                         {
3661                             offsetToJumpIsland = jumpIsland
3662                                 - (((long)image) + sect->offset + reloc->r_address);
3663                         }
3664                     }
3665                     else
3666                     {
3667                         word += (unsigned long) symbolAddress;
3668                     }
3669                 }
3670
3671                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
3672                 {
3673                     *wordPtr = word;
3674                     continue;
3675                 }
3676                 else if(reloc->r_type == PPC_RELOC_LO16)
3677                 {
3678                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
3679                     i++; continue;
3680                 }
3681                 else if(reloc->r_type == PPC_RELOC_HI16)
3682                 {
3683                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
3684                     i++; continue;
3685                 }
3686                 else if(reloc->r_type == PPC_RELOC_HA16)
3687                 {
3688                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
3689                         + ((word & (1<<15)) ? 1 : 0);
3690                     i++; continue;
3691                 }
3692                 else if(reloc->r_type == PPC_RELOC_BR24)
3693                 {
3694                     if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
3695                     {
3696                         // The branch offset is too large.
3697                         // Therefore, we try to use a jump island.
3698                         if(jumpIsland == 0)
3699                         {
3700                             barf("unconditional relative branch out of range: "
3701                                  "no jump island available");
3702                         }
3703
3704                         word = offsetToJumpIsland;
3705                         if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
3706                             barf("unconditional relative branch out of range: "
3707                                  "jump island out of range");
3708                     }
3709                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
3710                     continue;
3711                 }
3712             }
3713             barf("\nunknown relocation %d",reloc->r_type);
3714             return 0;
3715         }
3716     }
3717     return 1;
3718 }
3719
3720 static int ocGetNames_MachO(ObjectCode* oc)
3721 {
3722     char *image = (char*) oc->image;
3723     struct mach_header *header = (struct mach_header*) image;
3724     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
3725     unsigned i,curSymbol = 0;
3726     struct segment_command *segLC = NULL;
3727     struct section *sections;
3728     struct symtab_command *symLC = NULL;
3729     struct nlist *nlist;
3730     unsigned long commonSize = 0;
3731     char    *commonStorage = NULL;
3732     unsigned long commonCounter;
3733
3734     for(i=0;i<header->ncmds;i++)
3735     {
3736         if(lc->cmd == LC_SEGMENT)
3737             segLC = (struct segment_command*) lc;
3738         else if(lc->cmd == LC_SYMTAB)
3739             symLC = (struct symtab_command*) lc;
3740         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
3741     }
3742
3743     sections = (struct section*) (segLC+1);
3744     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
3745                   : NULL;
3746
3747     for(i=0;i<segLC->nsects;i++)
3748     {
3749         if(sections[i].size == 0)
3750             continue;
3751
3752         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
3753         {
3754             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
3755                                       "ocGetNames_MachO(common symbols)");
3756             sections[i].offset = zeroFillArea - image;
3757         }
3758
3759         if(!strcmp(sections[i].sectname,"__text"))
3760             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
3761                 (void*) (image + sections[i].offset),
3762                 (void*) (image + sections[i].offset + sections[i].size));
3763         else if(!strcmp(sections[i].sectname,"__const"))
3764             addSection(oc, SECTIONKIND_RWDATA,
3765                 (void*) (image + sections[i].offset),
3766                 (void*) (image + sections[i].offset + sections[i].size));
3767         else if(!strcmp(sections[i].sectname,"__data"))
3768             addSection(oc, SECTIONKIND_RWDATA,
3769                 (void*) (image + sections[i].offset),
3770                 (void*) (image + sections[i].offset + sections[i].size));
3771         else if(!strcmp(sections[i].sectname,"__bss")
3772                 || !strcmp(sections[i].sectname,"__common"))
3773             addSection(oc, SECTIONKIND_RWDATA,
3774                 (void*) (image + sections[i].offset),
3775                 (void*) (image + sections[i].offset + sections[i].size));
3776
3777         addProddableBlock(oc, (void*) (image + sections[i].offset),
3778                                         sections[i].size);
3779     }
3780
3781         // count external symbols defined here
3782     oc->n_symbols = 0;
3783     if(symLC)
3784     {
3785         for(i=0;i<symLC->nsyms;i++)
3786         {
3787             if(nlist[i].n_type & N_STAB)
3788                 ;
3789             else if(nlist[i].n_type & N_EXT)
3790             {
3791                 if((nlist[i].n_type & N_TYPE) == N_UNDF
3792                     && (nlist[i].n_value != 0))
3793                 {
3794                     commonSize += nlist[i].n_value;
3795                     oc->n_symbols++;
3796                 }
3797                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
3798                     oc->n_symbols++;
3799             }
3800         }
3801     }
3802     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3803                                    "ocGetNames_MachO(oc->symbols)");
3804
3805     if(symLC)
3806     {
3807         for(i=0;i<symLC->nsyms;i++)
3808         {
3809             if(nlist[i].n_type & N_STAB)
3810                 ;
3811             else if((nlist[i].n_type & N_TYPE) == N_SECT)
3812             {
3813                 if(nlist[i].n_type & N_EXT)
3814                 {
3815                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
3816                     ghciInsertStrHashTable(oc->fileName, symhash, nm,
3817                                             image
3818                                             + sections[nlist[i].n_sect-1].offset
3819                                             - sections[nlist[i].n_sect-1].addr
3820                                             + nlist[i].n_value);
3821                     oc->symbols[curSymbol++] = nm;
3822                 }
3823                 else
3824                 {
3825                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
3826                     ghciInsertStrHashTable(oc->fileName, oc->lochash, nm,
3827                                             image
3828                                             + sections[nlist[i].n_sect-1].offset
3829                                             - sections[nlist[i].n_sect-1].addr
3830                                             + nlist[i].n_value);
3831                 }
3832             }
3833         }
3834     }
3835
3836     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
3837     commonCounter = (unsigned long)commonStorage;
3838     if(symLC)
3839     {
3840         for(i=0;i<symLC->nsyms;i++)
3841         {
3842             if((nlist[i].n_type & N_TYPE) == N_UNDF
3843                     && (nlist[i].n_type & N_EXT) && (nlist[i].n_value != 0))
3844             {
3845                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
3846                 unsigned long sz = nlist[i].n_value;
3847
3848                 nlist[i].n_value = commonCounter;
3849
3850                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
3851                                        (void*)commonCounter);
3852                 oc->symbols[curSymbol++] = nm;
3853
3854                 commonCounter += sz;
3855             }
3856         }
3857     }
3858     return 1;
3859 }
3860
3861 static int ocResolve_MachO(ObjectCode* oc)
3862 {
3863     char *image = (char*) oc->image;
3864     struct mach_header *header = (struct mach_header*) image;
3865     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
3866     unsigned i;
3867     struct segment_command *segLC = NULL;
3868     struct section *sections, *la_ptrs = NULL, *nl_ptrs = NULL;
3869     struct symtab_command *symLC = NULL;
3870     struct dysymtab_command *dsymLC = NULL;
3871     struct nlist *nlist;
3872
3873     for(i=0;i<header->ncmds;i++)
3874     {
3875         if(lc->cmd == LC_SEGMENT)
3876             segLC = (struct segment_command*) lc;
3877         else if(lc->cmd == LC_SYMTAB)
3878             symLC = (struct symtab_command*) lc;
3879         else if(lc->cmd == LC_DYSYMTAB)
3880             dsymLC = (struct dysymtab_command*) lc;
3881         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
3882     }
3883
3884     sections = (struct section*) (segLC+1);
3885     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
3886                   : NULL;
3887
3888     for(i=0;i<segLC->nsects;i++)
3889     {
3890         if(!strcmp(sections[i].sectname,"__la_symbol_ptr"))
3891             la_ptrs = &sections[i];
3892         else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr"))
3893             nl_ptrs = &sections[i];
3894     }
3895
3896     if(dsymLC)
3897     {
3898         unsigned long *indirectSyms
3899             = (unsigned long*) (image + dsymLC->indirectsymoff);
3900
3901         if(la_ptrs)
3902             if(!resolveImports(oc,image,symLC,la_ptrs,indirectSyms,nlist))
3903                 return 0;
3904         if(nl_ptrs)
3905             if(!resolveImports(oc,image,symLC,nl_ptrs,indirectSyms,nlist))
3906                 return 0;
3907     }
3908
3909     for(i=0;i<segLC->nsects;i++)
3910     {
3911         if(!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
3912             return 0;
3913     }
3914
3915     /* Free the local symbol table; we won't need it again. */
3916     freeHashTable(oc->lochash, NULL);
3917     oc->lochash = NULL;
3918
3919 #if defined (powerpc_HOST_ARCH)
3920     ocFlushInstructionCache( oc );
3921 #endif
3922
3923     return 1;
3924 }
3925
3926 /*
3927  * The Mach-O object format uses leading underscores. But not everywhere.
3928  * There is a small number of runtime support functions defined in
3929  * libcc_dynamic.a whose name does not have a leading underscore.
3930  * As a consequence, we can't get their address from C code.
3931  * We have to use inline assembler just to take the address of a function.
3932  * Yuck.
3933  */
3934
3935 static void machoInitSymbolsWithoutUnderscore()
3936 {
3937     extern void* symbolsWithoutUnderscore[];
3938     void **p = symbolsWithoutUnderscore;
3939     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
3940
3941 #undef Sym
3942 #define Sym(x)  \
3943     __asm__ volatile(".long " # x);
3944
3945     RTS_MACHO_NOUNDERLINE_SYMBOLS
3946
3947     __asm__ volatile(".text");
3948
3949 #undef Sym
3950 #define Sym(x)  \
3951     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
3952
3953     RTS_MACHO_NOUNDERLINE_SYMBOLS
3954
3955 #undef Sym
3956 }
3957 #endif