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 "HsFFI.h"
  22
  23 #include "sm/Storage.h"
  24 #include "Stats.h"
  25 #include "Hash.h"
  26 #include "LinkerInternals.h"
  27 #include "RtsUtils.h"
  28 #include "Trace.h"
  29 #include "StgPrimFloat.h" // for __int_encodeFloat etc.
  30 #include "Stable.h"
  31
  32 #if !defined(mingw32_HOST_OS)
  33 #include "posix/Signals.h"
  34 #endif
  35
  36 // get protos for is*()
  37 #include <ctype.h>
  38
  39 #ifdef HAVE_SYS_TYPES_H
  40 #include <sys/types.h>
  41 #endif
  42
  43 #include <stdlib.h>
  44 #include <string.h>
  45 #include <stdio.h>
  46 #include <assert.h>
  47
  48 #ifdef HAVE_SYS_STAT_H
  49 #include <sys/stat.h>
  50 #endif
  51
  52 #if defined(HAVE_DLFCN_H)
  53 #include <dlfcn.h>
  54 #endif
  55
  56 #if defined(cygwin32_HOST_OS)
  57 #ifdef HAVE_DIRENT_H
  58 #include <dirent.h>
  59 #endif
  60
  61 #ifdef HAVE_SYS_TIME_H
  62 #include <sys/time.h>
  63 #endif
  64 #include <regex.h>
  65 #include <sys/fcntl.h>
  66 #include <sys/termios.h>
  67 #include <sys/utime.h>
  68 #include <sys/utsname.h>
  69 #include <sys/wait.h>
  70 #endif
  71
  72 #if defined(linux_HOST_OS) || defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS) || defined(darwin_HOST_OS)
  73 #define USE_MMAP
  74 #include <fcntl.h>
  75 #include <sys/mman.h>
  76
  77 #ifdef HAVE_UNISTD_H
  78 #include <unistd.h>
  79 #endif
  80
  81 #endif
  82
  83 #if defined(linux_HOST_OS) || defined(solaris2_HOST_OS) || defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  84 #  define OBJFORMAT_ELF
  85 #  include <regex.h>    // regex is already used by dlopen() so this is OK
  86                         // to use here without requiring an additional lib
  87 #elif defined(cygwin32_HOST_OS) || defined (mingw32_HOST_OS)
  88 #  define OBJFORMAT_PEi386
  89 #  include <windows.h>
  90 #  include <math.h>
  91 #elif defined(darwin_HOST_OS)
  92 #  define OBJFORMAT_MACHO
  93 #  include <regex.h>
  94 #  include <mach-o/loader.h>
  95 #  include <mach-o/nlist.h>
  96 #  include <mach-o/reloc.h>
  97 #if !defined(HAVE_DLFCN_H)
  98 #  include <mach-o/dyld.h>
  99 #endif
 100 #if defined(powerpc_HOST_ARCH)
 101 #  include <mach-o/ppc/reloc.h>
 102 #endif
 103 #if defined(x86_64_HOST_ARCH)
 104 #  include <mach-o/x86_64/reloc.h>
 105 #endif
 106 #endif
 107
 108 #if defined(x86_64_HOST_ARCH) && defined(darwin_HOST_OS)
 109 #define ALWAYS_PIC
 110 #endif
 111
 112 /* Hash table mapping symbol names to Symbol */
 113 static /*Str*/HashTable *symhash;
 114
 115 /* Hash table mapping symbol names to StgStablePtr */
 116 static /*Str*/HashTable *stablehash;
 117
 118 /* List of currently loaded objects */
 119 ObjectCode *objects = NULL;     /* initially empty */
 120
 121 static HsInt loadOc( ObjectCode* oc );
 122 static ObjectCode* mkOc( char *path, char *image, int imageSize,
 123                          char *archiveMemberName
 124 #ifndef USE_MMAP
 125 #ifdef darwin_HOST_OS
 126                        , int misalignment
 127 #endif
 128 #endif
 129                        );
 130
 131 #if defined(OBJFORMAT_ELF)
 132 static int ocVerifyImage_ELF    ( ObjectCode* oc );
 133 static int ocGetNames_ELF       ( ObjectCode* oc );
 134 static int ocResolve_ELF        ( ObjectCode* oc );
 135 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 136 static int ocAllocateSymbolExtras_ELF ( ObjectCode* oc );
 137 #endif
 138 #elif defined(OBJFORMAT_PEi386)
 139 static int ocVerifyImage_PEi386 ( ObjectCode* oc );
 140 static int ocGetNames_PEi386    ( ObjectCode* oc );
 141 static int ocResolve_PEi386     ( ObjectCode* oc );
 142 static void *lookupSymbolInDLLs ( unsigned char *lbl );
 143 static void zapTrailingAtSign   ( unsigned char *sym );
 144 #elif defined(OBJFORMAT_MACHO)
 145 static int ocVerifyImage_MachO    ( ObjectCode* oc );
 146 static int ocGetNames_MachO       ( ObjectCode* oc );
 147 static int ocResolve_MachO        ( ObjectCode* oc );
 148
 149 #ifndef USE_MMAP
 150 static int machoGetMisalignment( FILE * );
 151 #endif
 152 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 153 static int ocAllocateSymbolExtras_MachO ( ObjectCode* oc );
 154 #endif
 155 #ifdef powerpc_HOST_ARCH
 156 static void machoInitSymbolsWithoutUnderscore( void );
 157 #endif
 158 #endif
 159
 160 /* on x86_64 we have a problem with relocating symbol references in
 161  * code that was compiled without -fPIC.  By default, the small memory
 162  * model is used, which assumes that symbol references can fit in a
 163  * 32-bit slot.  The system dynamic linker makes this work for
 164  * references to shared libraries by either (a) allocating a jump
 165  * table slot for code references, or (b) moving the symbol at load
 166  * time (and copying its contents, if necessary) for data references.
 167  *
 168  * We unfortunately can't tell whether symbol references are to code
 169  * or data.  So for now we assume they are code (the vast majority
 170  * are), and allocate jump-table slots.  Unfortunately this will
 171  * SILENTLY generate crashing code for data references.  This hack is
 172  * enabled by X86_64_ELF_NONPIC_HACK.
 173  *
 174  * One workaround is to use shared Haskell libraries.  This is
 175  * coming.  Another workaround is to keep the static libraries but
 176  * compile them with -fPIC, because that will generate PIC references
 177  * to data which can be relocated.  The PIC code is still too green to
 178  * do this systematically, though.
 179  *
 180  * See bug #781
 181  * See thread http://www.haskell.org/pipermail/cvs-ghc/2007-September/038458.html
 182  *
 183  * Naming Scheme for Symbol Macros
 184  *
 185  * SymI_*: symbol is internal to the RTS. It resides in an object
 186  *         file/library that is statically.
 187  * SymE_*: symbol is external to the RTS library. It might be linked
 188  *         dynamically.
 189  *
 190  * Sym*_HasProto  : the symbol prototype is imported in an include file
 191  *                  or defined explicitly
 192  * Sym*_NeedsProto: the symbol is undefined and we add a dummy
 193  *                  default proto extern void sym(void);
 194  */
 195 #define X86_64_ELF_NONPIC_HACK 1
 196
 197 /* Link objects into the lower 2Gb on x86_64.  GHC assumes the
 198  * small memory model on this architecture (see gcc docs,
 199  * -mcmodel=small).
 200  *
 201  * MAP_32BIT not available on OpenBSD/amd64
 202  */
 203 #if defined(x86_64_HOST_ARCH) && defined(MAP_32BIT)
 204 #define TRY_MAP_32BIT MAP_32BIT
 205 #else
 206 #define TRY_MAP_32BIT 0
 207 #endif
 208
 209 /*
 210  * Due to the small memory model (see above), on x86_64 we have to map
 211  * all our non-PIC object files into the low 2Gb of the address space
 212  * (why 2Gb and not 4Gb?  Because all addresses must be reachable
 213  * using a 32-bit signed PC-relative offset). On Linux we can do this
 214  * using the MAP_32BIT flag to mmap(), however on other OSs
 215  * (e.g. *BSD, see #2063, and also on Linux inside Xen, see #2512), we
 216  * can't do this.  So on these systems, we have to pick a base address
 217  * in the low 2Gb of the address space and try to allocate memory from
 218  * there.
 219  *
 220  * We pick a default address based on the OS, but also make this
 221  * configurable via an RTS flag (+RTS -xm)
 222  */
 223 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
 224
 225 #if defined(MAP_32BIT)
 226 // Try to use MAP_32BIT
 227 #define MMAP_32BIT_BASE_DEFAULT 0
 228 #else
 229 // A guess: 1Gb.
 230 #define MMAP_32BIT_BASE_DEFAULT 0x40000000
 231 #endif
 232
 233 static void *mmap_32bit_base = (void *)MMAP_32BIT_BASE_DEFAULT;
 234 #endif
 235
 236 /* MAP_ANONYMOUS is MAP_ANON on some systems, e.g. OpenBSD */
 237 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
 238 #define MAP_ANONYMOUS MAP_ANON
 239 #endif
 240
 241 /* -----------------------------------------------------------------------------
 242  * Built-in symbols from the RTS
 243  */
 244
 245 typedef struct _RtsSymbolVal {
 246     char   *lbl;
 247     void   *addr;
 248 } RtsSymbolVal;
 249
 250 #define Maybe_Stable_Names      SymI_HasProto(stg_mkWeakzh)                     \
 251                                 SymI_HasProto(stg_mkWeakForeignEnvzh)           \
 252                                 SymI_HasProto(stg_makeStableNamezh)             \
 253                                 SymI_HasProto(stg_finalizzeWeakzh)
 254
 255 #if !defined (mingw32_HOST_OS)
 256 #define RTS_POSIX_ONLY_SYMBOLS                  \
 257       SymI_HasProto(__hscore_get_saved_termios) \
 258       SymI_HasProto(__hscore_set_saved_termios) \
 259       SymI_HasProto(shutdownHaskellAndSignal)   \
 260       SymI_HasProto(lockFile)                   \
 261       SymI_HasProto(unlockFile)                 \
 262       SymI_HasProto(signal_handlers)            \
 263       SymI_HasProto(stg_sig_install)            \
 264       SymI_NeedsProto(nocldstop)
 265 #endif
 266
 267 #if defined (cygwin32_HOST_OS)
 268 #define RTS_MINGW_ONLY_SYMBOLS /**/
 269 /* Don't have the ability to read import libs / archives, so
 270  * we have to stupidly list a lot of what libcygwin.a
 271  * exports; sigh.
 272  */
 273 #define RTS_CYGWIN_ONLY_SYMBOLS                          \
 274       SymI_HasProto(regfree)                             \
 275       SymI_HasProto(regexec)                             \
 276       SymI_HasProto(regerror)                            \
 277       SymI_HasProto(regcomp)                             \
 278       SymI_HasProto(__errno)                             \
 279       SymI_HasProto(access)                              \
 280       SymI_HasProto(chmod)                               \
 281       SymI_HasProto(chdir)                               \
 282       SymI_HasProto(close)                               \
 283       SymI_HasProto(creat)                               \
 284       SymI_HasProto(dup)                                 \
 285       SymI_HasProto(dup2)                                \
 286       SymI_HasProto(fstat)                               \
 287       SymI_HasProto(fcntl)                               \
 288       SymI_HasProto(getcwd)                              \
 289       SymI_HasProto(getenv)                              \
 290       SymI_HasProto(lseek)                               \
 291       SymI_HasProto(open)                                \
 292       SymI_HasProto(fpathconf)                           \
 293       SymI_HasProto(pathconf)                            \
 294       SymI_HasProto(stat)                                \
 295       SymI_HasProto(pow)                                 \
 296       SymI_HasProto(tanh)                                \
 297       SymI_HasProto(cosh)                                \
 298       SymI_HasProto(sinh)                                \
 299       SymI_HasProto(atan)                                \
 300       SymI_HasProto(acos)                                \
 301       SymI_HasProto(asin)                                \
 302       SymI_HasProto(tan)                                 \
 303       SymI_HasProto(cos)                                 \
 304       SymI_HasProto(sin)                                 \
 305       SymI_HasProto(exp)                                 \
 306       SymI_HasProto(log)                                 \
 307       SymI_HasProto(sqrt)                                \
 308       SymI_HasProto(localtime_r)                         \
 309       SymI_HasProto(gmtime_r)                            \
 310       SymI_HasProto(mktime)                              \
 311       SymI_NeedsProto(_imp___tzname)                     \
 312       SymI_HasProto(gettimeofday)                        \
 313       SymI_HasProto(timezone)                            \
 314       SymI_HasProto(tcgetattr)                           \
 315       SymI_HasProto(tcsetattr)                           \
 316       SymI_HasProto(memcpy)                              \
 317       SymI_HasProto(memmove)                             \
 318       SymI_HasProto(realloc)                             \
 319       SymI_HasProto(malloc)                              \
 320       SymI_HasProto(free)                                \
 321       SymI_HasProto(fork)                                \
 322       SymI_HasProto(lstat)                               \
 323       SymI_HasProto(isatty)                              \
 324       SymI_HasProto(mkdir)                               \
 325       SymI_HasProto(opendir)                             \
 326       SymI_HasProto(readdir)                             \
 327       SymI_HasProto(rewinddir)                           \
 328       SymI_HasProto(closedir)                            \
 329       SymI_HasProto(link)                                \
 330       SymI_HasProto(mkfifo)                              \
 331       SymI_HasProto(pipe)                                \
 332       SymI_HasProto(read)                                \
 333       SymI_HasProto(rename)                              \
 334       SymI_HasProto(rmdir)                               \
 335       SymI_HasProto(select)                              \
 336       SymI_HasProto(system)                              \
 337       SymI_HasProto(write)                               \
 338       SymI_HasProto(strcmp)                              \
 339       SymI_HasProto(strcpy)                              \
 340       SymI_HasProto(strncpy)                             \
 341       SymI_HasProto(strerror)                            \
 342       SymI_HasProto(sigaddset)                           \
 343       SymI_HasProto(sigemptyset)                         \
 344       SymI_HasProto(sigprocmask)                         \
 345       SymI_HasProto(umask)                               \
 346       SymI_HasProto(uname)                               \
 347       SymI_HasProto(unlink)                              \
 348       SymI_HasProto(utime)                               \
 349       SymI_HasProto(waitpid)
 350
 351 #elif !defined(mingw32_HOST_OS)
 352 #define RTS_MINGW_ONLY_SYMBOLS /**/
 353 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 354 #else /* defined(mingw32_HOST_OS) */
 355 #define RTS_POSIX_ONLY_SYMBOLS  /**/
 356 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 357
 358 #if HAVE_GETTIMEOFDAY
 359 #define RTS_MINGW_GETTIMEOFDAY_SYM SymI_NeedsProto(gettimeofday)
 360 #else
 361 #define RTS_MINGW_GETTIMEOFDAY_SYM /**/
 362 #endif
 363
 364 #if HAVE___MINGW_VFPRINTF
 365 #define RTS___MINGW_VFPRINTF_SYM SymI_HasProto(__mingw_vfprintf)
 366 #else
 367 #define RTS___MINGW_VFPRINTF_SYM /**/
 368 #endif
 369
 370 /* These are statically linked from the mingw libraries into the ghc
 371    executable, so we have to employ this hack. */
 372 #define RTS_MINGW_ONLY_SYMBOLS                           \
 373       SymI_HasProto(stg_asyncReadzh)                     \
 374       SymI_HasProto(stg_asyncWritezh)                    \
 375       SymI_HasProto(stg_asyncDoProczh)                   \
 376       SymI_HasProto(memset)                              \
 377       SymI_HasProto(inet_ntoa)                           \
 378       SymI_HasProto(inet_addr)                           \
 379       SymI_HasProto(htonl)                               \
 380       SymI_HasProto(recvfrom)                            \
 381       SymI_HasProto(listen)                              \
 382       SymI_HasProto(bind)                                \
 383       SymI_HasProto(shutdown)                            \
 384       SymI_HasProto(connect)                             \
 385       SymI_HasProto(htons)                               \
 386       SymI_HasProto(ntohs)                               \
 387       SymI_HasProto(getservbyname)                       \
 388       SymI_HasProto(getservbyport)                       \
 389       SymI_HasProto(getprotobynumber)                    \
 390       SymI_HasProto(getprotobyname)                      \
 391       SymI_HasProto(gethostbyname)                       \
 392       SymI_HasProto(gethostbyaddr)                       \
 393       SymI_HasProto(gethostname)                         \
 394       SymI_HasProto(strcpy)                              \
 395       SymI_HasProto(strncpy)                             \
 396       SymI_HasProto(abort)                               \
 397       SymI_NeedsProto(_alloca)                           \
 398       SymI_HasProto(isxdigit)                          \
 399       SymI_HasProto(isupper)                           \
 400       SymI_HasProto(ispunct)                           \
 401       SymI_HasProto(islower)                           \
 402       SymI_HasProto(isspace)                           \
 403       SymI_HasProto(isprint)                           \
 404       SymI_HasProto(isdigit)                           \
 405       SymI_HasProto(iscntrl)                           \
 406       SymI_HasProto(isalpha)                           \
 407       SymI_HasProto(isalnum)                           \
 408       SymI_HasProto(isascii)                           \
 409       RTS___MINGW_VFPRINTF_SYM                           \
 410       SymI_HasProto(strcmp)                              \
 411       SymI_HasProto(memmove)                             \
 412       SymI_HasProto(realloc)                             \
 413       SymI_HasProto(malloc)                              \
 414       SymI_HasProto(pow)                                 \
 415       SymI_HasProto(tanh)                                \
 416       SymI_HasProto(cosh)                                \
 417       SymI_HasProto(sinh)                                \
 418       SymI_HasProto(atan)                                \
 419       SymI_HasProto(acos)                                \
 420       SymI_HasProto(asin)                                \
 421       SymI_HasProto(tan)                                 \
 422       SymI_HasProto(cos)                                 \
 423       SymI_HasProto(sin)                                 \
 424       SymI_HasProto(exp)                                 \
 425       SymI_HasProto(log)                                 \
 426       SymI_HasProto(sqrt)                                \
 427       SymI_HasProto(powf)                                \
 428       SymI_HasProto(tanhf)                               \
 429       SymI_HasProto(coshf)                               \
 430       SymI_HasProto(sinhf)                               \
 431       SymI_HasProto(atanf)                               \
 432       SymI_HasProto(acosf)                               \
 433       SymI_HasProto(asinf)                               \
 434       SymI_HasProto(tanf)                                \
 435       SymI_HasProto(cosf)                                \
 436       SymI_HasProto(sinf)                                \
 437       SymI_HasProto(expf)                                \
 438       SymI_HasProto(logf)                                \
 439       SymI_HasProto(sqrtf)                               \
 440       SymI_HasProto(erf)                                \
 441       SymI_HasProto(erfc)                                \
 442       SymI_HasProto(erff)                                \
 443       SymI_HasProto(erfcf)                               \
 444       SymI_HasProto(memcpy)                              \
 445       SymI_HasProto(rts_InstallConsoleEvent)             \
 446       SymI_HasProto(rts_ConsoleHandlerDone)              \
 447       SymI_NeedsProto(mktime)                            \
 448       SymI_NeedsProto(_imp___timezone)                   \
 449       SymI_NeedsProto(_imp___tzname)                     \
 450       SymI_NeedsProto(_imp__tzname)                      \
 451       SymI_NeedsProto(_imp___iob)                        \
 452       SymI_NeedsProto(_imp___osver)                      \
 453       SymI_NeedsProto(localtime)                         \
 454       SymI_NeedsProto(gmtime)                            \
 455       SymI_NeedsProto(opendir)                           \
 456       SymI_NeedsProto(readdir)                           \
 457       SymI_NeedsProto(rewinddir)                         \
 458       SymI_NeedsProto(_imp____mb_cur_max)                \
 459       SymI_NeedsProto(_imp___pctype)                     \
 460       SymI_NeedsProto(__chkstk)                          \
 461       RTS_MINGW_GETTIMEOFDAY_SYM                         \
 462       SymI_NeedsProto(closedir)
 463 #endif
 464
 465
 466 #if defined(darwin_HOST_OS) && HAVE_PRINTF_LDBLSTUB
 467 #define RTS_DARWIN_ONLY_SYMBOLS                             \
 468      SymI_NeedsProto(asprintf$LDBLStub)                     \
 469      SymI_NeedsProto(err$LDBLStub)                          \
 470      SymI_NeedsProto(errc$LDBLStub)                         \
 471      SymI_NeedsProto(errx$LDBLStub)                         \
 472      SymI_NeedsProto(fprintf$LDBLStub)                      \
 473      SymI_NeedsProto(fscanf$LDBLStub)                       \
 474      SymI_NeedsProto(fwprintf$LDBLStub)                     \
 475      SymI_NeedsProto(fwscanf$LDBLStub)                      \
 476      SymI_NeedsProto(printf$LDBLStub)                       \
 477      SymI_NeedsProto(scanf$LDBLStub)                        \
 478      SymI_NeedsProto(snprintf$LDBLStub)                     \
 479      SymI_NeedsProto(sprintf$LDBLStub)                      \
 480      SymI_NeedsProto(sscanf$LDBLStub)                       \
 481      SymI_NeedsProto(strtold$LDBLStub)                      \
 482      SymI_NeedsProto(swprintf$LDBLStub)                     \
 483      SymI_NeedsProto(swscanf$LDBLStub)                      \
 484      SymI_NeedsProto(syslog$LDBLStub)                       \
 485      SymI_NeedsProto(vasprintf$LDBLStub)                    \
 486      SymI_NeedsProto(verr$LDBLStub)                         \
 487      SymI_NeedsProto(verrc$LDBLStub)                        \
 488      SymI_NeedsProto(verrx$LDBLStub)                        \
 489      SymI_NeedsProto(vfprintf$LDBLStub)                     \
 490      SymI_NeedsProto(vfscanf$LDBLStub)                      \
 491      SymI_NeedsProto(vfwprintf$LDBLStub)                    \
 492      SymI_NeedsProto(vfwscanf$LDBLStub)                     \
 493      SymI_NeedsProto(vprintf$LDBLStub)                      \
 494      SymI_NeedsProto(vscanf$LDBLStub)                       \
 495      SymI_NeedsProto(vsnprintf$LDBLStub)                    \
 496      SymI_NeedsProto(vsprintf$LDBLStub)                     \
 497      SymI_NeedsProto(vsscanf$LDBLStub)                      \
 498      SymI_NeedsProto(vswprintf$LDBLStub)                    \
 499      SymI_NeedsProto(vswscanf$LDBLStub)                     \
 500      SymI_NeedsProto(vsyslog$LDBLStub)                      \
 501      SymI_NeedsProto(vwarn$LDBLStub)                        \
 502      SymI_NeedsProto(vwarnc$LDBLStub)                       \
 503      SymI_NeedsProto(vwarnx$LDBLStub)                       \
 504      SymI_NeedsProto(vwprintf$LDBLStub)                     \
 505      SymI_NeedsProto(vwscanf$LDBLStub)                      \
 506      SymI_NeedsProto(warn$LDBLStub)                         \
 507      SymI_NeedsProto(warnc$LDBLStub)                        \
 508      SymI_NeedsProto(warnx$LDBLStub)                        \
 509      SymI_NeedsProto(wcstold$LDBLStub)                      \
 510      SymI_NeedsProto(wprintf$LDBLStub)                      \
 511      SymI_NeedsProto(wscanf$LDBLStub)
 512 #else
 513 #define RTS_DARWIN_ONLY_SYMBOLS
 514 #endif
 515
 516 #ifndef SMP
 517 # define MAIN_CAP_SYM SymI_HasProto(MainCapability)
 518 #else
 519 # define MAIN_CAP_SYM
 520 #endif
 521
 522 #if !defined(mingw32_HOST_OS)
 523 #define RTS_USER_SIGNALS_SYMBOLS \
 524    SymI_HasProto(setIOManagerControlFd) \
 525    SymI_HasProto(setIOManagerWakeupFd) \
 526    SymI_HasProto(ioManagerWakeup) \
 527    SymI_HasProto(blockUserSignals) \
 528    SymI_HasProto(unblockUserSignals)
 529 #else
 530 #define RTS_USER_SIGNALS_SYMBOLS     \
 531    SymI_HasProto(ioManagerWakeup) \
 532    SymI_HasProto(sendIOManagerEvent) \
 533    SymI_HasProto(readIOManagerEvent) \
 534    SymI_HasProto(getIOManagerEvent)  \
 535    SymI_HasProto(console_handler)
 536 #endif
 537
 538 #define RTS_LIBFFI_SYMBOLS                                  \
 539      SymE_NeedsProto(ffi_prep_cif)                          \
 540      SymE_NeedsProto(ffi_call)                              \
 541      SymE_NeedsProto(ffi_type_void)                         \
 542      SymE_NeedsProto(ffi_type_float)                        \
 543      SymE_NeedsProto(ffi_type_double)                       \
 544      SymE_NeedsProto(ffi_type_sint64)                       \
 545      SymE_NeedsProto(ffi_type_uint64)                       \
 546      SymE_NeedsProto(ffi_type_sint32)                       \
 547      SymE_NeedsProto(ffi_type_uint32)                       \
 548      SymE_NeedsProto(ffi_type_sint16)                       \
 549      SymE_NeedsProto(ffi_type_uint16)                       \
 550      SymE_NeedsProto(ffi_type_sint8)                        \
 551      SymE_NeedsProto(ffi_type_uint8)                        \
 552      SymE_NeedsProto(ffi_type_pointer)
 553
 554 #ifdef TABLES_NEXT_TO_CODE
 555 #define RTS_RET_SYMBOLS /* nothing */
 556 #else
 557 #define RTS_RET_SYMBOLS                                 \
 558       SymI_HasProto(stg_enter_ret)                      \
 559       SymI_HasProto(stg_gc_fun_ret)                     \
 560       SymI_HasProto(stg_ap_v_ret)                       \
 561       SymI_HasProto(stg_ap_f_ret)                       \
 562       SymI_HasProto(stg_ap_d_ret)                       \
 563       SymI_HasProto(stg_ap_l_ret)                       \
 564       SymI_HasProto(stg_ap_n_ret)                       \
 565       SymI_HasProto(stg_ap_p_ret)                       \
 566       SymI_HasProto(stg_ap_pv_ret)                      \
 567       SymI_HasProto(stg_ap_pp_ret)                      \
 568       SymI_HasProto(stg_ap_ppv_ret)                     \
 569       SymI_HasProto(stg_ap_ppp_ret)                     \
 570       SymI_HasProto(stg_ap_pppv_ret)                    \
 571       SymI_HasProto(stg_ap_pppp_ret)                    \
 572       SymI_HasProto(stg_ap_ppppp_ret)                   \
 573       SymI_HasProto(stg_ap_pppppp_ret)
 574 #endif
 575
 576 /* Modules compiled with -ticky may mention ticky counters */
 577 /* This list should marry up with the one in $(TOP)/includes/stg/Ticky.h */
 578 #define RTS_TICKY_SYMBOLS                       \
 579       SymI_NeedsProto(ticky_entry_ctrs)         \
 580       SymI_NeedsProto(top_ct)                   \
 581                                                 \
 582       SymI_HasProto(ENT_VIA_NODE_ctr)           \
 583       SymI_HasProto(ENT_STATIC_THK_ctr)         \
 584       SymI_HasProto(ENT_DYN_THK_ctr)            \
 585       SymI_HasProto(ENT_STATIC_FUN_DIRECT_ctr)  \
 586       SymI_HasProto(ENT_DYN_FUN_DIRECT_ctr)     \
 587       SymI_HasProto(ENT_STATIC_CON_ctr)         \
 588       SymI_HasProto(ENT_DYN_CON_ctr)            \
 589       SymI_HasProto(ENT_STATIC_IND_ctr)         \
 590       SymI_HasProto(ENT_DYN_IND_ctr)            \
 591       SymI_HasProto(ENT_PERM_IND_ctr)           \
 592       SymI_HasProto(ENT_PAP_ctr)                \
 593       SymI_HasProto(ENT_AP_ctr)                 \
 594       SymI_HasProto(ENT_AP_STACK_ctr)           \
 595       SymI_HasProto(ENT_BH_ctr)                 \
 596       SymI_HasProto(UNKNOWN_CALL_ctr)           \
 597       SymI_HasProto(SLOW_CALL_v_ctr)            \
 598       SymI_HasProto(SLOW_CALL_f_ctr)            \
 599       SymI_HasProto(SLOW_CALL_d_ctr)            \
 600       SymI_HasProto(SLOW_CALL_l_ctr)            \
 601       SymI_HasProto(SLOW_CALL_n_ctr)            \
 602       SymI_HasProto(SLOW_CALL_p_ctr)            \
 603       SymI_HasProto(SLOW_CALL_pv_ctr)           \
 604       SymI_HasProto(SLOW_CALL_pp_ctr)           \
 605       SymI_HasProto(SLOW_CALL_ppv_ctr)          \
 606       SymI_HasProto(SLOW_CALL_ppp_ctr)          \
 607       SymI_HasProto(SLOW_CALL_pppv_ctr)         \
 608       SymI_HasProto(SLOW_CALL_pppp_ctr)         \
 609       SymI_HasProto(SLOW_CALL_ppppp_ctr)                \
 610       SymI_HasProto(SLOW_CALL_pppppp_ctr)               \
 611       SymI_HasProto(SLOW_CALL_OTHER_ctr)                \
 612       SymI_HasProto(ticky_slow_call_unevald)            \
 613       SymI_HasProto(SLOW_CALL_ctr)                      \
 614       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_ctr)          \
 615       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_CHUNKS_ctr)   \
 616       SymI_HasProto(KNOWN_CALL_ctr)                     \
 617       SymI_HasProto(KNOWN_CALL_TOO_FEW_ARGS_ctr)        \
 618       SymI_HasProto(KNOWN_CALL_EXTRA_ARGS_ctr)          \
 619       SymI_HasProto(SLOW_CALL_FUN_TOO_FEW_ctr)          \
 620       SymI_HasProto(SLOW_CALL_FUN_CORRECT_ctr)          \
 621       SymI_HasProto(SLOW_CALL_FUN_TOO_MANY_ctr)         \
 622       SymI_HasProto(SLOW_CALL_PAP_TOO_FEW_ctr)          \
 623       SymI_HasProto(SLOW_CALL_PAP_CORRECT_ctr)          \
 624       SymI_HasProto(SLOW_CALL_PAP_TOO_MANY_ctr)         \
 625       SymI_HasProto(SLOW_CALL_UNEVALD_ctr)              \
 626       SymI_HasProto(UPDF_OMITTED_ctr)           \
 627       SymI_HasProto(UPDF_PUSHED_ctr)            \
 628       SymI_HasProto(CATCHF_PUSHED_ctr)          \
 629       SymI_HasProto(UPDF_RCC_PUSHED_ctr)        \
 630       SymI_HasProto(UPDF_RCC_OMITTED_ctr)       \
 631       SymI_HasProto(UPD_SQUEEZED_ctr)           \
 632       SymI_HasProto(UPD_CON_IN_NEW_ctr)         \
 633       SymI_HasProto(UPD_CON_IN_PLACE_ctr)       \
 634       SymI_HasProto(UPD_PAP_IN_NEW_ctr)         \
 635       SymI_HasProto(UPD_PAP_IN_PLACE_ctr)       \
 636       SymI_HasProto(ALLOC_HEAP_ctr)             \
 637       SymI_HasProto(ALLOC_HEAP_tot)             \
 638       SymI_HasProto(ALLOC_FUN_ctr)              \
 639       SymI_HasProto(ALLOC_FUN_adm)              \
 640       SymI_HasProto(ALLOC_FUN_gds)              \
 641       SymI_HasProto(ALLOC_FUN_slp)              \
 642       SymI_HasProto(UPD_NEW_IND_ctr)            \
 643       SymI_HasProto(UPD_NEW_PERM_IND_ctr)       \
 644       SymI_HasProto(UPD_OLD_IND_ctr)            \
 645       SymI_HasProto(UPD_OLD_PERM_IND_ctr)               \
 646       SymI_HasProto(UPD_BH_UPDATABLE_ctr)               \
 647       SymI_HasProto(UPD_BH_SINGLE_ENTRY_ctr)            \
 648       SymI_HasProto(UPD_CAF_BH_UPDATABLE_ctr)           \
 649       SymI_HasProto(UPD_CAF_BH_SINGLE_ENTRY_ctr)        \
 650       SymI_HasProto(GC_SEL_ABANDONED_ctr)               \
 651       SymI_HasProto(GC_SEL_MINOR_ctr)           \
 652       SymI_HasProto(GC_SEL_MAJOR_ctr)           \
 653       SymI_HasProto(GC_FAILED_PROMOTION_ctr)    \
 654       SymI_HasProto(ALLOC_UP_THK_ctr)           \
 655       SymI_HasProto(ALLOC_SE_THK_ctr)           \
 656       SymI_HasProto(ALLOC_THK_adm)              \
 657       SymI_HasProto(ALLOC_THK_gds)              \
 658       SymI_HasProto(ALLOC_THK_slp)              \
 659       SymI_HasProto(ALLOC_CON_ctr)              \
 660       SymI_HasProto(ALLOC_CON_adm)              \
 661       SymI_HasProto(ALLOC_CON_gds)              \
 662       SymI_HasProto(ALLOC_CON_slp)              \
 663       SymI_HasProto(ALLOC_TUP_ctr)              \
 664       SymI_HasProto(ALLOC_TUP_adm)              \
 665       SymI_HasProto(ALLOC_TUP_gds)              \
 666       SymI_HasProto(ALLOC_TUP_slp)              \
 667       SymI_HasProto(ALLOC_BH_ctr)               \
 668       SymI_HasProto(ALLOC_BH_adm)               \
 669       SymI_HasProto(ALLOC_BH_gds)               \
 670       SymI_HasProto(ALLOC_BH_slp)               \
 671       SymI_HasProto(ALLOC_PRIM_ctr)             \
 672       SymI_HasProto(ALLOC_PRIM_adm)             \
 673       SymI_HasProto(ALLOC_PRIM_gds)             \
 674       SymI_HasProto(ALLOC_PRIM_slp)             \
 675       SymI_HasProto(ALLOC_PAP_ctr)              \
 676       SymI_HasProto(ALLOC_PAP_adm)              \
 677       SymI_HasProto(ALLOC_PAP_gds)              \
 678       SymI_HasProto(ALLOC_PAP_slp)              \
 679       SymI_HasProto(ALLOC_TSO_ctr)              \
 680       SymI_HasProto(ALLOC_TSO_adm)              \
 681       SymI_HasProto(ALLOC_TSO_gds)              \
 682       SymI_HasProto(ALLOC_TSO_slp)              \
 683       SymI_HasProto(RET_NEW_ctr)                \
 684       SymI_HasProto(RET_OLD_ctr)                \
 685       SymI_HasProto(RET_UNBOXED_TUP_ctr)        \
 686       SymI_HasProto(RET_SEMI_loads_avoided)
 687
 688
 689 // On most platforms, the garbage collector rewrites references
 690 //      to small integer and char objects to a set of common, shared ones.
 691 //
 692 // We don't do this when compiling to Windows DLLs at the moment because
 693 //      it doesn't support cross package data references well.
 694 //
 695 #if defined(__PIC__) && defined(mingw32_HOST_OS)
 696 #define RTS_INTCHAR_SYMBOLS
 697 #else
 698 #define RTS_INTCHAR_SYMBOLS                             \
 699       SymI_HasProto(stg_CHARLIKE_closure)               \
 700       SymI_HasProto(stg_INTLIKE_closure)
 701 #endif
 702
 703
 704 #define RTS_SYMBOLS                                     \
 705       Maybe_Stable_Names                                \
 706       RTS_TICKY_SYMBOLS                                 \
 707       SymI_HasProto(StgReturn)                          \
 708       SymI_HasProto(stg_enter_info)                     \
 709       SymI_HasProto(stg_gc_void_info)                   \
 710       SymI_HasProto(__stg_gc_enter_1)                   \
 711       SymI_HasProto(stg_gc_noregs)                      \
 712       SymI_HasProto(stg_gc_unpt_r1_info)                \
 713       SymI_HasProto(stg_gc_unpt_r1)                     \
 714       SymI_HasProto(stg_gc_unbx_r1_info)                \
 715       SymI_HasProto(stg_gc_unbx_r1)                     \
 716       SymI_HasProto(stg_gc_f1_info)                     \
 717       SymI_HasProto(stg_gc_f1)                          \
 718       SymI_HasProto(stg_gc_d1_info)                     \
 719       SymI_HasProto(stg_gc_d1)                          \
 720       SymI_HasProto(stg_gc_l1_info)                     \
 721       SymI_HasProto(stg_gc_l1)                          \
 722       SymI_HasProto(__stg_gc_fun)                       \
 723       SymI_HasProto(stg_gc_fun_info)                    \
 724       SymI_HasProto(stg_gc_gen)                         \
 725       SymI_HasProto(stg_gc_gen_info)                    \
 726       SymI_HasProto(stg_gc_gen_hp)                      \
 727       SymI_HasProto(stg_gc_ut)                          \
 728       SymI_HasProto(stg_gen_yield)                      \
 729       SymI_HasProto(stg_yield_noregs)                   \
 730       SymI_HasProto(stg_yield_to_interpreter)           \
 731       SymI_HasProto(stg_gen_block)                      \
 732       SymI_HasProto(stg_block_noregs)                   \
 733       SymI_HasProto(stg_block_1)                        \
 734       SymI_HasProto(stg_block_takemvar)                 \
 735       SymI_HasProto(stg_block_putmvar)                  \
 736       MAIN_CAP_SYM                                      \
 737       SymI_HasProto(MallocFailHook)                     \
 738       SymI_HasProto(OnExitHook)                         \
 739       SymI_HasProto(OutOfHeapHook)                      \
 740       SymI_HasProto(StackOverflowHook)                  \
 741       SymI_HasProto(addDLL)                             \
 742       SymI_HasProto(__int_encodeDouble)                 \
 743       SymI_HasProto(__word_encodeDouble)                \
 744       SymI_HasProto(__2Int_encodeDouble)                \
 745       SymI_HasProto(__int_encodeFloat)                  \
 746       SymI_HasProto(__word_encodeFloat)                 \
 747       SymI_HasProto(stg_atomicallyzh)                   \
 748       SymI_HasProto(barf)                               \
 749       SymI_HasProto(debugBelch)                         \
 750       SymI_HasProto(errorBelch)                         \
 751       SymI_HasProto(sysErrorBelch)                      \
 752       SymI_HasProto(stg_getMaskingStatezh)              \
 753       SymI_HasProto(stg_maskAsyncExceptionszh)          \
 754       SymI_HasProto(stg_maskUninterruptiblezh)          \
 755       SymI_HasProto(stg_catchzh)                        \
 756       SymI_HasProto(stg_catchRetryzh)                   \
 757       SymI_HasProto(stg_catchSTMzh)                     \
 758       SymI_HasProto(stg_checkzh)                        \
 759       SymI_HasProto(closure_flags)                      \
 760       SymI_HasProto(cmp_thread)                         \
 761       SymI_HasProto(createAdjustor)                     \
 762       SymI_HasProto(stg_decodeDoublezu2Intzh)           \
 763       SymI_HasProto(stg_decodeFloatzuIntzh)             \
 764       SymI_HasProto(defaultsHook)                       \
 765       SymI_HasProto(stg_delayzh)                        \
 766       SymI_HasProto(stg_deRefWeakzh)                    \
 767       SymI_HasProto(stg_deRefStablePtrzh)               \
 768       SymI_HasProto(dirty_MUT_VAR)                      \
 769       SymI_HasProto(stg_forkzh)                         \
 770       SymI_HasProto(stg_forkOnzh)                       \
 771       SymI_HasProto(forkProcess)                        \
 772       SymI_HasProto(forkOS_createThread)                \
 773       SymI_HasProto(freeHaskellFunctionPtr)             \
 774       SymI_HasProto(getOrSetTypeableStore)              \
 775       SymI_HasProto(getOrSetGHCConcSignalSignalHandlerStore)            \
 776       SymI_HasProto(getOrSetGHCConcWindowsPendingDelaysStore)           \
 777       SymI_HasProto(getOrSetGHCConcWindowsIOManagerThreadStore) \
 778       SymI_HasProto(getOrSetGHCConcWindowsProddingStore)                \
 779       SymI_HasProto(getOrSetSystemEventThreadEventManagerStore)         \
 780       SymI_HasProto(getOrSetSystemEventThreadIOManagerThreadStore)              \
 781       SymI_HasProto(genSymZh)                           \
 782       SymI_HasProto(genericRaise)                       \
 783       SymI_HasProto(getProgArgv)                        \
 784       SymI_HasProto(getFullProgArgv)                    \
 785       SymI_HasProto(getStablePtr)                       \
 786       SymI_HasProto(hs_init)                            \
 787       SymI_HasProto(hs_exit)                            \
 788       SymI_HasProto(hs_set_argv)                        \
 789       SymI_HasProto(hs_add_root)                        \
 790       SymI_HasProto(hs_perform_gc)                      \
 791       SymI_HasProto(hs_free_stable_ptr)                 \
 792       SymI_HasProto(hs_free_fun_ptr)                    \
 793       SymI_HasProto(hs_hpc_rootModule)                  \
 794       SymI_HasProto(hs_hpc_module)                      \
 795       SymI_HasProto(initLinker)                         \
 796       SymI_HasProto(stg_unpackClosurezh)                \
 797       SymI_HasProto(stg_getApStackValzh)                \
 798       SymI_HasProto(stg_getSparkzh)                     \
 799       SymI_HasProto(stg_numSparkszh)                    \
 800       SymI_HasProto(stg_isCurrentThreadBoundzh)         \
 801       SymI_HasProto(stg_isEmptyMVarzh)                  \
 802       SymI_HasProto(stg_killThreadzh)                   \
 803       SymI_HasProto(loadArchive)                                \
 804       SymI_HasProto(loadObj)                            \
 805       SymI_HasProto(insertStableSymbol)                 \
 806       SymI_HasProto(insertSymbol)                       \
 807       SymI_HasProto(lookupSymbol)                       \
 808       SymI_HasProto(stg_makeStablePtrzh)                \
 809       SymI_HasProto(stg_mkApUpd0zh)                     \
 810       SymI_HasProto(stg_myThreadIdzh)                   \
 811       SymI_HasProto(stg_labelThreadzh)                  \
 812       SymI_HasProto(stg_newArrayzh)                     \
 813       SymI_HasProto(stg_newBCOzh)                       \
 814       SymI_HasProto(stg_newByteArrayzh)                 \
 815       SymI_HasProto_redirect(newCAF, newDynCAF)         \
 816       SymI_HasProto(stg_newMVarzh)                      \
 817       SymI_HasProto(stg_newMutVarzh)                    \
 818       SymI_HasProto(stg_newTVarzh)                      \
 819       SymI_HasProto(stg_noDuplicatezh)                  \
 820       SymI_HasProto(stg_atomicModifyMutVarzh)           \
 821       SymI_HasProto(stg_newPinnedByteArrayzh)           \
 822       SymI_HasProto(stg_newAlignedPinnedByteArrayzh)    \
 823       SymI_HasProto(newSpark)                           \
 824       SymI_HasProto(performGC)                          \
 825       SymI_HasProto(performMajorGC)                     \
 826       SymI_HasProto(prog_argc)                          \
 827       SymI_HasProto(prog_argv)                          \
 828       SymI_HasProto(stg_putMVarzh)                      \
 829       SymI_HasProto(stg_raisezh)                        \
 830       SymI_HasProto(stg_raiseIOzh)                      \
 831       SymI_HasProto(stg_readTVarzh)                     \
 832       SymI_HasProto(stg_readTVarIOzh)                   \
 833       SymI_HasProto(resumeThread)                       \
 834       SymI_HasProto(resolveObjs)                        \
 835       SymI_HasProto(stg_retryzh)                        \
 836       SymI_HasProto(rts_apply)                          \
 837       SymI_HasProto(rts_checkSchedStatus)               \
 838       SymI_HasProto(rts_eval)                           \
 839       SymI_HasProto(rts_evalIO)                         \
 840       SymI_HasProto(rts_evalLazyIO)                     \
 841       SymI_HasProto(rts_evalStableIO)                   \
 842       SymI_HasProto(rts_eval_)                          \
 843       SymI_HasProto(rts_getBool)                        \
 844       SymI_HasProto(rts_getChar)                        \
 845       SymI_HasProto(rts_getDouble)                      \
 846       SymI_HasProto(rts_getFloat)                       \
 847       SymI_HasProto(rts_getInt)                         \
 848       SymI_HasProto(rts_getInt8)                        \
 849       SymI_HasProto(rts_getInt16)                       \
 850       SymI_HasProto(rts_getInt32)                       \
 851       SymI_HasProto(rts_getInt64)                       \
 852       SymI_HasProto(rts_getPtr)                         \
 853       SymI_HasProto(rts_getFunPtr)                      \
 854       SymI_HasProto(rts_getStablePtr)                   \
 855       SymI_HasProto(rts_getThreadId)                    \
 856       SymI_HasProto(rts_getWord)                        \
 857       SymI_HasProto(rts_getWord8)                       \
 858       SymI_HasProto(rts_getWord16)                      \
 859       SymI_HasProto(rts_getWord32)                      \
 860       SymI_HasProto(rts_getWord64)                      \
 861       SymI_HasProto(rts_lock)                           \
 862       SymI_HasProto(rts_mkBool)                         \
 863       SymI_HasProto(rts_mkChar)                         \
 864       SymI_HasProto(rts_mkDouble)                       \
 865       SymI_HasProto(rts_mkFloat)                        \
 866       SymI_HasProto(rts_mkInt)                          \
 867       SymI_HasProto(rts_mkInt8)                         \
 868       SymI_HasProto(rts_mkInt16)                        \
 869       SymI_HasProto(rts_mkInt32)                        \
 870       SymI_HasProto(rts_mkInt64)                        \
 871       SymI_HasProto(rts_mkPtr)                          \
 872       SymI_HasProto(rts_mkFunPtr)                       \
 873       SymI_HasProto(rts_mkStablePtr)                    \
 874       SymI_HasProto(rts_mkString)                       \
 875       SymI_HasProto(rts_mkWord)                         \
 876       SymI_HasProto(rts_mkWord8)                        \
 877       SymI_HasProto(rts_mkWord16)                       \
 878       SymI_HasProto(rts_mkWord32)                       \
 879       SymI_HasProto(rts_mkWord64)                       \
 880       SymI_HasProto(rts_unlock)                         \
 881       SymI_HasProto(rts_unsafeGetMyCapability)          \
 882       SymI_HasProto(rtsSupportsBoundThreads)            \
 883       SymI_HasProto(rts_isProfiled)                     \
 884       SymI_HasProto(setProgArgv)                        \
 885       SymI_HasProto(startupHaskell)                     \
 886       SymI_HasProto(shutdownHaskell)                    \
 887       SymI_HasProto(shutdownHaskellAndExit)             \
 888       SymI_HasProto(stable_ptr_table)                   \
 889       SymI_HasProto(stackOverflow)                      \
 890       SymI_HasProto(stg_CAF_BLACKHOLE_info)             \
 891       SymI_HasProto(stg_BLACKHOLE_info)                 \
 892       SymI_HasProto(__stg_EAGER_BLACKHOLE_info)         \
 893       SymI_HasProto(stg_BLOCKING_QUEUE_CLEAN_info)      \
 894       SymI_HasProto(stg_BLOCKING_QUEUE_DIRTY_info)      \
 895       SymI_HasProto(startTimer)                         \
 896       SymI_HasProto(stg_MVAR_CLEAN_info)                \
 897       SymI_HasProto(stg_MVAR_DIRTY_info)                \
 898       SymI_HasProto(stg_IND_STATIC_info)                \
 899       SymI_HasProto(stg_ARR_WORDS_info)                 \
 900       SymI_HasProto(stg_MUT_ARR_PTRS_DIRTY_info)        \
 901       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN_info)       \
 902       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN0_info)      \
 903       SymI_HasProto(stg_WEAK_info)                      \
 904       SymI_HasProto(stg_ap_v_info)                      \
 905       SymI_HasProto(stg_ap_f_info)                      \
 906       SymI_HasProto(stg_ap_d_info)                      \
 907       SymI_HasProto(stg_ap_l_info)                      \
 908       SymI_HasProto(stg_ap_n_info)                      \
 909       SymI_HasProto(stg_ap_p_info)                      \
 910       SymI_HasProto(stg_ap_pv_info)                     \
 911       SymI_HasProto(stg_ap_pp_info)                     \
 912       SymI_HasProto(stg_ap_ppv_info)                    \
 913       SymI_HasProto(stg_ap_ppp_info)                    \
 914       SymI_HasProto(stg_ap_pppv_info)                   \
 915       SymI_HasProto(stg_ap_pppp_info)                   \
 916       SymI_HasProto(stg_ap_ppppp_info)                  \
 917       SymI_HasProto(stg_ap_pppppp_info)                 \
 918       SymI_HasProto(stg_ap_0_fast)                      \
 919       SymI_HasProto(stg_ap_v_fast)                      \
 920       SymI_HasProto(stg_ap_f_fast)                      \
 921       SymI_HasProto(stg_ap_d_fast)                      \
 922       SymI_HasProto(stg_ap_l_fast)                      \
 923       SymI_HasProto(stg_ap_n_fast)                      \
 924       SymI_HasProto(stg_ap_p_fast)                      \
 925       SymI_HasProto(stg_ap_pv_fast)                     \
 926       SymI_HasProto(stg_ap_pp_fast)                     \
 927       SymI_HasProto(stg_ap_ppv_fast)                    \
 928       SymI_HasProto(stg_ap_ppp_fast)                    \
 929       SymI_HasProto(stg_ap_pppv_fast)                   \
 930       SymI_HasProto(stg_ap_pppp_fast)                   \
 931       SymI_HasProto(stg_ap_ppppp_fast)                  \
 932       SymI_HasProto(stg_ap_pppppp_fast)                 \
 933       SymI_HasProto(stg_ap_1_upd_info)                  \
 934       SymI_HasProto(stg_ap_2_upd_info)                  \
 935       SymI_HasProto(stg_ap_3_upd_info)                  \
 936       SymI_HasProto(stg_ap_4_upd_info)                  \
 937       SymI_HasProto(stg_ap_5_upd_info)                  \
 938       SymI_HasProto(stg_ap_6_upd_info)                  \
 939       SymI_HasProto(stg_ap_7_upd_info)                  \
 940       SymI_HasProto(stg_exit)                           \
 941       SymI_HasProto(stg_sel_0_upd_info)                 \
 942       SymI_HasProto(stg_sel_10_upd_info)                \
 943       SymI_HasProto(stg_sel_11_upd_info)                \
 944       SymI_HasProto(stg_sel_12_upd_info)                \
 945       SymI_HasProto(stg_sel_13_upd_info)                \
 946       SymI_HasProto(stg_sel_14_upd_info)                \
 947       SymI_HasProto(stg_sel_15_upd_info)                \
 948       SymI_HasProto(stg_sel_1_upd_info)                 \
 949       SymI_HasProto(stg_sel_2_upd_info)                 \
 950       SymI_HasProto(stg_sel_3_upd_info)                 \
 951       SymI_HasProto(stg_sel_4_upd_info)                 \
 952       SymI_HasProto(stg_sel_5_upd_info)                 \
 953       SymI_HasProto(stg_sel_6_upd_info)                 \
 954       SymI_HasProto(stg_sel_7_upd_info)                 \
 955       SymI_HasProto(stg_sel_8_upd_info)                 \
 956       SymI_HasProto(stg_sel_9_upd_info)                 \
 957       SymI_HasProto(stg_upd_frame_info)                 \
 958       SymI_HasProto(stg_bh_upd_frame_info)              \
 959       SymI_HasProto(suspendThread)                      \
 960       SymI_HasProto(stg_takeMVarzh)                     \
 961       SymI_HasProto(stg_threadStatuszh)                 \
 962       SymI_HasProto(stg_tryPutMVarzh)                   \
 963       SymI_HasProto(stg_tryTakeMVarzh)                  \
 964       SymI_HasProto(stg_unmaskAsyncExceptionszh)        \
 965       SymI_HasProto(unloadObj)                          \
 966       SymI_HasProto(stg_unsafeThawArrayzh)              \
 967       SymI_HasProto(stg_waitReadzh)                     \
 968       SymI_HasProto(stg_waitWritezh)                    \
 969       SymI_HasProto(stg_writeTVarzh)                    \
 970       SymI_HasProto(stg_yieldzh)                        \
 971       SymI_NeedsProto(stg_interp_constr_entry)          \
 972       SymI_HasProto(stg_arg_bitmaps)                    \
 973       SymI_HasProto(alloc_blocks_lim)                   \
 974       SymI_HasProto(g0)                                 \
 975       SymI_HasProto(allocate)                           \
 976       SymI_HasProto(allocateExec)                       \
 977       SymI_HasProto(freeExec)                           \
 978       SymI_HasProto(getAllocations)                     \
 979       SymI_HasProto(revertCAFs)                         \
 980       SymI_HasProto(RtsFlags)                           \
 981       SymI_NeedsProto(rts_breakpoint_io_action)         \
 982       SymI_NeedsProto(rts_stop_next_breakpoint)         \
 983       SymI_NeedsProto(rts_stop_on_exception)            \
 984       SymI_HasProto(stopTimer)                          \
 985       SymI_HasProto(n_capabilities)                     \
 986       SymI_HasProto(stg_traceCcszh)                     \
 987       SymI_HasProto(stg_traceEventzh)                   \
 988       RTS_USER_SIGNALS_SYMBOLS                          \
 989       RTS_INTCHAR_SYMBOLS
 990
 991
 992 // 64-bit support functions in libgcc.a
 993 #if defined(__GNUC__) && SIZEOF_VOID_P <= 4
 994 #define RTS_LIBGCC_SYMBOLS                             \
 995       SymI_NeedsProto(__divdi3)                        \
 996       SymI_NeedsProto(__udivdi3)                       \
 997       SymI_NeedsProto(__moddi3)                        \
 998       SymI_NeedsProto(__umoddi3)                       \
 999       SymI_NeedsProto(__muldi3)                        \
1000       SymI_NeedsProto(__ashldi3)                       \
1001       SymI_NeedsProto(__ashrdi3)                       \
1002       SymI_NeedsProto(__lshrdi3)
1003 #else
1004 #define RTS_LIBGCC_SYMBOLS
1005 #endif
1006
1007 #if defined(darwin_HOST_OS) && defined(powerpc_HOST_ARCH)
1008       // Symbols that don't have a leading underscore
1009       // on Mac OS X. They have to receive special treatment,
1010       // see machoInitSymbolsWithoutUnderscore()
1011 #define RTS_MACHO_NOUNDERLINE_SYMBOLS           \
1012       SymI_NeedsProto(saveFP)                           \
1013       SymI_NeedsProto(restFP)
1014 #endif
1015
1016 /* entirely bogus claims about types of these symbols */
1017 #define SymI_NeedsProto(vvv)  extern void vvv(void);
1018 #if defined(__PIC__) && defined(mingw32_HOST_OS)
1019 #define SymE_HasProto(vvv)    SymE_HasProto(vvv);
1020 #define SymE_NeedsProto(vvv)    extern void _imp__ ## vvv (void);
1021 #else
1022 #define SymE_NeedsProto(vvv)  SymI_NeedsProto(vvv);
1023 #define SymE_HasProto(vvv)    SymI_HasProto(vvv)
1024 #endif
1025 #define SymI_HasProto(vvv) /**/
1026 #define SymI_HasProto_redirect(vvv,xxx) /**/
1027 RTS_SYMBOLS
1028 RTS_RET_SYMBOLS
1029 RTS_POSIX_ONLY_SYMBOLS
1030 RTS_MINGW_ONLY_SYMBOLS
1031 RTS_CYGWIN_ONLY_SYMBOLS
1032 RTS_DARWIN_ONLY_SYMBOLS
1033 RTS_LIBGCC_SYMBOLS
1034 RTS_LIBFFI_SYMBOLS
1035 #undef SymI_NeedsProto
1036 #undef SymI_HasProto
1037 #undef SymI_HasProto_redirect
1038 #undef SymE_HasProto
1039 #undef SymE_NeedsProto
1040
1041 #ifdef LEADING_UNDERSCORE
1042 #define MAYBE_LEADING_UNDERSCORE_STR(s) ("_" s)
1043 #else
1044 #define MAYBE_LEADING_UNDERSCORE_STR(s) (s)
1045 #endif
1046
1047 #define SymI_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1048                     (void*)(&(vvv)) },
1049 #define SymE_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1050             (void*)DLL_IMPORT_DATA_REF(vvv) },
1051
1052 #define SymI_NeedsProto(vvv) SymI_HasProto(vvv)
1053 #define SymE_NeedsProto(vvv) SymE_HasProto(vvv)
1054
1055 // SymI_HasProto_redirect allows us to redirect references to one symbol to
1056 // another symbol.  See newCAF/newDynCAF for an example.
1057 #define SymI_HasProto_redirect(vvv,xxx) \
1058     { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1059       (void*)(&(xxx)) },
1060
1061 static RtsSymbolVal rtsSyms[] = {
1062       RTS_SYMBOLS
1063       RTS_RET_SYMBOLS
1064       RTS_POSIX_ONLY_SYMBOLS
1065       RTS_MINGW_ONLY_SYMBOLS
1066       RTS_CYGWIN_ONLY_SYMBOLS
1067       RTS_DARWIN_ONLY_SYMBOLS
1068       RTS_LIBGCC_SYMBOLS
1069       RTS_LIBFFI_SYMBOLS
1070 #if defined(darwin_HOST_OS) && defined(i386_HOST_ARCH)
1071       // dyld stub code contains references to this,
1072       // but it should never be called because we treat
1073       // lazy pointers as nonlazy.
1074       { "dyld_stub_binding_helper", (void*)0xDEADBEEF },
1075 #endif
1076       { 0, 0 } /* sentinel */
1077 };
1078
1079
1080
1081 /* -----------------------------------------------------------------------------
1082  * Insert symbols into hash tables, checking for duplicates.
1083  */
1084
1085 static void ghciInsertStrHashTable ( char* obj_name,
1086                                      HashTable *table,
1087                                      char* key,
1088                                      void *data
1089                                    )
1090 {
1091    if (lookupHashTable(table, (StgWord)key) == NULL)
1092    {
1093       insertStrHashTable(table, (StgWord)key, data);
1094       return;
1095    }
1096    debugBelch(
1097       "\n\n"
1098       "GHCi runtime linker: fatal error: I found a duplicate definition for symbol\n"
1099       "   %s\n"
1100       "whilst processing object file\n"
1101       "   %s\n"
1102       "This could be caused by:\n"
1103       "   * Loading two different object files which export the same symbol\n"
1104       "   * Specifying the same object file twice on the GHCi command line\n"
1105       "   * An incorrect `package.conf' entry, causing some object to be\n"
1106       "     loaded twice.\n"
1107       "GHCi cannot safely continue in this situation.  Exiting now.  Sorry.\n"
1108       "\n",
1109       (char*)key,
1110       obj_name
1111    );
1112    stg_exit(1);
1113 }
1114 /* -----------------------------------------------------------------------------
1115  * initialize the object linker
1116  */
1117
1118
1119 static int linker_init_done = 0 ;
1120
1121 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1122 static void *dl_prog_handle;
1123 static regex_t re_invalid;
1124 static regex_t re_realso;
1125 #ifdef THREADED_RTS
1126 static Mutex dl_mutex; // mutex to protect dlopen/dlerror critical section
1127 #endif
1128 #endif
1129
1130 void
1131 initLinker( void )
1132 {
1133     RtsSymbolVal *sym;
1134 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1135     int compileResult;
1136 #endif
1137
1138     IF_DEBUG(linker, debugBelch("initLinker: start\n"));
1139
1140     /* Make initLinker idempotent, so we can call it
1141        before evey relevant operation; that means we
1142        don't need to initialise the linker separately */
1143     if (linker_init_done == 1) {
1144         IF_DEBUG(linker, debugBelch("initLinker: idempotent return\n"));
1145         return;
1146     } else {
1147         linker_init_done = 1;
1148     }
1149
1150 #if defined(THREADED_RTS) && (defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO))
1151     initMutex(&dl_mutex);
1152 #endif
1153     stablehash = allocStrHashTable();
1154     symhash = allocStrHashTable();
1155
1156     /* populate the symbol table with stuff from the RTS */
1157     for (sym = rtsSyms; sym->lbl != NULL; sym++) {
1158         ghciInsertStrHashTable("(GHCi built-in symbols)",
1159                                symhash, sym->lbl, sym->addr);
1160         IF_DEBUG(linker, debugBelch("initLinker: inserting rts symbol %s, %p\n", sym->lbl, sym->addr));
1161     }
1162 #   if defined(OBJFORMAT_MACHO) && defined(powerpc_HOST_ARCH)
1163     machoInitSymbolsWithoutUnderscore();
1164 #   endif
1165
1166 #   if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1167 #   if defined(RTLD_DEFAULT)
1168     dl_prog_handle = RTLD_DEFAULT;
1169 #   else
1170     dl_prog_handle = dlopen(NULL, RTLD_LAZY);
1171 #   endif /* RTLD_DEFAULT */
1172
1173     compileResult = regcomp(&re_invalid,
1174            "(([^ \t()])+\\.so([^ \t:()])*):([ \t])*invalid ELF header",
1175            REG_EXTENDED);
1176     ASSERT( compileResult == 0 );
1177     compileResult = regcomp(&re_realso,
1178            "GROUP *\\( *(([^ )])+)",
1179            REG_EXTENDED);
1180     ASSERT( compileResult == 0 );
1181 #   endif
1182
1183 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1184     if (RtsFlags.MiscFlags.linkerMemBase != 0) {
1185         // User-override for mmap_32bit_base
1186         mmap_32bit_base = (void*)RtsFlags.MiscFlags.linkerMemBase;
1187     }
1188 #endif
1189
1190 #if defined(mingw32_HOST_OS)
1191     /*
1192      * These two libraries cause problems when added to the static link,
1193      * but are necessary for resolving symbols in GHCi, hence we load
1194      * them manually here.
1195      */
1196     addDLL("msvcrt");
1197     addDLL("kernel32");
1198 #endif
1199
1200     IF_DEBUG(linker, debugBelch("initLinker: done\n"));
1201     return;
1202 }
1203
1204 void
1205 exitLinker( void ) {
1206 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1207    if (linker_init_done == 1) {
1208       regfree(&re_invalid);
1209       regfree(&re_realso);
1210 #ifdef THREADED_RTS
1211       closeMutex(&dl_mutex);
1212 #endif
1213    }
1214 #endif
1215 }
1216
1217 /* -----------------------------------------------------------------------------
1218  *                  Loading DLL or .so dynamic libraries
1219  * -----------------------------------------------------------------------------
1220  *
1221  * Add a DLL from which symbols may be found.  In the ELF case, just
1222  * do RTLD_GLOBAL-style add, so no further messing around needs to
1223  * happen in order that symbols in the loaded .so are findable --
1224  * lookupSymbol() will subsequently see them by dlsym on the program's
1225  * dl-handle.  Returns NULL if success, otherwise ptr to an err msg.
1226  *
1227  * In the PEi386 case, open the DLLs and put handles to them in a
1228  * linked list.  When looking for a symbol, try all handles in the
1229  * list.  This means that we need to load even DLLs that are guaranteed
1230  * to be in the ghc.exe image already, just so we can get a handle
1231  * to give to loadSymbol, so that we can find the symbols.  For such
1232  * libraries, the LoadLibrary call should be a no-op except for returning
1233  * the handle.
1234  *
1235  */
1236
1237 #if defined(OBJFORMAT_PEi386)
1238 /* A record for storing handles into DLLs. */
1239
1240 typedef
1241    struct _OpenedDLL {
1242       char*              name;
1243       struct _OpenedDLL* next;
1244       HINSTANCE instance;
1245    }
1246    OpenedDLL;
1247
1248 /* A list thereof. */
1249 static OpenedDLL* opened_dlls = NULL;
1250 #endif
1251
1252 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1253
1254 static const char *
1255 internal_dlopen(const char *dll_name)
1256 {
1257    void *hdl;
1258    const char *errmsg;
1259    char *errmsg_copy;
1260
1261    // omitted: RTLD_NOW
1262    // see http://www.haskell.org/pipermail/cvs-ghc/2007-September/038570.html
1263    IF_DEBUG(linker,
1264       debugBelch("internal_dlopen: dll_name = '%s'\n", dll_name));
1265
1266    //-------------- Begin critical section ------------------
1267    // This critical section is necessary because dlerror() is not
1268    // required to be reentrant (see POSIX -- IEEE Std 1003.1-2008)
1269    // Also, the error message returned must be copied to preserve it
1270    // (see POSIX also)
1271
1272    ACQUIRE_LOCK(&dl_mutex);
1273    hdl = dlopen(dll_name, RTLD_LAZY | RTLD_GLOBAL);
1274
1275    errmsg = NULL;
1276    if (hdl == NULL) {
1277       /* dlopen failed; return a ptr to the error msg. */
1278       errmsg = dlerror();
1279       if (errmsg == NULL) errmsg = "addDLL: unknown error";
1280       errmsg_copy = stgMallocBytes(strlen(errmsg)+1, "addDLL");
1281       strcpy(errmsg_copy, errmsg);
1282       errmsg = errmsg_copy;
1283    }
1284    RELEASE_LOCK(&dl_mutex);
1285    //--------------- End critical section -------------------
1286
1287    return errmsg;
1288 }
1289 #  endif
1290
1291 const char *
1292 addDLL( char *dll_name )
1293 {
1294 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1295    /* ------------------- ELF DLL loader ------------------- */
1296
1297 #define NMATCH 5
1298    regmatch_t match[NMATCH];
1299    const char *errmsg;
1300    FILE* fp;
1301    size_t match_length;
1302 #define MAXLINE 1000
1303    char line[MAXLINE];
1304    int result;
1305
1306    initLinker();
1307
1308    IF_DEBUG(linker, debugBelch("addDLL: dll_name = '%s'\n", dll_name));
1309    errmsg = internal_dlopen(dll_name);
1310
1311    if (errmsg == NULL) {
1312       return NULL;
1313    }
1314
1315    // GHC Trac ticket #2615
1316    // On some systems (e.g., Gentoo Linux) dynamic files (e.g. libc.so)
1317    // contain linker scripts rather than ELF-format object code. This
1318    // code handles the situation by recognizing the real object code
1319    // file name given in the linker script.
1320    //
1321    // If an "invalid ELF header" error occurs, it is assumed that the
1322    // .so file contains a linker script instead of ELF object code.
1323    // In this case, the code looks for the GROUP ( ... ) linker
1324    // directive. If one is found, the first file name inside the
1325    // parentheses is treated as the name of a dynamic library and the
1326    // code attempts to dlopen that file. If this is also unsuccessful,
1327    // an error message is returned.
1328
1329    // see if the error message is due to an invalid ELF header
1330    IF_DEBUG(linker, debugBelch("errmsg = '%s'\n", errmsg));
1331    result = regexec(&re_invalid, errmsg, (size_t) NMATCH, match, 0);
1332    IF_DEBUG(linker, debugBelch("result = %i\n", result));
1333    if (result == 0) {
1334       // success -- try to read the named file as a linker script
1335       match_length = (size_t) stg_min((match[1].rm_eo - match[1].rm_so),
1336                                  MAXLINE-1);
1337       strncpy(line, (errmsg+(match[1].rm_so)),match_length);
1338       line[match_length] = '\0'; // make sure string is null-terminated
1339       IF_DEBUG(linker, debugBelch ("file name = '%s'\n", line));
1340       if ((fp = fopen(line, "r")) == NULL) {
1341          return errmsg; // return original error if open fails
1342       }
1343       // try to find a GROUP ( ... ) command
1344       while (fgets(line, MAXLINE, fp) != NULL) {
1345          IF_DEBUG(linker, debugBelch("input line = %s", line));
1346          if (regexec(&re_realso, line, (size_t) NMATCH, match, 0) == 0) {
1347             // success -- try to dlopen the first named file
1348             IF_DEBUG(linker, debugBelch("match%s\n",""));
1349             line[match[1].rm_eo] = '\0';
1350             errmsg = internal_dlopen(line+match[1].rm_so);
1351             break;
1352          }
1353          // if control reaches here, no GROUP ( ... ) directive was found
1354          // and the original error message is returned to the caller
1355       }
1356       fclose(fp);
1357    }
1358    return errmsg;
1359
1360 #  elif defined(OBJFORMAT_PEi386)
1361    /* ------------------- Win32 DLL loader ------------------- */
1362
1363    char*      buf;
1364    OpenedDLL* o_dll;
1365    HINSTANCE  instance;
1366
1367    initLinker();
1368
1369    /* debugBelch("\naddDLL; dll_name = `%s'\n", dll_name); */
1370
1371    /* See if we've already got it, and ignore if so. */
1372    for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
1373       if (0 == strcmp(o_dll->name, dll_name))
1374          return NULL;
1375    }
1376
1377    /* The file name has no suffix (yet) so that we can try
1378       both foo.dll and foo.drv
1379
1380       The documentation for LoadLibrary says:
1381         If no file name extension is specified in the lpFileName
1382         parameter, the default library extension .dll is
1383         appended. However, the file name string can include a trailing
1384         point character (.) to indicate that the module name has no
1385         extension. */
1386
1387    buf = stgMallocBytes(strlen(dll_name) + 10, "addDLL");
1388    sprintf(buf, "%s.DLL", dll_name);
1389    instance = LoadLibrary(buf);
1390    if (instance == NULL) {
1391        if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1392        // KAA: allow loading of drivers (like winspool.drv)
1393        sprintf(buf, "%s.DRV", dll_name);
1394        instance = LoadLibrary(buf);
1395        if (instance == NULL) {
1396            if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1397            // #1883: allow loading of unix-style libfoo.dll DLLs
1398            sprintf(buf, "lib%s.DLL", dll_name);
1399            instance = LoadLibrary(buf);
1400            if (instance == NULL) {
1401                goto error;
1402            }
1403        }
1404    }
1405    stgFree(buf);
1406
1407    /* Add this DLL to the list of DLLs in which to search for symbols. */
1408    o_dll = stgMallocBytes( sizeof(OpenedDLL), "addDLL" );
1409    o_dll->name     = stgMallocBytes(1+strlen(dll_name), "addDLL");
1410    strcpy(o_dll->name, dll_name);
1411    o_dll->instance = instance;
1412    o_dll->next     = opened_dlls;
1413    opened_dlls     = o_dll;
1414
1415    return NULL;
1416
1417 error:
1418    stgFree(buf);
1419    sysErrorBelch(dll_name);
1420
1421    /* LoadLibrary failed; return a ptr to the error msg. */
1422    return "addDLL: could not load DLL";
1423
1424 #  else
1425    barf("addDLL: not implemented on this platform");
1426 #  endif
1427 }
1428
1429 /* -----------------------------------------------------------------------------
1430  * insert a stable symbol in the hash table
1431  */
1432
1433 void
1434 insertStableSymbol(char* obj_name, char* key, StgPtr p)
1435 {
1436   ghciInsertStrHashTable(obj_name, stablehash, key, getStablePtr(p));
1437 }
1438
1439
1440 /* -----------------------------------------------------------------------------
1441  * insert a symbol in the hash table
1442  */
1443 void
1444 insertSymbol(char* obj_name, char* key, void* data)
1445 {
1446   ghciInsertStrHashTable(obj_name, symhash, key, data);
1447 }
1448
1449 /* -----------------------------------------------------------------------------
1450  * lookup a symbol in the hash table
1451  */
1452 void *
1453 lookupSymbol( char *lbl )
1454 {
1455     void *val;
1456     IF_DEBUG(linker, debugBelch("lookupSymbol: looking up %s\n", lbl));
1457     initLinker() ;
1458     ASSERT(symhash != NULL);
1459     val = lookupStrHashTable(symhash, lbl);
1460
1461     if (val == NULL) {
1462         IF_DEBUG(linker, debugBelch("lookupSymbol: symbol not found\n"));
1463 #       if defined(OBJFORMAT_ELF)
1464         return dlsym(dl_prog_handle, lbl);
1465 #       elif defined(OBJFORMAT_MACHO)
1466 #       if HAVE_DLFCN_H
1467         /* On OS X 10.3 and later, we use dlsym instead of the old legacy
1468            interface.
1469
1470            HACK: On OS X, global symbols are prefixed with an underscore.
1471                  However, dlsym wants us to omit the leading underscore from the
1472                  symbol name. For now, we simply strip it off here (and ONLY
1473                  here).
1474         */
1475         IF_DEBUG(linker, debugBelch("lookupSymbol: looking up %s with dlsym\n", lbl));
1476         ASSERT(lbl[0] == '_');
1477         return dlsym(dl_prog_handle, lbl+1);
1478 #       else
1479         if(NSIsSymbolNameDefined(lbl)) {
1480             NSSymbol symbol = NSLookupAndBindSymbol(lbl);
1481             return NSAddressOfSymbol(symbol);
1482         } else {
1483             return NULL;
1484         }
1485 #       endif /* HAVE_DLFCN_H */
1486 #       elif defined(OBJFORMAT_PEi386)
1487         void* sym;
1488
1489         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1490         if (sym != NULL) { return sym; };
1491
1492         // Also try looking up the symbol without the @N suffix.  Some
1493         // DLLs have the suffixes on their symbols, some don't.
1494         zapTrailingAtSign ( (unsigned char*)lbl );
1495         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1496         if (sym != NULL) { return sym; };
1497         return NULL;
1498
1499 #       else
1500         ASSERT(2+2 == 5);
1501         return NULL;
1502 #       endif
1503     } else {
1504         IF_DEBUG(linker, debugBelch("lookupSymbol: value of %s is %p\n", lbl, val));
1505         return val;
1506     }
1507 }
1508
1509 /* -----------------------------------------------------------------------------
1510  * Debugging aid: look in GHCi's object symbol tables for symbols
1511  * within DELTA bytes of the specified address, and show their names.
1512  */
1513 #ifdef DEBUG
1514 void ghci_enquire ( char* addr );
1515
1516 void ghci_enquire ( char* addr )
1517 {
1518    int   i;
1519    char* sym;
1520    char* a;
1521    const int DELTA = 64;
1522    ObjectCode* oc;
1523
1524    initLinker();
1525
1526    for (oc = objects; oc; oc = oc->next) {
1527       for (i = 0; i < oc->n_symbols; i++) {
1528          sym = oc->symbols[i];
1529          if (sym == NULL) continue;
1530          a = NULL;
1531          if (a == NULL) {
1532             a = lookupStrHashTable(symhash, sym);
1533          }
1534          if (a == NULL) {
1535              // debugBelch("ghci_enquire: can't find %s\n", sym);
1536          }
1537          else if (addr-DELTA <= a && a <= addr+DELTA) {
1538             debugBelch("%p + %3d  ==  `%s'\n", addr, (int)(a - addr), sym);
1539          }
1540       }
1541    }
1542 }
1543 #endif
1544
1545 #ifdef USE_MMAP
1546 #define ROUND_UP(x,size) ((x + size - 1) & ~(size - 1))
1547
1548 static void *
1549 mmapForLinker (size_t bytes, nat flags, int fd)
1550 {
1551    void *map_addr = NULL;
1552    void *result;
1553    int pagesize, size;
1554    static nat fixed = 0;
1555
1556    pagesize = getpagesize();
1557    size = ROUND_UP(bytes, pagesize);
1558
1559 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1560 mmap_again:
1561
1562    if (mmap_32bit_base != 0) {
1563        map_addr = mmap_32bit_base;
1564    }
1565 #endif
1566
1567    result = mmap(map_addr, size, PROT_EXEC|PROT_READ|PROT_WRITE,
1568                     MAP_PRIVATE|TRY_MAP_32BIT|fixed|flags, fd, 0);
1569
1570    if (result == MAP_FAILED) {
1571        sysErrorBelch("mmap %lu bytes at %p",(lnat)size,map_addr);
1572        errorBelch("Try specifying an address with +RTS -xm<addr> -RTS");
1573        stg_exit(EXIT_FAILURE);
1574    }
1575
1576 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1577    if (mmap_32bit_base != 0) {
1578        if (result == map_addr) {
1579            mmap_32bit_base = (StgWord8*)map_addr + size;
1580        } else {
1581            if ((W_)result > 0x80000000) {
1582                // oops, we were given memory over 2Gb
1583 #if defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS)
1584                // Some platforms require MAP_FIXED.  This is normally
1585                // a bad idea, because MAP_FIXED will overwrite
1586                // existing mappings.
1587                munmap(result,size);
1588                fixed = MAP_FIXED;
1589                goto mmap_again;
1590 #else
1591                barf("loadObj: failed to mmap() memory below 2Gb; asked for %lu bytes at %p.  Try specifying an address with +RTS -xm<addr> -RTS", size, map_addr, result);
1592 #endif
1593            } else {
1594                // hmm, we were given memory somewhere else, but it's
1595                // still under 2Gb so we can use it.  Next time, ask
1596                // for memory right after the place we just got some
1597                mmap_32bit_base = (StgWord8*)result + size;
1598            }
1599        }
1600    } else {
1601        if ((W_)result > 0x80000000) {
1602            // oops, we were given memory over 2Gb
1603            // ... try allocating memory somewhere else?;
1604            debugTrace(DEBUG_linker,"MAP_32BIT didn't work; gave us %lu bytes at 0x%p", bytes, result);
1605            munmap(result, size);
1606
1607            // Set a base address and try again... (guess: 1Gb)
1608            mmap_32bit_base = (void*)0x40000000;
1609            goto mmap_again;
1610        }
1611    }
1612 #endif
1613
1614    return result;
1615 }
1616 #endif // USE_MMAP
1617
1618 static ObjectCode*
1619 mkOc( char *path, char *image, int imageSize,
1620       char *archiveMemberName
1621 #ifndef USE_MMAP
1622 #ifdef darwin_HOST_OS
1623     , int misalignment
1624 #endif
1625 #endif
1626     ) {
1627    ObjectCode* oc;
1628
1629    oc = stgMallocBytes(sizeof(ObjectCode), "loadArchive(oc)");
1630
1631 #  if defined(OBJFORMAT_ELF)
1632    oc->formatName = "ELF";
1633 #  elif defined(OBJFORMAT_PEi386)
1634    oc->formatName = "PEi386";
1635 #  elif defined(OBJFORMAT_MACHO)
1636    oc->formatName = "Mach-O";
1637 #  else
1638    stgFree(oc);
1639    barf("loadObj: not implemented on this platform");
1640 #  endif
1641
1642    oc->image = image;
1643    /* sigh, strdup() isn't a POSIX function, so do it the long way */
1644    oc->fileName = stgMallocBytes( strlen(path)+1, "loadObj" );
1645    strcpy(oc->fileName, path);
1646
1647    if (archiveMemberName) {
1648        oc->archiveMemberName = stgMallocBytes( strlen(archiveMemberName)+1, "loadObj" );
1649        strcpy(oc->archiveMemberName, archiveMemberName);
1650    }
1651    else {
1652        oc->archiveMemberName = NULL;
1653    }
1654
1655    oc->fileSize          = imageSize;
1656    oc->symbols           = NULL;
1657    oc->sections          = NULL;
1658    oc->proddables        = NULL;
1659
1660 #ifndef USE_MMAP
1661 #ifdef darwin_HOST_OS
1662    oc->misalignment = misalignment;
1663 #endif
1664 #endif
1665
1666    /* chain it onto the list of objects */
1667    oc->next              = objects;
1668    objects               = oc;
1669
1670    return oc;
1671 }
1672
1673 HsInt
1674 loadArchive( char *path )
1675 {
1676     ObjectCode* oc;
1677     char *image;
1678     int memberSize;
1679     FILE *f;
1680     int n;
1681     size_t thisFileNameSize;
1682     char *fileName;
1683     size_t fileNameSize;
1684     int isObject, isGnuIndex;
1685     char tmp[12];
1686     char *gnuFileIndex;
1687     int gnuFileIndexSize;
1688
1689     IF_DEBUG(linker, debugBelch("loadArchive: Loading archive `%s'\n", path));
1690
1691     gnuFileIndex = NULL;
1692     gnuFileIndexSize = 0;
1693
1694     fileNameSize = 32;
1695     fileName = stgMallocBytes(fileNameSize, "loadArchive(fileName)");
1696
1697     f = fopen(path, "rb");
1698     if (!f)
1699         barf("loadObj: can't read `%s'", path);
1700
1701     n = fread ( tmp, 1, 8, f );
1702     if (strncmp(tmp, "!<arch>\n", 8) != 0)
1703         barf("loadArchive: Not an archive: `%s'", path);
1704
1705     while(1) {
1706         n = fread ( fileName, 1, 16, f );
1707         if (n != 16) {
1708             if (feof(f)) {
1709                 break;
1710             }
1711             else {
1712                 barf("loadArchive: Failed reading file name from `%s'", path);
1713             }
1714         }
1715         n = fread ( tmp, 1, 12, f );
1716         if (n != 12)
1717             barf("loadArchive: Failed reading mod time from `%s'", path);
1718         n = fread ( tmp, 1, 6, f );
1719         if (n != 6)
1720             barf("loadArchive: Failed reading owner from `%s'", path);
1721         n = fread ( tmp, 1, 6, f );
1722         if (n != 6)
1723             barf("loadArchive: Failed reading group from `%s'", path);
1724         n = fread ( tmp, 1, 8, f );
1725         if (n != 8)
1726             barf("loadArchive: Failed reading mode from `%s'", path);
1727         n = fread ( tmp, 1, 10, f );
1728         if (n != 10)
1729             barf("loadArchive: Failed reading size from `%s'", path);
1730         tmp[10] = '\0';
1731         for (n = 0; isdigit(tmp[n]); n++);
1732         tmp[n] = '\0';
1733         memberSize = atoi(tmp);
1734         n = fread ( tmp, 1, 2, f );
1735         if (strncmp(tmp, "\x60\x0A", 2) != 0)
1736             barf("loadArchive: Failed reading magic from `%s' at %ld. Got %c%c",
1737                  path, ftell(f), tmp[0], tmp[1]);
1738
1739         isGnuIndex = 0;
1740         /* Check for BSD-variant large filenames */
1741         if (0 == strncmp(fileName, "#1/", 3)) {
1742             fileName[16] = '\0';
1743             if (isdigit(fileName[3])) {
1744                 for (n = 4; isdigit(fileName[n]); n++);
1745                 fileName[n] = '\0';
1746                 thisFileNameSize = atoi(fileName + 3);
1747                 memberSize -= thisFileNameSize;
1748                 if (thisFileNameSize >= fileNameSize) {
1749                     /* Double it to avoid potentially continually
1750                        increasing it by 1 */
1751                     fileNameSize = thisFileNameSize * 2;
1752                     fileName = stgReallocBytes(fileName, fileNameSize, "loadArchive(fileName)");
1753                 }
1754                 n = fread ( fileName, 1, thisFileNameSize, f );
1755                 if (n != (int)thisFileNameSize) {
1756                     barf("loadArchive: Failed reading filename from `%s'",
1757                          path);
1758                 }
1759                 fileName[thisFileNameSize] = 0;
1760             }
1761             else {
1762                 barf("loadArchive: BSD-variant filename size not found while reading filename from `%s'", path);
1763             }
1764         }
1765         /* Check for GNU file index file */
1766         else if (0 == strncmp(fileName, "//", 2)) {
1767             fileName[0] = '\0';
1768             thisFileNameSize = 0;
1769             isGnuIndex = 1;
1770         }
1771         /* Check for a file in the GNU file index */
1772         else if (fileName[0] == '/') {
1773             if (isdigit(fileName[1])) {
1774                 int i;
1775
1776                 for (n = 2; isdigit(fileName[n]); n++);
1777                 fileName[n] = '\0';
1778                 n = atoi(fileName + 1);
1779
1780                 if (gnuFileIndex == NULL) {
1781                     barf("loadArchive: GNU-variant filename without an index while reading from `%s'", path);
1782                 }
1783                 if (n < 0 || n > gnuFileIndexSize) {
1784                     barf("loadArchive: GNU-variant filename offset %d out of range [0..%d] while reading filename from `%s'", n, gnuFileIndexSize, path);
1785                 }
1786                 if (n != 0 && gnuFileIndex[n - 1] != '\n') {
1787                     barf("loadArchive: GNU-variant filename offset %d invalid (range [0..%d]) while reading filename from `%s'", n, gnuFileIndexSize, path);
1788                 }
1789                 for (i = n; gnuFileIndex[i] != '/'; i++);
1790                 thisFileNameSize = i - n;
1791                 if (thisFileNameSize >= fileNameSize) {
1792                     /* Double it to avoid potentially continually
1793                        increasing it by 1 */
1794                     fileNameSize = thisFileNameSize * 2;
1795                     fileName = stgReallocBytes(fileName, fileNameSize, "loadArchive(fileName)");
1796                 }
1797                 memcpy(fileName, gnuFileIndex + n, thisFileNameSize);
1798                 fileName[thisFileNameSize] = '\0';
1799             }
1800             else if (fileName[1] == ' ') {
1801                 fileName[0] = '\0';
1802                 thisFileNameSize = 0;
1803             }
1804             else {
1805                 barf("loadArchive: GNU-variant filename offset not found while reading filename from `%s'", path);
1806             }
1807         }
1808         /* Finally, the case where the filename field actually contains
1809            the filename */
1810         else {
1811             /* GNU ar terminates filenames with a '/', this allowing
1812                spaces in filenames. So first look to see if there is a
1813                terminating '/'. */
1814             for (thisFileNameSize = 0;
1815                  thisFileNameSize < 16;
1816                  thisFileNameSize++) {
1817                 if (fileName[thisFileNameSize] == '/') {
1818                     fileName[thisFileNameSize] = '\0';
1819                     break;
1820                 }
1821             }
1822             /* If we didn't find a '/', then a space teminates the
1823                filename. Note that if we don't find one, then
1824                thisFileNameSize ends up as 16, and we already have the
1825                '\0' at the end. */
1826             if (thisFileNameSize == 16) {
1827                 for (thisFileNameSize = 0;
1828                      thisFileNameSize < 16;
1829                      thisFileNameSize++) {
1830                     if (fileName[thisFileNameSize] == ' ') {
1831                         fileName[thisFileNameSize] = '\0';
1832                         break;
1833                     }
1834                 }
1835             }
1836         }
1837
1838         IF_DEBUG(linker,
1839                  debugBelch("loadArchive: Found member file `%s'\n", fileName));
1840
1841         isObject = thisFileNameSize >= 2
1842                 && fileName[thisFileNameSize - 2] == '.'
1843                 && fileName[thisFileNameSize - 1] == 'o';
1844
1845         if (isObject) {
1846             char *archiveMemberName;
1847
1848             IF_DEBUG(linker, debugBelch("loadArchive: Member is an object file...loading...\n"));
1849
1850             /* We can't mmap from the archive directly, as object
1851                files need to be 8-byte aligned but files in .ar
1852                archives are 2-byte aligned. When possible we use mmap
1853                to get some anonymous memory, as on 64-bit platforms if
1854                we use malloc then we can be given memory above 2^32.
1855                In the mmap case we're probably wasting lots of space;
1856                we could do better. */
1857 #ifdef USE_MMAP
1858             image = mmapForLinker(memberSize, MAP_ANONYMOUS, -1);
1859 #else
1860             image = stgMallocBytes(memberSize, "loadArchive(image)");
1861 #endif
1862             n = fread ( image, 1, memberSize, f );
1863             if (n != memberSize) {
1864                 barf("loadArchive: error whilst reading `%s'", path);
1865             }
1866
1867             archiveMemberName = stgMallocBytes(strlen(path) + thisFileNameSize + 3,
1868                                                "loadArchive(file)");
1869             sprintf(archiveMemberName, "%s(%.*s)",
1870                     path, (int)thisFileNameSize, fileName);
1871
1872             oc = mkOc(path, image, memberSize, archiveMemberName
1873 #ifndef USE_MMAP
1874 #ifdef darwin_HOST_OS
1875                      , 0
1876 #endif
1877 #endif
1878                      );
1879
1880             stgFree(archiveMemberName);
1881
1882             if (0 == loadOc(oc)) {
1883                 stgFree(fileName);
1884                 return 0;
1885             }
1886         }
1887         else if (isGnuIndex) {
1888             if (gnuFileIndex != NULL) {
1889                 barf("loadArchive: GNU-variant index found, but already have an index, while reading filename from `%s'", path);
1890             }
1891             IF_DEBUG(linker, debugBelch("loadArchive: Found GNU-variant file index\n"));
1892 #ifdef USE_MMAP
1893             gnuFileIndex = mmapForLinker(memberSize + 1, MAP_ANONYMOUS, -1);
1894 #else
1895             gnuFileIndex = stgMallocBytes(memberSize + 1, "loadArchive(image)");
1896 #endif
1897             n = fread ( gnuFileIndex, 1, memberSize, f );
1898             if (n != memberSize) {
1899                 barf("loadArchive: error whilst reading `%s'", path);
1900             }
1901             gnuFileIndex[memberSize] = '/';
1902             gnuFileIndexSize = memberSize;
1903         }
1904         else {
1905             n = fseek(f, memberSize, SEEK_CUR);
1906             if (n != 0)
1907                 barf("loadArchive: error whilst seeking by %d in `%s'",
1908                      memberSize, path);
1909         }
1910         /* .ar files are 2-byte aligned */
1911         if (memberSize % 2) {
1912             n = fread ( tmp, 1, 1, f );
1913             if (n != 1) {
1914                 if (feof(f)) {
1915                     break;
1916                 }
1917                 else {
1918                     barf("loadArchive: Failed reading padding from `%s'", path);
1919                 }
1920             }
1921         }
1922     }
1923
1924     fclose(f);
1925
1926     stgFree(fileName);
1927     if (gnuFileIndex != NULL) {
1928 #ifdef USE_MMAP
1929         munmap(gnuFileIndex, gnuFileIndexSize + 1);
1930 #else
1931         stgFree(gnuFileIndex);
1932 #endif
1933     }
1934
1935     return 1;
1936 }
1937
1938 /* -----------------------------------------------------------------------------
1939  * Load an obj (populate the global symbol table, but don't resolve yet)
1940  *
1941  * Returns: 1 if ok, 0 on error.
1942  */
1943 HsInt
1944 loadObj( char *path )
1945 {
1946    ObjectCode* oc;
1947    char *image;
1948    int fileSize;
1949    struct stat st;
1950    int r;
1951 #ifdef USE_MMAP
1952    int fd;
1953 #else
1954    FILE *f;
1955 #endif
1956    IF_DEBUG(linker, debugBelch("loadObj %s\n", path));
1957
1958    initLinker();
1959
1960    /* debugBelch("loadObj %s\n", path ); */
1961
1962    /* Check that we haven't already loaded this object.
1963       Ignore requests to load multiple times */
1964    {
1965        ObjectCode *o;
1966        int is_dup = 0;
1967        for (o = objects; o; o = o->next) {
1968           if (0 == strcmp(o->fileName, path)) {
1969              is_dup = 1;
1970              break; /* don't need to search further */
1971           }
1972        }
1973        if (is_dup) {
1974           IF_DEBUG(linker, debugBelch(
1975             "GHCi runtime linker: warning: looks like you're trying to load the\n"
1976             "same object file twice:\n"
1977             "   %s\n"
1978             "GHCi will ignore this, but be warned.\n"
1979             , path));
1980           return 1; /* success */
1981        }
1982    }
1983
1984    r = stat(path, &st);
1985    if (r == -1) {
1986        IF_DEBUG(linker, debugBelch("File doesn't exist\n"));
1987        return 0;
1988    }
1989
1990    fileSize = st.st_size;
1991
1992 #ifdef USE_MMAP
1993    /* On many architectures malloc'd memory isn't executable, so we need to use mmap. */
1994
1995 #if defined(openbsd_HOST_OS)
1996    fd = open(path, O_RDONLY, S_IRUSR);
1997 #else
1998    fd = open(path, O_RDONLY);
1999 #endif
2000    if (fd == -1)
2001       barf("loadObj: can't open `%s'", path);
2002
2003    image = mmapForLinker(fileSize, 0, fd);
2004
2005    close(fd);
2006
2007 #else /* !USE_MMAP */
2008    /* load the image into memory */
2009    f = fopen(path, "rb");
2010    if (!f)
2011        barf("loadObj: can't read `%s'", path);
2012
2013 #   if defined(mingw32_HOST_OS)
2014         // TODO: We would like to use allocateExec here, but allocateExec
2015         //       cannot currently allocate blocks large enough.
2016     image = VirtualAlloc(NULL, fileSize, MEM_RESERVE | MEM_COMMIT,
2017                              PAGE_EXECUTE_READWRITE);
2018 #   elif defined(darwin_HOST_OS)
2019     // In a Mach-O .o file, all sections can and will be misaligned
2020     // if the total size of the headers is not a multiple of the
2021     // desired alignment. This is fine for .o files that only serve
2022     // as input for the static linker, but it's not fine for us,
2023     // as SSE (used by gcc for floating point) and Altivec require
2024     // 16-byte alignment.
2025     // We calculate the correct alignment from the header before
2026     // reading the file, and then we misalign image on purpose so
2027     // that the actual sections end up aligned again.
2028    misalignment = machoGetMisalignment(f);
2029    image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
2030    image += misalignment;
2031 #  else
2032    image = stgMallocBytes(fileSize, "loadObj(image)");
2033 #  endif
2034
2035    {
2036        int n;
2037        n = fread ( image, 1, fileSize, f );
2038        if (n != fileSize)
2039            barf("loadObj: error whilst reading `%s'", path);
2040    }
2041    fclose(f);
2042 #endif /* USE_MMAP */
2043
2044    oc = mkOc(path, image, fileSize, NULL
2045 #ifndef USE_MMAP
2046 #ifdef darwin_HOST_OS
2047             , misalignment
2048 #endif
2049 #endif
2050             );
2051
2052    return loadOc(oc);
2053 }
2054
2055 static HsInt
2056 loadOc( ObjectCode* oc ) {
2057    int r;
2058
2059    IF_DEBUG(linker, debugBelch("loadOc\n"));
2060
2061 #  if defined(OBJFORMAT_MACHO) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
2062    r = ocAllocateSymbolExtras_MachO ( oc );
2063    if (!r) {
2064        IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO failed\n"));
2065        return r;
2066    }
2067 #  elif defined(OBJFORMAT_ELF) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
2068    r = ocAllocateSymbolExtras_ELF ( oc );
2069    if (!r) {
2070        IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_ELF failed\n"));
2071        return r;
2072    }
2073 #endif
2074
2075    /* verify the in-memory image */
2076 #  if defined(OBJFORMAT_ELF)
2077    r = ocVerifyImage_ELF ( oc );
2078 #  elif defined(OBJFORMAT_PEi386)
2079    r = ocVerifyImage_PEi386 ( oc );
2080 #  elif defined(OBJFORMAT_MACHO)
2081    r = ocVerifyImage_MachO ( oc );
2082 #  else
2083    barf("loadObj: no verify method");
2084 #  endif
2085    if (!r) {
2086        IF_DEBUG(linker, debugBelch("ocVerifyImage_* failed\n"));
2087        return r;
2088    }
2089
2090    /* build the symbol list for this image */
2091 #  if defined(OBJFORMAT_ELF)
2092    r = ocGetNames_ELF ( oc );
2093 #  elif defined(OBJFORMAT_PEi386)
2094    r = ocGetNames_PEi386 ( oc );
2095 #  elif defined(OBJFORMAT_MACHO)
2096    r = ocGetNames_MachO ( oc );
2097 #  else
2098    barf("loadObj: no getNames method");
2099 #  endif
2100    if (!r) {
2101        IF_DEBUG(linker, debugBelch("ocGetNames_* failed\n"));
2102        return r;
2103    }
2104
2105    /* loaded, but not resolved yet */
2106    oc->status = OBJECT_LOADED;
2107    IF_DEBUG(linker, debugBelch("loadObj done.\n"));
2108
2109    return 1;
2110 }
2111
2112 /* -----------------------------------------------------------------------------
2113  * resolve all the currently unlinked objects in memory
2114  *
2115  * Returns: 1 if ok, 0 on error.
2116  */
2117 HsInt
2118 resolveObjs( void )
2119 {
2120     ObjectCode *oc;
2121     int r;
2122
2123     IF_DEBUG(linker, debugBelch("resolveObjs: start\n"));
2124     initLinker();
2125
2126     for (oc = objects; oc; oc = oc->next) {
2127         if (oc->status != OBJECT_RESOLVED) {
2128 #           if defined(OBJFORMAT_ELF)
2129             r = ocResolve_ELF ( oc );
2130 #           elif defined(OBJFORMAT_PEi386)
2131             r = ocResolve_PEi386 ( oc );
2132 #           elif defined(OBJFORMAT_MACHO)
2133             r = ocResolve_MachO ( oc );
2134 #           else
2135             barf("resolveObjs: not implemented on this platform");
2136 #           endif
2137             if (!r) { return r; }
2138             oc->status = OBJECT_RESOLVED;
2139         }
2140     }
2141     IF_DEBUG(linker, debugBelch("resolveObjs: done\n"));
2142     return 1;
2143 }
2144
2145 /* -----------------------------------------------------------------------------
2146  * delete an object from the pool
2147  */
2148 HsInt
2149 unloadObj( char *path )
2150 {
2151     ObjectCode *oc, *prev;
2152     HsBool unloadedAnyObj = HS_BOOL_FALSE;
2153
2154     ASSERT(symhash != NULL);
2155     ASSERT(objects != NULL);
2156
2157     initLinker();
2158
2159     prev = NULL;
2160     for (oc = objects; oc; prev = oc, oc = oc->next) {
2161         if (!strcmp(oc->fileName,path)) {
2162
2163             /* Remove all the mappings for the symbols within this
2164              * object..
2165              */
2166             {
2167                 int i;
2168                 for (i = 0; i < oc->n_symbols; i++) {
2169                    if (oc->symbols[i] != NULL) {
2170                        removeStrHashTable(symhash, oc->symbols[i], NULL);
2171                    }
2172                 }
2173             }
2174
2175             if (prev == NULL) {
2176                 objects = oc->next;
2177             } else {
2178                 prev->next = oc->next;
2179             }
2180
2181             // We're going to leave this in place, in case there are
2182             // any pointers from the heap into it:
2183                 // #ifdef mingw32_HOST_OS
2184                 //  VirtualFree(oc->image);
2185                 // #else
2186             //  stgFree(oc->image);
2187             // #endif
2188             stgFree(oc->fileName);
2189             stgFree(oc->symbols);
2190             stgFree(oc->sections);
2191             stgFree(oc);
2192
2193             /* This could be a member of an archive so continue
2194              * unloading other members. */
2195             unloadedAnyObj = HS_BOOL_TRUE;
2196         }
2197     }
2198
2199     if (unloadedAnyObj) {
2200         return 1;
2201     }
2202     else {
2203         errorBelch("unloadObj: can't find `%s' to unload", path);
2204         return 0;
2205     }
2206 }
2207
2208 /* -----------------------------------------------------------------------------
2209  * Sanity checking.  For each ObjectCode, maintain a list of address ranges
2210  * which may be prodded during relocation, and abort if we try and write
2211  * outside any of these.
2212  */
2213 static void addProddableBlock ( ObjectCode* oc, void* start, int size )
2214 {
2215    ProddableBlock* pb
2216       = stgMallocBytes(sizeof(ProddableBlock), "addProddableBlock");
2217    IF_DEBUG(linker, debugBelch("addProddableBlock %p %p %d\n", oc, start, size));
2218    ASSERT(size > 0);
2219    pb->start      = start;
2220    pb->size       = size;
2221    pb->next       = oc->proddables;
2222    oc->proddables = pb;
2223 }
2224
2225 static void checkProddableBlock ( ObjectCode* oc, void* addr )
2226 {
2227    ProddableBlock* pb;
2228    for (pb = oc->proddables; pb != NULL; pb = pb->next) {
2229       char* s = (char*)(pb->start);
2230       char* e = s + pb->size - 1;
2231       char* a = (char*)addr;
2232       /* Assumes that the biggest fixup involves a 4-byte write.  This
2233          probably needs to be changed to 8 (ie, +7) on 64-bit
2234          plats. */
2235       if (a >= s && (a+3) <= e) return;
2236    }
2237    barf("checkProddableBlock: invalid fixup in runtime linker");
2238 }
2239
2240 /* -----------------------------------------------------------------------------
2241  * Section management.
2242  */
2243 static void addSection ( ObjectCode* oc, SectionKind kind,
2244                          void* start, void* end )
2245 {
2246    Section* s   = stgMallocBytes(sizeof(Section), "addSection");
2247    s->start     = start;
2248    s->end       = end;
2249    s->kind      = kind;
2250    s->next      = oc->sections;
2251    oc->sections = s;
2252    /*
2253    debugBelch("addSection: %p-%p (size %d), kind %d\n",
2254                    start, ((char*)end)-1, end - start + 1, kind );
2255    */
2256 }
2257
2258
2259 /* --------------------------------------------------------------------------
2260  * Symbol Extras.
2261  * This is about allocating a small chunk of memory for every symbol in the
2262  * object file. We make sure that the SymboLExtras are always "in range" of
2263  * limited-range PC-relative instructions on various platforms by allocating
2264  * them right next to the object code itself.
2265  */
2266
2267 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
2268
2269 /*
2270   ocAllocateSymbolExtras
2271
2272   Allocate additional space at the end of the object file image to make room
2273   for jump islands (powerpc, x86_64) and GOT entries (x86_64).
2274
2275   PowerPC relative branch instructions have a 24 bit displacement field.
2276   As PPC code is always 4-byte-aligned, this yields a +-32MB range.
2277   If a particular imported symbol is outside this range, we have to redirect
2278   the jump to a short piece of new code that just loads the 32bit absolute
2279   address and jumps there.
2280   On x86_64, PC-relative jumps and PC-relative accesses to the GOT are limited
2281   to 32 bits (+-2GB).
2282
2283   This function just allocates space for one SymbolExtra for every
2284   undefined symbol in the object file. The code for the jump islands is
2285   filled in by makeSymbolExtra below.
2286 */
2287
2288 static int ocAllocateSymbolExtras( ObjectCode* oc, int count, int first )
2289 {
2290 #ifdef USE_MMAP
2291   int pagesize, n, m;
2292 #endif
2293   int aligned;
2294 #ifndef USE_MMAP
2295   int misalignment = 0;
2296 #ifdef darwin_HOST_OS
2297   misalignment = oc->misalignment;
2298 #endif
2299 #endif
2300
2301   if( count > 0 )
2302   {
2303     // round up to the nearest 4
2304     aligned = (oc->fileSize + 3) & ~3;
2305
2306 #ifdef USE_MMAP
2307     pagesize = getpagesize();
2308     n = ROUND_UP( oc->fileSize, pagesize );
2309     m = ROUND_UP( aligned + sizeof (SymbolExtra) * count, pagesize );
2310
2311     /* we try to use spare space at the end of the last page of the
2312      * image for the jump islands, but if there isn't enough space
2313      * then we have to map some (anonymously, remembering MAP_32BIT).
2314      */
2315     if( m > n ) // we need to allocate more pages
2316     {
2317         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
2318                                           MAP_ANONYMOUS, -1);
2319     }
2320     else
2321     {
2322         oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
2323     }
2324 #else
2325     oc->image -= misalignment;
2326     oc->image = stgReallocBytes( oc->image,
2327                                  misalignment +
2328                                  aligned + sizeof (SymbolExtra) * count,
2329                                  "ocAllocateSymbolExtras" );
2330     oc->image += misalignment;
2331
2332     oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
2333 #endif /* USE_MMAP */
2334
2335     memset( oc->symbol_extras, 0, sizeof (SymbolExtra) * count );
2336   }
2337   else
2338     oc->symbol_extras = NULL;
2339
2340   oc->first_symbol_extra = first;
2341   oc->n_symbol_extras = count;
2342
2343   return 1;
2344 }
2345
2346 static SymbolExtra* makeSymbolExtra( ObjectCode* oc,
2347                                      unsigned long symbolNumber,
2348                                      unsigned long target )
2349 {
2350   SymbolExtra *extra;
2351
2352   ASSERT( symbolNumber >= oc->first_symbol_extra
2353         && symbolNumber - oc->first_symbol_extra < oc->n_symbol_extras);
2354
2355   extra = &oc->symbol_extras[symbolNumber - oc->first_symbol_extra];
2356
2357 #ifdef powerpc_HOST_ARCH
2358   // lis r12, hi16(target)
2359   extra->jumpIsland.lis_r12     = 0x3d80;
2360   extra->jumpIsland.hi_addr     = target >> 16;
2361
2362   // ori r12, r12, lo16(target)
2363   extra->jumpIsland.ori_r12_r12 = 0x618c;
2364   extra->jumpIsland.lo_addr     = target & 0xffff;
2365
2366   // mtctr r12
2367   extra->jumpIsland.mtctr_r12   = 0x7d8903a6;
2368
2369   // bctr
2370   extra->jumpIsland.bctr        = 0x4e800420;
2371 #endif
2372 #ifdef x86_64_HOST_ARCH
2373         // jmp *-14(%rip)
2374   static uint8_t jmp[] = { 0xFF, 0x25, 0xF2, 0xFF, 0xFF, 0xFF };
2375   extra->addr = target;
2376   memcpy(extra->jumpIsland, jmp, 6);
2377 #endif
2378
2379   return extra;
2380 }
2381
2382 #endif
2383
2384 /* --------------------------------------------------------------------------
2385  * PowerPC specifics (instruction cache flushing)
2386  * ------------------------------------------------------------------------*/
2387
2388 #ifdef powerpc_HOST_ARCH
2389 /*
2390    ocFlushInstructionCache
2391
2392    Flush the data & instruction caches.
2393    Because the PPC has split data/instruction caches, we have to
2394    do that whenever we modify code at runtime.
2395  */
2396 static void ocFlushInstructionCacheFrom(void* begin, size_t length)
2397 {
2398     size_t         n = (length + 3) / 4;
2399     unsigned long* p = begin;
2400
2401     while (n--)
2402     {
2403         __asm__ volatile ( "dcbf 0,%0\n\t"
2404                            "sync\n\t"
2405                            "icbi 0,%0"
2406                            :
2407                            : "r" (p)
2408                          );
2409         p++;
2410     }
2411     __asm__ volatile ( "sync\n\t"
2412                        "isync"
2413                      );
2414 }
2415 static void ocFlushInstructionCache( ObjectCode *oc )
2416 {
2417     /* The main object code */
2418     ocFlushInstructionCacheFrom(oc->image + oc->misalignment, oc->fileSize);
2419
2420     /* Jump Islands */
2421     ocFlushInstructionCacheFrom(oc->symbol_extras, sizeof(SymbolExtra) * oc->n_symbol_extras);
2422 }
2423 #endif
2424
2425 /* --------------------------------------------------------------------------
2426  * PEi386 specifics (Win32 targets)
2427  * ------------------------------------------------------------------------*/
2428
2429 /* The information for this linker comes from
2430       Microsoft Portable Executable
2431       and Common Object File Format Specification
2432       revision 5.1 January 1998
2433    which SimonM says comes from the MS Developer Network CDs.
2434
2435    It can be found there (on older CDs), but can also be found
2436    online at:
2437
2438       http://www.microsoft.com/hwdev/hardware/PECOFF.asp
2439
2440    (this is Rev 6.0 from February 1999).
2441
2442    Things move, so if that fails, try searching for it via
2443
2444       http://www.google.com/search?q=PE+COFF+specification
2445
2446    The ultimate reference for the PE format is the Winnt.h
2447    header file that comes with the Platform SDKs; as always,
2448    implementations will drift wrt their documentation.
2449
2450    A good background article on the PE format is Matt Pietrek's
2451    March 1994 article in Microsoft System Journal (MSJ)
2452    (Vol.9, No. 3): "Peering Inside the PE: A Tour of the
2453    Win32 Portable Executable File Format." The info in there
2454    has recently been updated in a two part article in
2455    MSDN magazine, issues Feb and March 2002,
2456    "Inside Windows: An In-Depth Look into the Win32 Portable
2457    Executable File Format"
2458
2459    John Levine's book "Linkers and Loaders" contains useful
2460    info on PE too.
2461 */
2462
2463
2464 #if defined(OBJFORMAT_PEi386)
2465
2466
2467
2468 typedef unsigned char  UChar;
2469 typedef unsigned short UInt16;
2470 typedef unsigned int   UInt32;
2471 typedef          int   Int32;
2472
2473
2474 typedef
2475    struct {
2476       UInt16 Machine;
2477       UInt16 NumberOfSections;
2478       UInt32 TimeDateStamp;
2479       UInt32 PointerToSymbolTable;
2480       UInt32 NumberOfSymbols;
2481       UInt16 SizeOfOptionalHeader;
2482       UInt16 Characteristics;
2483    }
2484    COFF_header;
2485
2486 #define sizeof_COFF_header 20
2487
2488
2489 typedef
2490    struct {
2491       UChar  Name[8];
2492       UInt32 VirtualSize;
2493       UInt32 VirtualAddress;
2494       UInt32 SizeOfRawData;
2495       UInt32 PointerToRawData;
2496       UInt32 PointerToRelocations;
2497       UInt32 PointerToLinenumbers;
2498       UInt16 NumberOfRelocations;
2499       UInt16 NumberOfLineNumbers;
2500       UInt32 Characteristics;
2501    }
2502    COFF_section;
2503
2504 #define sizeof_COFF_section 40
2505
2506
2507 typedef
2508    struct {
2509       UChar  Name[8];
2510       UInt32 Value;
2511       UInt16 SectionNumber;
2512       UInt16 Type;
2513       UChar  StorageClass;
2514       UChar  NumberOfAuxSymbols;
2515    }
2516    COFF_symbol;
2517
2518 #define sizeof_COFF_symbol 18
2519
2520
2521 typedef
2522    struct {
2523       UInt32 VirtualAddress;
2524       UInt32 SymbolTableIndex;
2525       UInt16 Type;
2526    }
2527    COFF_reloc;
2528
2529 #define sizeof_COFF_reloc 10
2530
2531
2532 /* From PE spec doc, section 3.3.2 */
2533 /* Note use of MYIMAGE_* since IMAGE_* are already defined in
2534    windows.h -- for the same purpose, but I want to know what I'm
2535    getting, here. */
2536 #define MYIMAGE_FILE_RELOCS_STRIPPED     0x0001
2537 #define MYIMAGE_FILE_EXECUTABLE_IMAGE    0x0002
2538 #define MYIMAGE_FILE_DLL                 0x2000
2539 #define MYIMAGE_FILE_SYSTEM              0x1000
2540 #define MYIMAGE_FILE_BYTES_REVERSED_HI   0x8000
2541 #define MYIMAGE_FILE_BYTES_REVERSED_LO   0x0080
2542 #define MYIMAGE_FILE_32BIT_MACHINE       0x0100
2543
2544 /* From PE spec doc, section 5.4.2 and 5.4.4 */
2545 #define MYIMAGE_SYM_CLASS_EXTERNAL       2
2546 #define MYIMAGE_SYM_CLASS_STATIC         3
2547 #define MYIMAGE_SYM_UNDEFINED            0
2548
2549 /* From PE spec doc, section 4.1 */
2550 #define MYIMAGE_SCN_CNT_CODE             0x00000020
2551 #define MYIMAGE_SCN_CNT_INITIALIZED_DATA 0x00000040
2552 #define MYIMAGE_SCN_LNK_NRELOC_OVFL      0x01000000
2553
2554 /* From PE spec doc, section 5.2.1 */
2555 #define MYIMAGE_REL_I386_DIR32           0x0006
2556 #define MYIMAGE_REL_I386_REL32           0x0014
2557
2558
2559 /* We use myindex to calculate array addresses, rather than
2560    simply doing the normal subscript thing.  That's because
2561    some of the above structs have sizes which are not
2562    a whole number of words.  GCC rounds their sizes up to a
2563    whole number of words, which means that the address calcs
2564    arising from using normal C indexing or pointer arithmetic
2565    are just plain wrong.  Sigh.
2566 */
2567 static UChar *
2568 myindex ( int scale, void* base, int index )
2569 {
2570    return
2571       ((UChar*)base) + scale * index;
2572 }
2573
2574
2575 static void
2576 printName ( UChar* name, UChar* strtab )
2577 {
2578    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2579       UInt32 strtab_offset = * (UInt32*)(name+4);
2580       debugBelch("%s", strtab + strtab_offset );
2581    } else {
2582       int i;
2583       for (i = 0; i < 8; i++) {
2584          if (name[i] == 0) break;
2585          debugBelch("%c", name[i] );
2586       }
2587    }
2588 }
2589
2590
2591 static void
2592 copyName ( UChar* name, UChar* strtab, UChar* dst, int dstSize )
2593 {
2594    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2595       UInt32 strtab_offset = * (UInt32*)(name+4);
2596       strncpy ( (char*)dst, (char*)strtab+strtab_offset, dstSize );
2597       dst[dstSize-1] = 0;
2598    } else {
2599       int i = 0;
2600       while (1) {
2601          if (i >= 8) break;
2602          if (name[i] == 0) break;
2603          dst[i] = name[i];
2604          i++;
2605       }
2606       dst[i] = 0;
2607    }
2608 }
2609
2610
2611 static UChar *
2612 cstring_from_COFF_symbol_name ( UChar* name, UChar* strtab )
2613 {
2614    UChar* newstr;
2615    /* If the string is longer than 8 bytes, look in the
2616       string table for it -- this will be correctly zero terminated.
2617    */
2618    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2619       UInt32 strtab_offset = * (UInt32*)(name+4);
2620       return ((UChar*)strtab) + strtab_offset;
2621    }
2622    /* Otherwise, if shorter than 8 bytes, return the original,
2623       which by defn is correctly terminated.
2624    */
2625    if (name[7]==0) return name;
2626    /* The annoying case: 8 bytes.  Copy into a temporary
2627       (XXX which is never freed ...)
2628    */
2629    newstr = stgMallocBytes(9, "cstring_from_COFF_symbol_name");
2630    ASSERT(newstr);
2631    strncpy((char*)newstr,(char*)name,8);
2632    newstr[8] = 0;
2633    return newstr;
2634 }
2635
2636 /* Getting the name of a section is mildly tricky, so we make a
2637    function for it.  Sadly, in one case we have to copy the string
2638    (when it is exactly 8 bytes long there's no trailing '\0'), so for
2639    consistency we *always* copy the string; the caller must free it
2640 */
2641 static char *
2642 cstring_from_section_name (UChar* name, UChar* strtab)
2643 {
2644     char *newstr;
2645
2646     if (name[0]=='/') {
2647         int strtab_offset = strtol((char*)name+1,NULL,10);
2648         int len = strlen(((char*)strtab) + strtab_offset);
2649
2650         newstr = stgMallocBytes(len, "cstring_from_section_symbol_name");
2651         strcpy((char*)newstr, (char*)((UChar*)strtab) + strtab_offset);
2652         return newstr;
2653     }
2654     else
2655     {
2656         newstr = stgMallocBytes(9, "cstring_from_section_symbol_name");
2657         ASSERT(newstr);
2658         strncpy((char*)newstr,(char*)name,8);
2659         newstr[8] = 0;
2660         return newstr;
2661     }
2662 }
2663
2664 /* Just compares the short names (first 8 chars) */
2665 static COFF_section *
2666 findPEi386SectionCalled ( ObjectCode* oc,  UChar* name )
2667 {
2668    int i;
2669    COFF_header* hdr
2670       = (COFF_header*)(oc->image);
2671    COFF_section* sectab
2672       = (COFF_section*) (
2673            ((UChar*)(oc->image))
2674            + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2675         );
2676    for (i = 0; i < hdr->NumberOfSections; i++) {
2677       UChar* n1;
2678       UChar* n2;
2679       COFF_section* section_i
2680          = (COFF_section*)
2681            myindex ( sizeof_COFF_section, sectab, i );
2682       n1 = (UChar*) &(section_i->Name);
2683       n2 = name;
2684       if (n1[0]==n2[0] && n1[1]==n2[1] && n1[2]==n2[2] &&
2685           n1[3]==n2[3] && n1[4]==n2[4] && n1[5]==n2[5] &&
2686           n1[6]==n2[6] && n1[7]==n2[7])
2687          return section_i;
2688    }
2689
2690    return NULL;
2691 }
2692
2693
2694 static void
2695 zapTrailingAtSign ( UChar* sym )
2696 {
2697 #  define my_isdigit(c) ((c) >= '0' && (c) <= '9')
2698    int i, j;
2699    if (sym[0] == 0) return;
2700    i = 0;
2701    while (sym[i] != 0) i++;
2702    i--;
2703    j = i;
2704    while (j > 0 && my_isdigit(sym[j])) j--;
2705    if (j > 0 && sym[j] == '@' && j != i) sym[j] = 0;
2706 #  undef my_isdigit
2707 }
2708
2709 static void *
2710 lookupSymbolInDLLs ( UChar *lbl )
2711 {
2712     OpenedDLL* o_dll;
2713     void *sym;
2714
2715     for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
2716         /* debugBelch("look in %s for %s\n", o_dll->name, lbl); */
2717
2718         if (lbl[0] == '_') {
2719             /* HACK: if the name has an initial underscore, try stripping
2720                it off & look that up first. I've yet to verify whether there's
2721                a Rule that governs whether an initial '_' *should always* be
2722                stripped off when mapping from import lib name to the DLL name.
2723             */
2724             sym = GetProcAddress(o_dll->instance, (char*)(lbl+1));
2725             if (sym != NULL) {
2726                 /*debugBelch("found %s in %s\n", lbl+1,o_dll->name);*/
2727                 return sym;
2728             }
2729         }
2730         sym = GetProcAddress(o_dll->instance, (char*)lbl);
2731         if (sym != NULL) {
2732             /*debugBelch("found %s in %s\n", lbl,o_dll->name);*/
2733             return sym;
2734            }
2735     }
2736     return NULL;
2737 }
2738
2739
2740 static int
2741 ocVerifyImage_PEi386 ( ObjectCode* oc )
2742 {
2743    int i;
2744    UInt32 j, noRelocs;
2745    COFF_header*  hdr;
2746    COFF_section* sectab;
2747    COFF_symbol*  symtab;
2748    UChar*        strtab;
2749    /* debugBelch("\nLOADING %s\n", oc->fileName); */
2750    hdr = (COFF_header*)(oc->image);
2751    sectab = (COFF_section*) (
2752                ((UChar*)(oc->image))
2753                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2754             );
2755    symtab = (COFF_symbol*) (
2756                ((UChar*)(oc->image))
2757                + hdr->PointerToSymbolTable
2758             );
2759    strtab = ((UChar*)symtab)
2760             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2761
2762    if (hdr->Machine != 0x14c) {
2763       errorBelch("%s: Not x86 PEi386", oc->fileName);
2764       return 0;
2765    }
2766    if (hdr->SizeOfOptionalHeader != 0) {
2767       errorBelch("%s: PEi386 with nonempty optional header", oc->fileName);
2768       return 0;
2769    }
2770    if ( /* (hdr->Characteristics & MYIMAGE_FILE_RELOCS_STRIPPED) || */
2771         (hdr->Characteristics & MYIMAGE_FILE_EXECUTABLE_IMAGE) ||
2772         (hdr->Characteristics & MYIMAGE_FILE_DLL) ||
2773         (hdr->Characteristics & MYIMAGE_FILE_SYSTEM) ) {
2774       errorBelch("%s: Not a PEi386 object file", oc->fileName);
2775       return 0;
2776    }
2777    if ( (hdr->Characteristics & MYIMAGE_FILE_BYTES_REVERSED_HI)
2778         /* || !(hdr->Characteristics & MYIMAGE_FILE_32BIT_MACHINE) */ ) {
2779       errorBelch("%s: Invalid PEi386 word size or endiannness: %d",
2780                  oc->fileName,
2781                  (int)(hdr->Characteristics));
2782       return 0;
2783    }
2784    /* If the string table size is way crazy, this might indicate that
2785       there are more than 64k relocations, despite claims to the
2786       contrary.  Hence this test. */
2787    /* debugBelch("strtab size %d\n", * (UInt32*)strtab); */
2788 #if 0
2789    if ( (*(UInt32*)strtab) > 600000 ) {
2790       /* Note that 600k has no special significance other than being
2791          big enough to handle the almost-2MB-sized lumps that
2792          constitute HSwin32*.o. */
2793       debugBelch("PEi386 object has suspiciously large string table; > 64k relocs?");
2794       return 0;
2795    }
2796 #endif
2797
2798    /* No further verification after this point; only debug printing. */
2799    i = 0;
2800    IF_DEBUG(linker, i=1);
2801    if (i == 0) return 1;
2802
2803    debugBelch( "sectab offset = %d\n", ((UChar*)sectab) - ((UChar*)hdr) );
2804    debugBelch( "symtab offset = %d\n", ((UChar*)symtab) - ((UChar*)hdr) );
2805    debugBelch( "strtab offset = %d\n", ((UChar*)strtab) - ((UChar*)hdr) );
2806
2807    debugBelch("\n" );
2808    debugBelch( "Machine:           0x%x\n", (UInt32)(hdr->Machine) );
2809    debugBelch( "# sections:        %d\n",   (UInt32)(hdr->NumberOfSections) );
2810    debugBelch( "time/date:         0x%x\n", (UInt32)(hdr->TimeDateStamp) );
2811    debugBelch( "symtab offset:     %d\n",   (UInt32)(hdr->PointerToSymbolTable) );
2812    debugBelch( "# symbols:         %d\n",   (UInt32)(hdr->NumberOfSymbols) );
2813    debugBelch( "sz of opt hdr:     %d\n",   (UInt32)(hdr->SizeOfOptionalHeader) );
2814    debugBelch( "characteristics:   0x%x\n", (UInt32)(hdr->Characteristics) );
2815
2816    /* Print the section table. */
2817    debugBelch("\n" );
2818    for (i = 0; i < hdr->NumberOfSections; i++) {
2819       COFF_reloc* reltab;
2820       COFF_section* sectab_i
2821          = (COFF_section*)
2822            myindex ( sizeof_COFF_section, sectab, i );
2823       debugBelch(
2824                 "\n"
2825                 "section %d\n"
2826                 "     name `",
2827                 i
2828               );
2829       printName ( sectab_i->Name, strtab );
2830       debugBelch(
2831                 "'\n"
2832                 "    vsize %d\n"
2833                 "    vaddr %d\n"
2834                 "  data sz %d\n"
2835                 " data off %d\n"
2836                 "  num rel %d\n"
2837                 "  off rel %d\n"
2838                 "  ptr raw 0x%x\n",
2839                 sectab_i->VirtualSize,
2840                 sectab_i->VirtualAddress,
2841                 sectab_i->SizeOfRawData,
2842                 sectab_i->PointerToRawData,
2843                 sectab_i->NumberOfRelocations,
2844                 sectab_i->PointerToRelocations,
2845                 sectab_i->PointerToRawData
2846               );
2847       reltab = (COFF_reloc*) (
2848                   ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2849                );
2850
2851       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2852         /* If the relocation field (a short) has overflowed, the
2853          * real count can be found in the first reloc entry.
2854          *
2855          * See Section 4.1 (last para) of the PE spec (rev6.0).
2856          */
2857         COFF_reloc* rel = (COFF_reloc*)
2858                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2859         noRelocs = rel->VirtualAddress;
2860         j = 1;
2861       } else {
2862         noRelocs = sectab_i->NumberOfRelocations;
2863         j = 0;
2864       }
2865
2866       for (; j < noRelocs; j++) {
2867          COFF_symbol* sym;
2868          COFF_reloc* rel = (COFF_reloc*)
2869                            myindex ( sizeof_COFF_reloc, reltab, j );
2870          debugBelch(
2871                    "        type 0x%-4x   vaddr 0x%-8x   name `",
2872                    (UInt32)rel->Type,
2873                    rel->VirtualAddress );
2874          sym = (COFF_symbol*)
2875                myindex ( sizeof_COFF_symbol, symtab, rel->SymbolTableIndex );
2876          /* Hmm..mysterious looking offset - what's it for? SOF */
2877          printName ( sym->Name, strtab -10 );
2878          debugBelch("'\n" );
2879       }
2880
2881       debugBelch("\n" );
2882    }
2883    debugBelch("\n" );
2884    debugBelch("string table has size 0x%x\n", * (UInt32*)strtab );
2885    debugBelch("---START of string table---\n");
2886    for (i = 4; i < *(Int32*)strtab; i++) {
2887       if (strtab[i] == 0)
2888          debugBelch("\n"); else
2889          debugBelch("%c", strtab[i] );
2890    }
2891    debugBelch("--- END  of string table---\n");
2892
2893    debugBelch("\n" );
2894    i = 0;
2895    while (1) {
2896       COFF_symbol* symtab_i;
2897       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2898       symtab_i = (COFF_symbol*)
2899                  myindex ( sizeof_COFF_symbol, symtab, i );
2900       debugBelch(
2901                 "symbol %d\n"
2902                 "     name `",
2903                 i
2904               );
2905       printName ( symtab_i->Name, strtab );
2906       debugBelch(
2907                 "'\n"
2908                 "    value 0x%x\n"
2909                 "   1+sec# %d\n"
2910                 "     type 0x%x\n"
2911                 "   sclass 0x%x\n"
2912                 "     nAux %d\n",
2913                 symtab_i->Value,
2914                 (Int32)(symtab_i->SectionNumber),
2915                 (UInt32)symtab_i->Type,
2916                 (UInt32)symtab_i->StorageClass,
2917                 (UInt32)symtab_i->NumberOfAuxSymbols
2918               );
2919       i += symtab_i->NumberOfAuxSymbols;
2920       i++;
2921    }
2922
2923    debugBelch("\n" );
2924    return 1;
2925 }
2926
2927
2928 static int
2929 ocGetNames_PEi386 ( ObjectCode* oc )
2930 {
2931    COFF_header*  hdr;
2932    COFF_section* sectab;
2933    COFF_symbol*  symtab;
2934    UChar*        strtab;
2935
2936    UChar* sname;
2937    void*  addr;
2938    int    i;
2939
2940    hdr = (COFF_header*)(oc->image);
2941    sectab = (COFF_section*) (
2942                ((UChar*)(oc->image))
2943                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2944             );
2945    symtab = (COFF_symbol*) (
2946                ((UChar*)(oc->image))
2947                + hdr->PointerToSymbolTable
2948             );
2949    strtab = ((UChar*)(oc->image))
2950             + hdr->PointerToSymbolTable
2951             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2952
2953    /* Allocate space for any (local, anonymous) .bss sections. */
2954
2955    for (i = 0; i < hdr->NumberOfSections; i++) {
2956       UInt32 bss_sz;
2957       UChar* zspace;
2958       COFF_section* sectab_i
2959          = (COFF_section*)
2960            myindex ( sizeof_COFF_section, sectab, i );
2961
2962       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
2963
2964       if (0 != strcmp(secname, ".bss")) {
2965           stgFree(secname);
2966           continue;
2967       }
2968
2969       stgFree(secname);
2970
2971       /* sof 10/05: the PE spec text isn't too clear regarding what
2972        * the SizeOfRawData field is supposed to hold for object
2973        * file sections containing just uninitialized data -- for executables,
2974        * it is supposed to be zero; unclear what it's supposed to be
2975        * for object files. However, VirtualSize is guaranteed to be
2976        * zero for object files, which definitely suggests that SizeOfRawData
2977        * will be non-zero (where else would the size of this .bss section be
2978        * stored?) Looking at the COFF_section info for incoming object files,
2979        * this certainly appears to be the case.
2980        *
2981        * => I suspect we've been incorrectly handling .bss sections in (relocatable)
2982        * object files up until now. This turned out to bite us with ghc-6.4.1's use
2983        * of gcc-3.4.x, which has started to emit initially-zeroed-out local 'static'
2984        * variable decls into to the .bss section. (The specific function in Q which
2985        * triggered this is libraries/base/cbits/dirUtils.c:__hscore_getFolderPath())
2986        */
2987       if (sectab_i->VirtualSize == 0 && sectab_i->SizeOfRawData == 0) continue;
2988       /* This is a non-empty .bss section.  Allocate zeroed space for
2989          it, and set its PointerToRawData field such that oc->image +
2990          PointerToRawData == addr_of_zeroed_space.  */
2991       bss_sz = sectab_i->VirtualSize;
2992       if ( bss_sz < sectab_i->SizeOfRawData) { bss_sz = sectab_i->SizeOfRawData; }
2993       zspace = stgCallocBytes(1, bss_sz, "ocGetNames_PEi386(anonymous bss)");
2994       sectab_i->PointerToRawData = ((UChar*)zspace) - ((UChar*)(oc->image));
2995       addProddableBlock(oc, zspace, bss_sz);
2996       /* debugBelch("BSS anon section at 0x%x\n", zspace); */
2997    }
2998
2999    /* Copy section information into the ObjectCode. */
3000
3001    for (i = 0; i < hdr->NumberOfSections; i++) {
3002       UChar* start;
3003       UChar* end;
3004       UInt32 sz;
3005
3006       SectionKind kind
3007          = SECTIONKIND_OTHER;
3008       COFF_section* sectab_i
3009          = (COFF_section*)
3010            myindex ( sizeof_COFF_section, sectab, i );
3011
3012       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3013
3014       IF_DEBUG(linker, debugBelch("section name = %s\n", secname ));
3015
3016 #     if 0
3017       /* I'm sure this is the Right Way to do it.  However, the
3018          alternative of testing the sectab_i->Name field seems to
3019          work ok with Cygwin.
3020       */
3021       if (sectab_i->Characteristics & MYIMAGE_SCN_CNT_CODE ||
3022           sectab_i->Characteristics & MYIMAGE_SCN_CNT_INITIALIZED_DATA)
3023          kind = SECTIONKIND_CODE_OR_RODATA;
3024 #     endif
3025
3026       if (0==strcmp(".text",(char*)secname) ||
3027           0==strcmp(".rdata",(char*)secname)||
3028           0==strcmp(".rodata",(char*)secname))
3029          kind = SECTIONKIND_CODE_OR_RODATA;
3030       if (0==strcmp(".data",(char*)secname) ||
3031           0==strcmp(".bss",(char*)secname))
3032          kind = SECTIONKIND_RWDATA;
3033
3034       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
3035       sz = sectab_i->SizeOfRawData;
3036       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
3037
3038       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
3039       end   = start + sz - 1;
3040
3041       if (kind == SECTIONKIND_OTHER
3042           /* Ignore sections called which contain stabs debugging
3043              information. */
3044           && 0 != strcmp(".stab", (char*)secname)
3045           && 0 != strcmp(".stabstr", (char*)secname)
3046           /* ignore constructor section for now */
3047           && 0 != strcmp(".ctors", (char*)secname)
3048           /* ignore section generated from .ident */
3049           && 0!= strncmp(".debug", (char*)secname, 6)
3050           /* ignore unknown section that appeared in gcc 3.4.5(?) */
3051           && 0!= strcmp(".reloc", (char*)secname)
3052           && 0 != strcmp(".rdata$zzz", (char*)secname)
3053          ) {
3054          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", secname, oc->fileName);
3055          stgFree(secname);
3056          return 0;
3057       }
3058
3059       if (kind != SECTIONKIND_OTHER && end >= start) {
3060          addSection(oc, kind, start, end);
3061          addProddableBlock(oc, start, end - start + 1);
3062       }
3063
3064       stgFree(secname);
3065    }
3066
3067    /* Copy exported symbols into the ObjectCode. */
3068
3069    oc->n_symbols = hdr->NumberOfSymbols;
3070    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
3071                                   "ocGetNames_PEi386(oc->symbols)");
3072    /* Call me paranoid; I don't care. */
3073    for (i = 0; i < oc->n_symbols; i++)
3074       oc->symbols[i] = NULL;
3075
3076    i = 0;
3077    while (1) {
3078       COFF_symbol* symtab_i;
3079       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
3080       symtab_i = (COFF_symbol*)
3081                  myindex ( sizeof_COFF_symbol, symtab, i );
3082
3083       addr  = NULL;
3084
3085       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
3086           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
3087          /* This symbol is global and defined, viz, exported */
3088          /* for MYIMAGE_SYMCLASS_EXTERNAL
3089                 && !MYIMAGE_SYM_UNDEFINED,
3090             the address of the symbol is:
3091                 address of relevant section + offset in section
3092          */
3093          COFF_section* sectabent
3094             = (COFF_section*) myindex ( sizeof_COFF_section,
3095                                         sectab,
3096                                         symtab_i->SectionNumber-1 );
3097          addr = ((UChar*)(oc->image))
3098                 + (sectabent->PointerToRawData
3099                    + symtab_i->Value);
3100       }
3101       else
3102       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
3103           && symtab_i->Value > 0) {
3104          /* This symbol isn't in any section at all, ie, global bss.
3105             Allocate zeroed space for it. */
3106          addr = stgCallocBytes(1, symtab_i->Value,
3107                                "ocGetNames_PEi386(non-anonymous bss)");
3108          addSection(oc, SECTIONKIND_RWDATA, addr,
3109                         ((UChar*)addr) + symtab_i->Value - 1);
3110          addProddableBlock(oc, addr, symtab_i->Value);
3111          /* debugBelch("BSS      section at 0x%x\n", addr); */
3112       }
3113
3114       if (addr != NULL ) {
3115          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
3116          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
3117          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
3118          ASSERT(i >= 0 && i < oc->n_symbols);
3119          /* cstring_from_COFF_symbol_name always succeeds. */
3120          oc->symbols[i] = (char*)sname;
3121          ghciInsertStrHashTable(oc->fileName, symhash, (char*)sname, addr);
3122       } else {
3123 #        if 0
3124          debugBelch(
3125                    "IGNORING symbol %d\n"
3126                    "     name `",
3127                    i
3128                  );
3129          printName ( symtab_i->Name, strtab );
3130          debugBelch(
3131                    "'\n"
3132                    "    value 0x%x\n"
3133                    "   1+sec# %d\n"
3134                    "     type 0x%x\n"
3135                    "   sclass 0x%x\n"
3136                    "     nAux %d\n",
3137                    symtab_i->Value,
3138                    (Int32)(symtab_i->SectionNumber),
3139                    (UInt32)symtab_i->Type,
3140                    (UInt32)symtab_i->StorageClass,
3141                    (UInt32)symtab_i->NumberOfAuxSymbols
3142                  );
3143 #        endif
3144       }
3145
3146       i += symtab_i->NumberOfAuxSymbols;
3147       i++;
3148    }
3149
3150    return 1;
3151 }
3152
3153
3154 static int
3155 ocResolve_PEi386 ( ObjectCode* oc )
3156 {
3157    COFF_header*  hdr;
3158    COFF_section* sectab;
3159    COFF_symbol*  symtab;
3160    UChar*        strtab;
3161
3162    UInt32        A;
3163    UInt32        S;
3164    UInt32*       pP;
3165
3166    int i;
3167    UInt32 j, noRelocs;
3168
3169    /* ToDo: should be variable-sized?  But is at least safe in the
3170       sense of buffer-overrun-proof. */
3171    UChar symbol[1000];
3172    /* debugBelch("resolving for %s\n", oc->fileName); */
3173
3174    hdr = (COFF_header*)(oc->image);
3175    sectab = (COFF_section*) (
3176                ((UChar*)(oc->image))
3177                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
3178             );
3179    symtab = (COFF_symbol*) (
3180                ((UChar*)(oc->image))
3181                + hdr->PointerToSymbolTable
3182             );
3183    strtab = ((UChar*)(oc->image))
3184             + hdr->PointerToSymbolTable
3185             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
3186
3187    for (i = 0; i < hdr->NumberOfSections; i++) {
3188       COFF_section* sectab_i
3189          = (COFF_section*)
3190            myindex ( sizeof_COFF_section, sectab, i );
3191       COFF_reloc* reltab
3192          = (COFF_reloc*) (
3193               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
3194            );
3195
3196       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3197
3198       /* Ignore sections called which contain stabs debugging
3199          information. */
3200       if (0 == strcmp(".stab", (char*)secname)
3201           || 0 == strcmp(".stabstr", (char*)secname)
3202           || 0 == strcmp(".ctors", (char*)secname)
3203           || 0 == strncmp(".debug", (char*)secname, 6)
3204           || 0 == strcmp(".rdata$zzz", (char*)secname)) {
3205           stgFree(secname);
3206           continue;
3207       }
3208
3209       stgFree(secname);
3210
3211       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
3212         /* If the relocation field (a short) has overflowed, the
3213          * real count can be found in the first reloc entry.
3214          *
3215          * See Section 4.1 (last para) of the PE spec (rev6.0).
3216          *
3217          * Nov2003 update: the GNU linker still doesn't correctly
3218          * handle the generation of relocatable object files with
3219          * overflown relocations. Hence the output to warn of potential
3220          * troubles.
3221          */
3222         COFF_reloc* rel = (COFF_reloc*)
3223                            myindex ( sizeof_COFF_reloc, reltab, 0 );
3224         noRelocs = rel->VirtualAddress;
3225
3226         /* 10/05: we now assume (and check for) a GNU ld that is capable
3227          * of handling object files with (>2^16) of relocs.
3228          */
3229 #if 0
3230         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
3231                    noRelocs);
3232 #endif
3233         j = 1;
3234       } else {
3235         noRelocs = sectab_i->NumberOfRelocations;
3236         j = 0;
3237       }
3238
3239
3240       for (; j < noRelocs; j++) {
3241          COFF_symbol* sym;
3242          COFF_reloc* reltab_j
3243             = (COFF_reloc*)
3244               myindex ( sizeof_COFF_reloc, reltab, j );
3245
3246          /* the location to patch */
3247          pP = (UInt32*)(
3248                  ((UChar*)(oc->image))
3249                  + (sectab_i->PointerToRawData
3250                     + reltab_j->VirtualAddress
3251                     - sectab_i->VirtualAddress )
3252               );
3253          /* the existing contents of pP */
3254          A = *pP;
3255          /* the symbol to connect to */
3256          sym = (COFF_symbol*)
3257                myindex ( sizeof_COFF_symbol,
3258                          symtab, reltab_j->SymbolTableIndex );
3259          IF_DEBUG(linker,
3260                   debugBelch(
3261                             "reloc sec %2d num %3d:  type 0x%-4x   "
3262                             "vaddr 0x%-8x   name `",
3263                             i, j,
3264                             (UInt32)reltab_j->Type,
3265                             reltab_j->VirtualAddress );
3266                             printName ( sym->Name, strtab );
3267                             debugBelch("'\n" ));
3268
3269          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
3270             COFF_section* section_sym
3271                = findPEi386SectionCalled ( oc, sym->Name );
3272             if (!section_sym) {
3273                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
3274                return 0;
3275             }
3276             S = ((UInt32)(oc->image))
3277                 + (section_sym->PointerToRawData
3278                    + sym->Value);
3279          } else {
3280             copyName ( sym->Name, strtab, symbol, 1000-1 );
3281             S = (UInt32) lookupSymbol( (char*)symbol );
3282             if ((void*)S != NULL) goto foundit;
3283             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3284             return 0;
3285            foundit:;
3286          }
3287          checkProddableBlock(oc, pP);
3288          switch (reltab_j->Type) {
3289             case MYIMAGE_REL_I386_DIR32:
3290                *pP = A + S;
3291                break;
3292             case MYIMAGE_REL_I386_REL32:
3293                /* Tricky.  We have to insert a displacement at
3294                   pP which, when added to the PC for the _next_
3295                   insn, gives the address of the target (S).
3296                   Problem is to know the address of the next insn
3297                   when we only know pP.  We assume that this
3298                   literal field is always the last in the insn,
3299                   so that the address of the next insn is pP+4
3300                   -- hence the constant 4.
3301                   Also I don't know if A should be added, but so
3302                   far it has always been zero.
3303
3304                   SOF 05/2005: 'A' (old contents of *pP) have been observed
3305                   to contain values other than zero (the 'wx' object file
3306                   that came with wxhaskell-0.9.4; dunno how it was compiled..).
3307                   So, add displacement to old value instead of asserting
3308                   A to be zero. Fixes wxhaskell-related crashes, and no other
3309                   ill effects have been observed.
3310
3311                   Update: the reason why we're seeing these more elaborate
3312                   relocations is due to a switch in how the NCG compiles SRTs
3313                   and offsets to them from info tables. SRTs live in .(ro)data,
3314                   while info tables live in .text, causing GAS to emit REL32/DISP32
3315                   relocations with non-zero values. Adding the displacement is
3316                   the right thing to do.
3317                */
3318                *pP = S - ((UInt32)pP) - 4 + A;
3319                break;
3320             default:
3321                debugBelch("%s: unhandled PEi386 relocation type %d",
3322                      oc->fileName, reltab_j->Type);
3323                return 0;
3324          }
3325
3326       }
3327    }
3328
3329    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
3330    return 1;
3331 }
3332
3333 #endif /* defined(OBJFORMAT_PEi386) */
3334
3335
3336 /* --------------------------------------------------------------------------
3337  * ELF specifics
3338  * ------------------------------------------------------------------------*/
3339
3340 #if defined(OBJFORMAT_ELF)
3341
3342 #define FALSE 0
3343 #define TRUE  1
3344
3345 #if defined(sparc_HOST_ARCH)
3346 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
3347 #elif defined(i386_HOST_ARCH)
3348 #  define ELF_TARGET_386    /* Used inside <elf.h> */
3349 #elif defined(x86_64_HOST_ARCH)
3350 #  define ELF_TARGET_X64_64
3351 #  define ELF_64BIT
3352 #endif
3353
3354 #if !defined(openbsd_HOST_OS)
3355 #  include <elf.h>
3356 #else
3357 /* openbsd elf has things in different places, with diff names */
3358 #  include <elf_abi.h>
3359 #  include <machine/reloc.h>
3360 #  define R_386_32    RELOC_32
3361 #  define R_386_PC32  RELOC_PC32
3362 #endif
3363
3364 /* If elf.h doesn't define it */
3365 #  ifndef R_X86_64_PC64
3366 #    define R_X86_64_PC64 24
3367 #  endif
3368
3369 /*
3370  * Define a set of types which can be used for both ELF32 and ELF64
3371  */
3372
3373 #ifdef ELF_64BIT
3374 #define ELFCLASS    ELFCLASS64
3375 #define Elf_Addr    Elf64_Addr
3376 #define Elf_Word    Elf64_Word
3377 #define Elf_Sword   Elf64_Sword
3378 #define Elf_Ehdr    Elf64_Ehdr
3379 #define Elf_Phdr    Elf64_Phdr
3380 #define Elf_Shdr    Elf64_Shdr
3381 #define Elf_Sym     Elf64_Sym
3382 #define Elf_Rel     Elf64_Rel
3383 #define Elf_Rela    Elf64_Rela
3384 #ifndef ELF_ST_TYPE
3385 #define ELF_ST_TYPE ELF64_ST_TYPE
3386 #endif
3387 #ifndef ELF_ST_BIND
3388 #define ELF_ST_BIND ELF64_ST_BIND
3389 #endif
3390 #ifndef ELF_R_TYPE
3391 #define ELF_R_TYPE  ELF64_R_TYPE
3392 #endif
3393 #ifndef ELF_R_SYM
3394 #define ELF_R_SYM   ELF64_R_SYM
3395 #endif
3396 #else
3397 #define ELFCLASS    ELFCLASS32
3398 #define Elf_Addr    Elf32_Addr
3399 #define Elf_Word    Elf32_Word
3400 #define Elf_Sword   Elf32_Sword
3401 #define Elf_Ehdr    Elf32_Ehdr
3402 #define Elf_Phdr    Elf32_Phdr
3403 #define Elf_Shdr    Elf32_Shdr
3404 #define Elf_Sym     Elf32_Sym
3405 #define Elf_Rel     Elf32_Rel
3406 #define Elf_Rela    Elf32_Rela
3407 #ifndef ELF_ST_TYPE
3408 #define ELF_ST_TYPE ELF32_ST_TYPE
3409 #endif
3410 #ifndef ELF_ST_BIND
3411 #define ELF_ST_BIND ELF32_ST_BIND
3412 #endif
3413 #ifndef ELF_R_TYPE
3414 #define ELF_R_TYPE  ELF32_R_TYPE
3415 #endif
3416 #ifndef ELF_R_SYM
3417 #define ELF_R_SYM   ELF32_R_SYM
3418 #endif
3419 #endif
3420
3421
3422 /*
3423  * Functions to allocate entries in dynamic sections.  Currently we simply
3424  * preallocate a large number, and we don't check if a entry for the given
3425  * target already exists (a linear search is too slow).  Ideally these
3426  * entries would be associated with symbols.
3427  */
3428
3429 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
3430 #define GOT_SIZE            0x20000
3431 #define FUNCTION_TABLE_SIZE 0x10000
3432 #define PLT_SIZE            0x08000
3433
3434 #ifdef ELF_NEED_GOT
3435 static Elf_Addr got[GOT_SIZE];
3436 static unsigned int gotIndex;
3437 static Elf_Addr gp_val = (Elf_Addr)got;
3438
3439 static Elf_Addr
3440 allocateGOTEntry(Elf_Addr target)
3441 {
3442    Elf_Addr *entry;
3443
3444    if (gotIndex >= GOT_SIZE)
3445       barf("Global offset table overflow");
3446
3447    entry = &got[gotIndex++];
3448    *entry = target;
3449    return (Elf_Addr)entry;
3450 }
3451 #endif
3452
3453 #ifdef ELF_FUNCTION_DESC
3454 typedef struct {
3455    Elf_Addr ip;
3456    Elf_Addr gp;
3457 } FunctionDesc;
3458
3459 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
3460 static unsigned int functionTableIndex;
3461
3462 static Elf_Addr
3463 allocateFunctionDesc(Elf_Addr target)
3464 {
3465    FunctionDesc *entry;
3466
3467    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
3468       barf("Function table overflow");
3469
3470    entry = &functionTable[functionTableIndex++];
3471    entry->ip = target;
3472    entry->gp = (Elf_Addr)gp_val;
3473    return (Elf_Addr)entry;
3474 }
3475
3476 static Elf_Addr
3477 copyFunctionDesc(Elf_Addr target)
3478 {
3479    FunctionDesc *olddesc = (FunctionDesc *)target;
3480    FunctionDesc *newdesc;
3481
3482    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
3483    newdesc->gp = olddesc->gp;
3484    return (Elf_Addr)newdesc;
3485 }
3486 #endif
3487
3488 #ifdef ELF_NEED_PLT
3489
3490 typedef struct {
3491    unsigned char code[sizeof(plt_code)];
3492 } PLTEntry;
3493
3494 static Elf_Addr
3495 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
3496 {
3497    PLTEntry *plt = (PLTEntry *)oc->plt;
3498    PLTEntry *entry;
3499
3500    if (oc->pltIndex >= PLT_SIZE)
3501       barf("Procedure table overflow");
3502
3503    entry = &plt[oc->pltIndex++];
3504    memcpy(entry->code, plt_code, sizeof(entry->code));
3505    PLT_RELOC(entry->code, target);
3506    return (Elf_Addr)entry;
3507 }
3508
3509 static unsigned int
3510 PLTSize(void)
3511 {
3512    return (PLT_SIZE * sizeof(PLTEntry));
3513 }
3514 #endif
3515
3516
3517 /*
3518  * Generic ELF functions
3519  */
3520
3521 static char *
3522 findElfSection ( void* objImage, Elf_Word sh_type )
3523 {
3524    char* ehdrC = (char*)objImage;
3525    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
3526    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
3527    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3528    char* ptr = NULL;
3529    int i;
3530
3531    for (i = 0; i < ehdr->e_shnum; i++) {
3532       if (shdr[i].sh_type == sh_type
3533           /* Ignore the section header's string table. */
3534           && i != ehdr->e_shstrndx
3535           /* Ignore string tables named .stabstr, as they contain
3536              debugging info. */
3537           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3538          ) {
3539          ptr = ehdrC + shdr[i].sh_offset;
3540          break;
3541       }
3542    }
3543    return ptr;
3544 }
3545
3546 static int
3547 ocVerifyImage_ELF ( ObjectCode* oc )
3548 {
3549    Elf_Shdr* shdr;
3550    Elf_Sym*  stab;
3551    int i, j, nent, nstrtab, nsymtabs;
3552    char* sh_strtab;
3553    char* strtab;
3554
3555    char*     ehdrC = (char*)(oc->image);
3556    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
3557
3558    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
3559        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
3560        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
3561        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
3562       errorBelch("%s: not an ELF object", oc->fileName);
3563       return 0;
3564    }
3565
3566    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
3567       errorBelch("%s: unsupported ELF format", oc->fileName);
3568       return 0;
3569    }
3570
3571    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
3572        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
3573    } else
3574    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
3575        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
3576    } else {
3577        errorBelch("%s: unknown endiannness", oc->fileName);
3578        return 0;
3579    }
3580
3581    if (ehdr->e_type != ET_REL) {
3582       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
3583       return 0;
3584    }
3585    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
3586
3587    IF_DEBUG(linker,debugBelch( "Architecture is " ));
3588    switch (ehdr->e_machine) {
3589       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
3590 #ifdef EM_SPARC32PLUS
3591       case EM_SPARC32PLUS:
3592 #endif
3593       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
3594 #ifdef EM_IA_64
3595       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
3596 #endif
3597       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
3598 #ifdef EM_X86_64
3599       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
3600 #elif defined(EM_AMD64)
3601       case EM_AMD64: IF_DEBUG(linker,debugBelch( "amd64" )); break;
3602 #endif
3603       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
3604                      errorBelch("%s: unknown architecture (e_machine == %d)"
3605                                 , oc->fileName, ehdr->e_machine);
3606                      return 0;
3607    }
3608
3609    IF_DEBUG(linker,debugBelch(
3610              "\nSection header table: start %ld, n_entries %d, ent_size %d\n",
3611              (long)ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
3612
3613    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
3614
3615    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3616
3617    if (ehdr->e_shstrndx == SHN_UNDEF) {
3618       errorBelch("%s: no section header string table", oc->fileName);
3619       return 0;
3620    } else {
3621       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
3622                           ehdr->e_shstrndx));
3623       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3624    }
3625
3626    for (i = 0; i < ehdr->e_shnum; i++) {
3627       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
3628       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
3629       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
3630       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
3631       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
3632                ehdrC + shdr[i].sh_offset,
3633                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
3634
3635       if (shdr[i].sh_type == SHT_REL) {
3636           IF_DEBUG(linker,debugBelch("Rel  " ));
3637       } else if (shdr[i].sh_type == SHT_RELA) {
3638           IF_DEBUG(linker,debugBelch("RelA " ));
3639       } else {
3640           IF_DEBUG(linker,debugBelch("     "));
3641       }
3642       if (sh_strtab) {
3643           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
3644       }
3645    }
3646
3647    IF_DEBUG(linker,debugBelch( "\nString tables" ));
3648    strtab = NULL;
3649    nstrtab = 0;
3650    for (i = 0; i < ehdr->e_shnum; i++) {
3651       if (shdr[i].sh_type == SHT_STRTAB
3652           /* Ignore the section header's string table. */
3653           && i != ehdr->e_shstrndx
3654           /* Ignore string tables named .stabstr, as they contain
3655              debugging info. */
3656           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3657          ) {
3658          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
3659          strtab = ehdrC + shdr[i].sh_offset;
3660          nstrtab++;
3661       }
3662    }
3663    if (nstrtab != 1) {
3664       errorBelch("%s: no string tables, or too many", oc->fileName);
3665       return 0;
3666    }
3667
3668    nsymtabs = 0;
3669    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
3670    for (i = 0; i < ehdr->e_shnum; i++) {
3671       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3672       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
3673       nsymtabs++;
3674       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3675       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3676       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%ld rem)\n",
3677                nent,
3678                (long)shdr[i].sh_size % sizeof(Elf_Sym)
3679              ));
3680       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
3681          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
3682          return 0;
3683       }
3684       for (j = 0; j < nent; j++) {
3685          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
3686          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
3687                              (int)stab[j].st_shndx,
3688                              (int)stab[j].st_size,
3689                              (char*)stab[j].st_value ));
3690
3691          IF_DEBUG(linker,debugBelch("type=" ));
3692          switch (ELF_ST_TYPE(stab[j].st_info)) {
3693             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
3694             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
3695             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
3696             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
3697             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
3698             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
3699          }
3700          IF_DEBUG(linker,debugBelch("  " ));
3701
3702          IF_DEBUG(linker,debugBelch("bind=" ));
3703          switch (ELF_ST_BIND(stab[j].st_info)) {
3704             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
3705             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
3706             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
3707             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
3708          }
3709          IF_DEBUG(linker,debugBelch("  " ));
3710
3711          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
3712       }
3713    }
3714
3715    if (nsymtabs == 0) {
3716       errorBelch("%s: didn't find any symbol tables", oc->fileName);
3717       return 0;
3718    }
3719
3720    return 1;
3721 }
3722
3723 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
3724 {
3725     *is_bss = FALSE;
3726
3727     if (hdr->sh_type == SHT_PROGBITS
3728         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
3729         /* .text-style section */
3730         return SECTIONKIND_CODE_OR_RODATA;
3731     }
3732
3733     if (hdr->sh_type == SHT_PROGBITS
3734             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3735             /* .data-style section */
3736             return SECTIONKIND_RWDATA;
3737     }
3738
3739     if (hdr->sh_type == SHT_PROGBITS
3740         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
3741         /* .rodata-style section */
3742         return SECTIONKIND_CODE_OR_RODATA;
3743     }
3744
3745     if (hdr->sh_type == SHT_NOBITS
3746         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3747         /* .bss-style section */
3748         *is_bss = TRUE;
3749         return SECTIONKIND_RWDATA;
3750     }
3751
3752     return SECTIONKIND_OTHER;
3753 }
3754
3755
3756 static int
3757 ocGetNames_ELF ( ObjectCode* oc )
3758 {
3759    int i, j, k, nent;
3760    Elf_Sym* stab;
3761
3762    char*     ehdrC    = (char*)(oc->image);
3763    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
3764    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
3765    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3766
3767    ASSERT(symhash != NULL);
3768
3769    if (!strtab) {
3770       errorBelch("%s: no strtab", oc->fileName);
3771       return 0;
3772    }
3773
3774    k = 0;
3775    for (i = 0; i < ehdr->e_shnum; i++) {
3776       /* Figure out what kind of section it is.  Logic derived from
3777          Figure 1.14 ("Special Sections") of the ELF document
3778          ("Portable Formats Specification, Version 1.1"). */
3779       int         is_bss = FALSE;
3780       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
3781
3782       if (is_bss && shdr[i].sh_size > 0) {
3783          /* This is a non-empty .bss section.  Allocate zeroed space for
3784             it, and set its .sh_offset field such that
3785             ehdrC + .sh_offset == addr_of_zeroed_space.  */
3786          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
3787                                        "ocGetNames_ELF(BSS)");
3788          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
3789          /*
3790          debugBelch("BSS section at 0x%x, size %d\n",
3791                          zspace, shdr[i].sh_size);
3792          */
3793       }
3794
3795       /* fill in the section info */
3796       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
3797          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
3798          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
3799                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
3800       }
3801
3802       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3803
3804       /* copy stuff into this module's object symbol table */
3805       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3806       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3807
3808       oc->n_symbols = nent;
3809       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3810                                    "ocGetNames_ELF(oc->symbols)");
3811
3812       for (j = 0; j < nent; j++) {
3813
3814          char  isLocal = FALSE; /* avoids uninit-var warning */
3815          char* ad      = NULL;
3816          char* nm      = strtab + stab[j].st_name;
3817          int   secno   = stab[j].st_shndx;
3818
3819          /* Figure out if we want to add it; if so, set ad to its
3820             address.  Otherwise leave ad == NULL. */
3821
3822          if (secno == SHN_COMMON) {
3823             isLocal = FALSE;
3824             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
3825             /*
3826             debugBelch("COMMON symbol, size %d name %s\n",
3827                             stab[j].st_size, nm);
3828             */
3829             /* Pointless to do addProddableBlock() for this area,
3830                since the linker should never poke around in it. */
3831          }
3832          else
3833          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
3834                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
3835               )
3836               /* and not an undefined symbol */
3837               && stab[j].st_shndx != SHN_UNDEF
3838               /* and not in a "special section" */
3839               && stab[j].st_shndx < SHN_LORESERVE
3840               &&
3841               /* and it's a not a section or string table or anything silly */
3842               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
3843                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
3844                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
3845               )
3846             ) {
3847             /* Section 0 is the undefined section, hence > and not >=. */
3848             ASSERT(secno > 0 && secno < ehdr->e_shnum);
3849             /*
3850             if (shdr[secno].sh_type == SHT_NOBITS) {
3851                debugBelch("   BSS symbol, size %d off %d name %s\n",
3852                                stab[j].st_size, stab[j].st_value, nm);
3853             }
3854             */
3855             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
3856             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
3857                isLocal = TRUE;
3858             } else {
3859 #ifdef ELF_FUNCTION_DESC
3860                /* dlsym() and the initialisation table both give us function
3861                 * descriptors, so to be consistent we store function descriptors
3862                 * in the symbol table */
3863                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
3864                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
3865 #endif
3866                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s\n",
3867                                       ad, oc->fileName, nm ));
3868                isLocal = FALSE;
3869             }
3870          }
3871
3872          /* And the decision is ... */
3873
3874          if (ad != NULL) {
3875             ASSERT(nm != NULL);
3876             oc->symbols[j] = nm;
3877             /* Acquire! */
3878             if (isLocal) {
3879                /* Ignore entirely. */
3880             } else {
3881                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
3882             }
3883          } else {
3884             /* Skip. */
3885             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
3886                                    strtab + stab[j].st_name ));
3887             /*
3888             debugBelch(
3889                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
3890                     (int)ELF_ST_BIND(stab[j].st_info),
3891                     (int)ELF_ST_TYPE(stab[j].st_info),
3892                     (int)stab[j].st_shndx,
3893                     strtab + stab[j].st_name
3894                    );
3895             */
3896             oc->symbols[j] = NULL;
3897          }
3898
3899       }
3900    }
3901
3902    return 1;
3903 }
3904
3905 /* Do ELF relocations which lack an explicit addend.  All x86-linux
3906    relocations appear to be of this form. */
3907 static int
3908 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
3909                          Elf_Shdr* shdr, int shnum,
3910                          Elf_Sym*  stab, char* strtab )
3911 {
3912    int j;
3913    char *symbol;
3914    Elf_Word* targ;
3915    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
3916    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
3917    int target_shndx = shdr[shnum].sh_info;
3918    int symtab_shndx = shdr[shnum].sh_link;
3919
3920    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
3921    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
3922    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
3923                           target_shndx, symtab_shndx ));
3924
3925    /* Skip sections that we're not interested in. */
3926    {
3927        int is_bss;
3928        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
3929        if (kind == SECTIONKIND_OTHER) {
3930            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
3931            return 1;
3932        }
3933    }
3934
3935    for (j = 0; j < nent; j++) {
3936       Elf_Addr offset = rtab[j].r_offset;
3937       Elf_Addr info   = rtab[j].r_info;
3938
3939       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
3940       Elf_Word* pP = (Elf_Word*)P;
3941       Elf_Addr  A  = *pP;
3942       Elf_Addr  S;
3943       void*     S_tmp;
3944       Elf_Addr  value;
3945       StgStablePtr stablePtr;
3946       StgPtr stableVal;
3947
3948       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
3949                              j, (void*)offset, (void*)info ));
3950       if (!info) {
3951          IF_DEBUG(linker,debugBelch( " ZERO" ));
3952          S = 0;
3953       } else {
3954          Elf_Sym sym = stab[ELF_R_SYM(info)];
3955          /* First see if it is a local symbol. */
3956          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
3957             /* Yes, so we can get the address directly from the ELF symbol
3958                table. */
3959             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
3960             S = (Elf_Addr)
3961                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
3962                        + stab[ELF_R_SYM(info)].st_value);
3963
3964          } else {
3965             symbol = strtab + sym.st_name;
3966             stablePtr = (StgStablePtr)lookupHashTable(stablehash, (StgWord)symbol);
3967             if (NULL == stablePtr) {
3968               /* No, so look up the name in our global table. */
3969               S_tmp = lookupSymbol( symbol );
3970               S = (Elf_Addr)S_tmp;
3971             } else {
3972               stableVal = deRefStablePtr( stablePtr );
3973               S_tmp = stableVal;
3974               S = (Elf_Addr)S_tmp;
3975             }
3976          }
3977          if (!S) {
3978             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3979             return 0;
3980          }
3981          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
3982       }
3983
3984       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
3985                              (void*)P, (void*)S, (void*)A ));
3986       checkProddableBlock ( oc, pP );
3987
3988       value = S + A;
3989
3990       switch (ELF_R_TYPE(info)) {
3991 #        ifdef i386_HOST_ARCH
3992          case R_386_32:   *pP = value;     break;
3993          case R_386_PC32: *pP = value - P; break;
3994 #        endif
3995          default:
3996             errorBelch("%s: unhandled ELF relocation(Rel) type %lu\n",
3997                   oc->fileName, (lnat)ELF_R_TYPE(info));
3998             return 0;
3999       }
4000
4001    }
4002    return 1;
4003 }
4004
4005 /* Do ELF relocations for which explicit addends are supplied.
4006    sparc-solaris relocations appear to be of this form. */
4007 static int
4008 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
4009                           Elf_Shdr* shdr, int shnum,
4010                           Elf_Sym*  stab, char* strtab )
4011 {
4012    int j;
4013    char *symbol = NULL;
4014    Elf_Addr targ;
4015    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
4016    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
4017    int target_shndx = shdr[shnum].sh_info;
4018    int symtab_shndx = shdr[shnum].sh_link;
4019
4020    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
4021    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
4022    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
4023                           target_shndx, symtab_shndx ));
4024
4025    for (j = 0; j < nent; j++) {
4026 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
4027       /* This #ifdef only serves to avoid unused-var warnings. */
4028       Elf_Addr  offset = rtab[j].r_offset;
4029       Elf_Addr  P      = targ + offset;
4030 #endif
4031       Elf_Addr  info   = rtab[j].r_info;
4032       Elf_Addr  A      = rtab[j].r_addend;
4033       Elf_Addr  S;
4034       void*     S_tmp;
4035       Elf_Addr  value;
4036 #     if defined(sparc_HOST_ARCH)
4037       Elf_Word* pP = (Elf_Word*)P;
4038       Elf_Word  w1, w2;
4039 #     elif defined(powerpc_HOST_ARCH)
4040       Elf_Sword delta;
4041 #     endif
4042
4043       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
4044                              j, (void*)offset, (void*)info,
4045                                 (void*)A ));
4046       if (!info) {
4047          IF_DEBUG(linker,debugBelch( " ZERO" ));
4048          S = 0;
4049       } else {
4050          Elf_Sym sym = stab[ELF_R_SYM(info)];
4051          /* First see if it is a local symbol. */
4052          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
4053             /* Yes, so we can get the address directly from the ELF symbol
4054                table. */
4055             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
4056             S = (Elf_Addr)
4057                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
4058                        + stab[ELF_R_SYM(info)].st_value);
4059 #ifdef ELF_FUNCTION_DESC
4060             /* Make a function descriptor for this function */
4061             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
4062                S = allocateFunctionDesc(S + A);
4063                A = 0;
4064             }
4065 #endif
4066          } else {
4067             /* No, so look up the name in our global table. */
4068             symbol = strtab + sym.st_name;
4069             S_tmp = lookupSymbol( symbol );
4070             S = (Elf_Addr)S_tmp;
4071
4072 #ifdef ELF_FUNCTION_DESC
4073             /* If a function, already a function descriptor - we would
4074                have to copy it to add an offset. */
4075             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
4076                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
4077 #endif
4078          }
4079          if (!S) {
4080            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
4081            return 0;
4082          }
4083          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
4084       }
4085
4086       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
4087                                         (void*)P, (void*)S, (void*)A ));
4088       /* checkProddableBlock ( oc, (void*)P ); */
4089
4090       value = S + A;
4091
4092       switch (ELF_R_TYPE(info)) {
4093 #        if defined(sparc_HOST_ARCH)
4094          case R_SPARC_WDISP30:
4095             w1 = *pP & 0xC0000000;
4096             w2 = (Elf_Word)((value - P) >> 2);
4097             ASSERT((w2 & 0xC0000000) == 0);
4098             w1 |= w2;
4099             *pP = w1;
4100             break;
4101          case R_SPARC_HI22:
4102             w1 = *pP & 0xFFC00000;
4103             w2 = (Elf_Word)(value >> 10);
4104             ASSERT((w2 & 0xFFC00000) == 0);
4105             w1 |= w2;
4106             *pP = w1;
4107             break;
4108          case R_SPARC_LO10:
4109             w1 = *pP & ~0x3FF;
4110             w2 = (Elf_Word)(value & 0x3FF);
4111             ASSERT((w2 & ~0x3FF) == 0);
4112             w1 |= w2;
4113             *pP = w1;
4114             break;
4115
4116          /* According to the Sun documentation:
4117             R_SPARC_UA32
4118             This relocation type resembles R_SPARC_32, except it refers to an
4119             unaligned word. That is, the word to be relocated must be treated
4120             as four separate bytes with arbitrary alignment, not as a word
4121             aligned according to the architecture requirements.
4122          */
4123          case R_SPARC_UA32:
4124             w2  = (Elf_Word)value;
4125
4126             // SPARC doesn't do misaligned writes of 32 bit words,
4127             //       so we have to do this one byte-at-a-time.
4128             char *pPc   = (char*)pP;
4129             pPc[0]      = (char) ((Elf_Word)(w2 & 0xff000000) >> 24);
4130             pPc[1]      = (char) ((Elf_Word)(w2 & 0x00ff0000) >> 16);
4131             pPc[2]      = (char) ((Elf_Word)(w2 & 0x0000ff00) >> 8);
4132             pPc[3]      = (char) ((Elf_Word)(w2 & 0x000000ff));
4133             break;
4134
4135          case R_SPARC_32:
4136             w2 = (Elf_Word)value;
4137             *pP = w2;
4138             break;
4139 #        elif defined(powerpc_HOST_ARCH)
4140          case R_PPC_ADDR16_LO:
4141             *(Elf32_Half*) P = value;
4142             break;
4143
4144          case R_PPC_ADDR16_HI:
4145             *(Elf32_Half*) P = value >> 16;
4146             break;
4147
4148          case R_PPC_ADDR16_HA:
4149             *(Elf32_Half*) P = (value + 0x8000) >> 16;
4150             break;
4151
4152          case R_PPC_ADDR32:
4153             *(Elf32_Word *) P = value;
4154             break;
4155
4156          case R_PPC_REL32:
4157             *(Elf32_Word *) P = value - P;
4158             break;
4159
4160          case R_PPC_REL24:
4161             delta = value - P;
4162
4163             if( delta << 6 >> 6 != delta )
4164             {
4165                value = (Elf_Addr) (&makeSymbolExtra( oc, ELF_R_SYM(info), value )
4166                                         ->jumpIsland);
4167                delta = value - P;
4168
4169                if( value == 0 || delta << 6 >> 6 != delta )
4170                {
4171                   barf( "Unable to make SymbolExtra for #%d",
4172                         ELF_R_SYM(info) );
4173                   return 0;
4174                }
4175             }
4176
4177             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
4178                                           | (delta & 0x3fffffc);
4179             break;
4180 #        endif
4181
4182 #if x86_64_HOST_ARCH
4183       case R_X86_64_64:
4184           *(Elf64_Xword *)P = value;
4185           break;
4186
4187       case R_X86_64_PC32:
4188       {
4189 #if defined(ALWAYS_PIC)
4190           barf("R_X86_64_PC32 relocation, but ALWAYS_PIC.");
4191 #else
4192           StgInt64 off = value - P;
4193           if (off >= 0x7fffffffL || off < -0x80000000L) {
4194 #if X86_64_ELF_NONPIC_HACK
4195               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4196                                                 -> jumpIsland;
4197               off = pltAddress + A - P;
4198 #else
4199               barf("R_X86_64_PC32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4200                    symbol, off, oc->fileName );
4201 #endif
4202           }
4203           *(Elf64_Word *)P = (Elf64_Word)off;
4204 #endif
4205           break;
4206       }
4207
4208       case R_X86_64_PC64:
4209       {
4210           StgInt64 off = value - P;
4211           *(Elf64_Word *)P = (Elf64_Word)off;
4212           break;
4213       }
4214
4215       case R_X86_64_32:
4216 #if defined(ALWAYS_PIC)
4217           barf("R_X86_64_32 relocation, but ALWAYS_PIC.");
4218 #else
4219           if (value >= 0x7fffffffL) {
4220 #if X86_64_ELF_NONPIC_HACK
4221               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4222                                                 -> jumpIsland;
4223               value = pltAddress + A;
4224 #else
4225               barf("R_X86_64_32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4226                    symbol, value, oc->fileName );
4227 #endif
4228           }
4229           *(Elf64_Word *)P = (Elf64_Word)value;
4230 #endif
4231           break;
4232
4233       case R_X86_64_32S:
4234 #if defined(ALWAYS_PIC)
4235           barf("R_X86_64_32S relocation, but ALWAYS_PIC.");
4236 #else
4237           if ((StgInt64)value > 0x7fffffffL || (StgInt64)value < -0x80000000L) {
4238 #if X86_64_ELF_NONPIC_HACK
4239               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4240                                                 -> jumpIsland;
4241               value = pltAddress + A;
4242 #else
4243               barf("R_X86_64_32S relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4244                    symbol, value, oc->fileName );
4245 #endif
4246           }
4247           *(Elf64_Sword *)P = (Elf64_Sword)value;
4248 #endif
4249           break;
4250
4251       case R_X86_64_GOTPCREL:
4252       {
4253           StgInt64 gotAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)->addr;
4254           StgInt64 off = gotAddress + A - P;
4255           *(Elf64_Word *)P = (Elf64_Word)off;
4256           break;
4257       }
4258
4259       case R_X86_64_PLT32:
4260       {
4261 #if defined(ALWAYS_PIC)
4262           barf("R_X86_64_PLT32 relocation, but ALWAYS_PIC.");
4263 #else
4264           StgInt64 off = value - P;
4265           if (off >= 0x7fffffffL || off < -0x80000000L) {
4266               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4267                                                     -> jumpIsland;
4268               off = pltAddress + A - P;
4269           }
4270           *(Elf64_Word *)P = (Elf64_Word)off;
4271 #endif
4272           break;
4273       }
4274 #endif
4275
4276          default:
4277             errorBelch("%s: unhandled ELF relocation(RelA) type %lu\n",
4278                   oc->fileName, (lnat)ELF_R_TYPE(info));
4279             return 0;
4280       }
4281
4282    }
4283    return 1;
4284 }
4285
4286 static int
4287 ocResolve_ELF ( ObjectCode* oc )
4288 {
4289    char *strtab;
4290    int   shnum, ok;
4291    Elf_Sym*  stab  = NULL;
4292    char*     ehdrC = (char*)(oc->image);
4293    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
4294    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
4295
4296    /* first find "the" symbol table */
4297    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
4298
4299    /* also go find the string table */
4300    strtab = findElfSection ( ehdrC, SHT_STRTAB );
4301
4302    if (stab == NULL || strtab == NULL) {
4303       errorBelch("%s: can't find string or symbol table", oc->fileName);
4304       return 0;
4305    }
4306
4307    /* Process the relocation sections. */
4308    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
4309       if (shdr[shnum].sh_type == SHT_REL) {
4310          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
4311                                        shnum, stab, strtab );
4312          if (!ok) return ok;
4313       }
4314       else
4315       if (shdr[shnum].sh_type == SHT_RELA) {
4316          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
4317                                         shnum, stab, strtab );
4318          if (!ok) return ok;
4319       }
4320    }
4321
4322 #if defined(powerpc_HOST_ARCH)
4323    ocFlushInstructionCache( oc );
4324 #endif
4325
4326    return 1;
4327 }
4328
4329 /*
4330  * PowerPC & X86_64 ELF specifics
4331  */
4332
4333 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
4334
4335 static int ocAllocateSymbolExtras_ELF( ObjectCode *oc )
4336 {
4337   Elf_Ehdr *ehdr;
4338   Elf_Shdr* shdr;
4339   int i;
4340
4341   ehdr = (Elf_Ehdr *) oc->image;
4342   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
4343
4344   for( i = 0; i < ehdr->e_shnum; i++ )
4345     if( shdr[i].sh_type == SHT_SYMTAB )
4346       break;
4347
4348   if( i == ehdr->e_shnum )
4349   {
4350     errorBelch( "This ELF file contains no symtab" );
4351     return 0;
4352   }
4353
4354   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
4355   {
4356     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
4357       (int) shdr[i].sh_entsize, (int) sizeof( Elf_Sym ) );
4358
4359     return 0;
4360   }
4361
4362   return ocAllocateSymbolExtras( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
4363 }
4364
4365 #endif /* powerpc */
4366
4367 #endif /* ELF */
4368
4369 /* --------------------------------------------------------------------------
4370  * Mach-O specifics
4371  * ------------------------------------------------------------------------*/
4372
4373 #if defined(OBJFORMAT_MACHO)
4374
4375 /*
4376   Support for MachO linking on Darwin/MacOS X
4377   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
4378
4379   I hereby formally apologize for the hackish nature of this code.
4380   Things that need to be done:
4381   *) implement ocVerifyImage_MachO
4382   *) add still more sanity checks.
4383 */
4384
4385 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4386 #define mach_header mach_header_64
4387 #define segment_command segment_command_64
4388 #define section section_64
4389 #define nlist nlist_64
4390 #endif
4391
4392 #ifdef powerpc_HOST_ARCH
4393 static int ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4394 {
4395     struct mach_header *header = (struct mach_header *) oc->image;
4396     struct load_command *lc = (struct load_command *) (header + 1);
4397     unsigned i;
4398
4399     for( i = 0; i < header->ncmds; i++ )
4400     {
4401         if( lc->cmd == LC_SYMTAB )
4402         {
4403                 // Find out the first and last undefined external
4404                 // symbol, so we don't have to allocate too many
4405                 // jump islands.
4406             struct symtab_command *symLC = (struct symtab_command *) lc;
4407             unsigned min = symLC->nsyms, max = 0;
4408             struct nlist *nlist =
4409                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
4410                       : NULL;
4411             for(i=0;i<symLC->nsyms;i++)
4412             {
4413                 if(nlist[i].n_type & N_STAB)
4414                     ;
4415                 else if(nlist[i].n_type & N_EXT)
4416                 {
4417                     if((nlist[i].n_type & N_TYPE) == N_UNDF
4418                         && (nlist[i].n_value == 0))
4419                     {
4420                         if(i < min)
4421                             min = i;
4422                         if(i > max)
4423                             max = i;
4424                     }
4425                 }
4426             }
4427             if(max >= min)
4428                 return ocAllocateSymbolExtras(oc, max - min + 1, min);
4429
4430             break;
4431         }
4432
4433         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4434     }
4435     return ocAllocateSymbolExtras(oc,0,0);
4436 }
4437 #endif
4438 #ifdef x86_64_HOST_ARCH
4439 static int ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4440 {
4441     struct mach_header *header = (struct mach_header *) oc->image;
4442     struct load_command *lc = (struct load_command *) (header + 1);
4443     unsigned i;
4444
4445     for( i = 0; i < header->ncmds; i++ )
4446     {
4447         if( lc->cmd == LC_SYMTAB )
4448         {
4449                 // Just allocate one entry for every symbol
4450             struct symtab_command *symLC = (struct symtab_command *) lc;
4451
4452             return ocAllocateSymbolExtras(oc, symLC->nsyms, 0);
4453         }
4454
4455         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4456     }
4457     return ocAllocateSymbolExtras(oc,0,0);
4458 }
4459 #endif
4460
4461 static int ocVerifyImage_MachO(ObjectCode* oc)
4462 {
4463     char *image = (char*) oc->image;
4464     struct mach_header *header = (struct mach_header*) image;
4465
4466 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4467     if(header->magic != MH_MAGIC_64) {
4468         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4469                    oc->fileName, MH_MAGIC_64, header->magic);
4470         return 0;
4471     }
4472 #else
4473     if(header->magic != MH_MAGIC) {
4474         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4475                    oc->fileName, MH_MAGIC, header->magic);
4476         return 0;
4477     }
4478 #endif
4479     // FIXME: do some more verifying here
4480     return 1;
4481 }
4482
4483 static int resolveImports(
4484     ObjectCode* oc,
4485     char *image,
4486     struct symtab_command *symLC,
4487     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
4488     unsigned long *indirectSyms,
4489     struct nlist *nlist)
4490 {
4491     unsigned i;
4492     size_t itemSize = 4;
4493
4494     IF_DEBUG(linker, debugBelch("resolveImports: start\n"));
4495
4496 #if i386_HOST_ARCH
4497     int isJumpTable = 0;
4498     if(!strcmp(sect->sectname,"__jump_table"))
4499     {
4500         isJumpTable = 1;
4501         itemSize = 5;
4502         ASSERT(sect->reserved2 == itemSize);
4503     }
4504 #endif
4505
4506     for(i=0; i*itemSize < sect->size;i++)
4507     {
4508         // according to otool, reserved1 contains the first index into the indirect symbol table
4509         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
4510         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4511         void *addr = NULL;
4512
4513         IF_DEBUG(linker, debugBelch("resolveImports: resolving %s\n", nm));
4514         if ((symbol->n_type & N_TYPE) == N_UNDF
4515             && (symbol->n_type & N_EXT) && (symbol->n_value != 0)) {
4516             addr = (void*) (symbol->n_value);
4517             IF_DEBUG(linker, debugBelch("resolveImports: undefined external %s has value %p\n", nm, addr));
4518         } else {
4519             addr = lookupSymbol(nm);
4520             IF_DEBUG(linker, debugBelch("resolveImports: looking up %s, %p\n", nm, addr));
4521         }
4522         if (!addr)
4523         {
4524             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
4525             return 0;
4526         }
4527         ASSERT(addr);
4528
4529 #if i386_HOST_ARCH
4530         if(isJumpTable)
4531         {
4532             checkProddableBlock(oc,image + sect->offset + i*itemSize);
4533             *(image + sect->offset + i*itemSize) = 0xe9; // jmp
4534             *(unsigned*)(image + sect->offset + i*itemSize + 1)
4535                 = (char*)addr - (image + sect->offset + i*itemSize + 5);
4536         }
4537         else
4538 #endif
4539         {
4540             checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
4541             ((void**)(image + sect->offset))[i] = addr;
4542         }
4543     }
4544
4545     IF_DEBUG(linker, debugBelch("resolveImports: done\n"));
4546     return 1;
4547 }
4548
4549 static unsigned long relocateAddress(
4550     ObjectCode* oc,
4551     int nSections,
4552     struct section* sections,
4553     unsigned long address)
4554 {
4555     int i;
4556     IF_DEBUG(linker, debugBelch("relocateAddress: start\n"));
4557     for (i = 0; i < nSections; i++)
4558     {
4559             IF_DEBUG(linker, debugBelch("    relocating address in section %d\n", i));
4560         if (sections[i].addr <= address
4561             && address < sections[i].addr + sections[i].size)
4562         {
4563             return (unsigned long)oc->image
4564                     + sections[i].offset + address - sections[i].addr;
4565         }
4566     }
4567     barf("Invalid Mach-O file:"
4568          "Address out of bounds while relocating object file");
4569     return 0;
4570 }
4571
4572 static int relocateSection(
4573     ObjectCode* oc,
4574     char *image,
4575     struct symtab_command *symLC, struct nlist *nlist,
4576     int nSections, struct section* sections, struct section *sect)
4577 {
4578     struct relocation_info *relocs;
4579     int i, n;
4580
4581     IF_DEBUG(linker, debugBelch("relocateSection: start\n"));
4582
4583     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
4584         return 1;
4585     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
4586         return 1;
4587     else if(!strcmp(sect->sectname,"__la_sym_ptr2"))
4588         return 1;
4589     else if(!strcmp(sect->sectname,"__la_sym_ptr3"))
4590         return 1;
4591
4592     n = sect->nreloc;
4593     IF_DEBUG(linker, debugBelch("relocateSection: number of relocations: %d\n", n));
4594
4595     relocs = (struct relocation_info*) (image + sect->reloff);
4596
4597     for(i=0;i<n;i++)
4598     {
4599 #ifdef x86_64_HOST_ARCH
4600         struct relocation_info *reloc = &relocs[i];
4601
4602         char    *thingPtr = image + sect->offset + reloc->r_address;
4603         uint64_t thing;
4604         /* We shouldn't need to initialise this, but gcc on OS X 64 bit
4605            complains that it may be used uninitialized if we don't */
4606         uint64_t value = 0;
4607         uint64_t baseValue;
4608         int type = reloc->r_type;
4609
4610         checkProddableBlock(oc,thingPtr);
4611         switch(reloc->r_length)
4612         {
4613             case 0:
4614                 thing = *(uint8_t*)thingPtr;
4615                 baseValue = (uint64_t)thingPtr + 1;
4616                 break;
4617             case 1:
4618                 thing = *(uint16_t*)thingPtr;
4619                 baseValue = (uint64_t)thingPtr + 2;
4620                 break;
4621             case 2:
4622                 thing = *(uint32_t*)thingPtr;
4623                 baseValue = (uint64_t)thingPtr + 4;
4624                 break;
4625             case 3:
4626                 thing = *(uint64_t*)thingPtr;
4627                 baseValue = (uint64_t)thingPtr + 8;
4628                 break;
4629             default:
4630                 barf("Unknown size.");
4631         }
4632
4633         IF_DEBUG(linker,
4634                  debugBelch("relocateSection: length = %d, thing = %d, baseValue = %p\n",
4635                             reloc->r_length, thing, baseValue));
4636
4637         if (type == X86_64_RELOC_GOT
4638            || type == X86_64_RELOC_GOT_LOAD)
4639         {
4640             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4641             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4642
4643             IF_DEBUG(linker, debugBelch("relocateSection: making jump island for %s, extern = %d, X86_64_RELOC_GOT\n", nm, reloc->r_extern));
4644             ASSERT(reloc->r_extern);
4645             value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, (unsigned long)lookupSymbol(nm))->addr;
4646
4647             type = X86_64_RELOC_SIGNED;
4648         }
4649         else if(reloc->r_extern)
4650         {
4651             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4652             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4653
4654             IF_DEBUG(linker, debugBelch("relocateSection: looking up external symbol %s\n", nm));
4655             IF_DEBUG(linker, debugBelch("               : type  = %d\n", symbol->n_type));
4656             IF_DEBUG(linker, debugBelch("               : sect  = %d\n", symbol->n_sect));
4657             IF_DEBUG(linker, debugBelch("               : desc  = %d\n", symbol->n_desc));
4658             IF_DEBUG(linker, debugBelch("               : value = %d\n", symbol->n_value));
4659             if ((symbol->n_type & N_TYPE) == N_SECT) {
4660                 value = relocateAddress(oc, nSections, sections,
4661                                         symbol->n_value);
4662                 IF_DEBUG(linker, debugBelch("relocateSection, defined external symbol %s, relocated address %p\n", nm, value));
4663             }
4664             else {
4665                 value = (uint64_t) lookupSymbol(nm);
4666                 IF_DEBUG(linker, debugBelch("relocateSection: external symbol %s, address %p\n", nm, value));
4667             }
4668         }
4669         else
4670         {
4671             value = sections[reloc->r_symbolnum-1].offset
4672                   - sections[reloc->r_symbolnum-1].addr
4673                   + (uint64_t) image;
4674         }
4675
4676         IF_DEBUG(linker, debugBelch("relocateSection: value = %p\n", value));
4677
4678         if (type == X86_64_RELOC_BRANCH)
4679         {
4680             if((int32_t)(value - baseValue) != (int64_t)(value - baseValue))
4681             {
4682                 ASSERT(reloc->r_extern);
4683                 value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)
4684                                         -> jumpIsland;
4685             }
4686             ASSERT((int32_t)(value - baseValue) == (int64_t)(value - baseValue));
4687             type = X86_64_RELOC_SIGNED;
4688         }
4689
4690         switch(type)
4691         {
4692             case X86_64_RELOC_UNSIGNED:
4693                 ASSERT(!reloc->r_pcrel);
4694                 thing += value;
4695                 break;
4696             case X86_64_RELOC_SIGNED:
4697             case X86_64_RELOC_SIGNED_1:
4698             case X86_64_RELOC_SIGNED_2:
4699             case X86_64_RELOC_SIGNED_4:
4700                 ASSERT(reloc->r_pcrel);
4701                 thing += value - baseValue;
4702                 break;
4703             case X86_64_RELOC_SUBTRACTOR:
4704                 ASSERT(!reloc->r_pcrel);
4705                 thing -= value;
4706                 break;
4707             default:
4708                 barf("unkown relocation");
4709         }
4710
4711         switch(reloc->r_length)
4712         {
4713             case 0:
4714                 *(uint8_t*)thingPtr = thing;
4715                 break;
4716             case 1:
4717                 *(uint16_t*)thingPtr = thing;
4718                 break;
4719             case 2:
4720                 *(uint32_t*)thingPtr = thing;
4721                 break;
4722             case 3:
4723                 *(uint64_t*)thingPtr = thing;
4724                 break;
4725         }
4726 #else
4727         if(relocs[i].r_address & R_SCATTERED)
4728         {
4729             struct scattered_relocation_info *scat =
4730                 (struct scattered_relocation_info*) &relocs[i];
4731
4732             if(!scat->r_pcrel)
4733             {
4734                 if(scat->r_length == 2)
4735                 {
4736                     unsigned long word = 0;
4737                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
4738                     checkProddableBlock(oc,wordPtr);
4739
4740                     // Note on relocation types:
4741                     // i386 uses the GENERIC_RELOC_* types,
4742                     // while ppc uses special PPC_RELOC_* types.
4743                     // *_RELOC_VANILLA and *_RELOC_PAIR have the same value
4744                     // in both cases, all others are different.
4745                     // Therefore, we use GENERIC_RELOC_VANILLA
4746                     // and GENERIC_RELOC_PAIR instead of the PPC variants,
4747                     // and use #ifdefs for the other types.
4748
4749                     // Step 1: Figure out what the relocated value should be
4750                     if(scat->r_type == GENERIC_RELOC_VANILLA)
4751                     {
4752                         word = *wordPtr + (unsigned long) relocateAddress(
4753                                                                 oc,
4754                                                                 nSections,
4755                                                                 sections,
4756                                                                 scat->r_value)
4757                                         - scat->r_value;
4758                     }
4759 #ifdef powerpc_HOST_ARCH
4760                     else if(scat->r_type == PPC_RELOC_SECTDIFF
4761                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
4762                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
4763                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF
4764                         || scat->r_type == PPC_RELOC_LOCAL_SECTDIFF)
4765 #else
4766                     else if(scat->r_type == GENERIC_RELOC_SECTDIFF
4767                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
4768 #endif
4769                     {
4770                         struct scattered_relocation_info *pair =
4771                                 (struct scattered_relocation_info*) &relocs[i+1];
4772
4773                         if(!pair->r_scattered || pair->r_type != GENERIC_RELOC_PAIR)
4774                             barf("Invalid Mach-O file: "
4775                                  "RELOC_*_SECTDIFF not followed by RELOC_PAIR");
4776
4777                         word = (unsigned long)
4778                                (relocateAddress(oc, nSections, sections, scat->r_value)
4779                               - relocateAddress(oc, nSections, sections, pair->r_value));
4780                         i++;
4781                     }
4782 #ifdef powerpc_HOST_ARCH
4783                     else if(scat->r_type == PPC_RELOC_HI16
4784                          || scat->r_type == PPC_RELOC_LO16
4785                          || scat->r_type == PPC_RELOC_HA16
4786                          || scat->r_type == PPC_RELOC_LO14)
4787                     {   // these are generated by label+offset things
4788                         struct relocation_info *pair = &relocs[i+1];
4789                         if((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR)
4790                             barf("Invalid Mach-O file: "
4791                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
4792
4793                         if(scat->r_type == PPC_RELOC_LO16)
4794                         {
4795                             word = ((unsigned short*) wordPtr)[1];
4796                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4797                         }
4798                         else if(scat->r_type == PPC_RELOC_LO14)
4799                         {
4800                             barf("Unsupported Relocation: PPC_RELOC_LO14");
4801                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
4802                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4803                         }
4804                         else if(scat->r_type == PPC_RELOC_HI16)
4805                         {
4806                             word = ((unsigned short*) wordPtr)[1] << 16;
4807                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4808                         }
4809                         else if(scat->r_type == PPC_RELOC_HA16)
4810                         {
4811                             word = ((unsigned short*) wordPtr)[1] << 16;
4812                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
4813                         }
4814
4815
4816                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
4817                                                 - scat->r_value;
4818
4819                         i++;
4820                     }
4821  #endif
4822                     else
4823                     {
4824                         barf ("Don't know how to handle this Mach-O "
4825                               "scattered relocation entry: "
4826                               "object file %s; entry type %ld; "
4827                               "address %#lx\n",
4828                               OC_INFORMATIVE_FILENAME(oc),
4829                               scat->r_type,
4830                               scat->r_address);
4831                         return 0;
4832                      }
4833
4834 #ifdef powerpc_HOST_ARCH
4835                     if(scat->r_type == GENERIC_RELOC_VANILLA
4836                         || scat->r_type == PPC_RELOC_SECTDIFF)
4837 #else
4838                     if(scat->r_type == GENERIC_RELOC_VANILLA
4839                         || scat->r_type == GENERIC_RELOC_SECTDIFF
4840                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
4841 #endif
4842                     {
4843                         *wordPtr = word;
4844                     }
4845 #ifdef powerpc_HOST_ARCH
4846                     else if(scat->r_type == PPC_RELOC_LO16_SECTDIFF || scat->r_type == PPC_RELOC_LO16)
4847                     {
4848                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
4849                     }
4850                     else if(scat->r_type == PPC_RELOC_HI16_SECTDIFF || scat->r_type == PPC_RELOC_HI16)
4851                     {
4852                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
4853                     }
4854                     else if(scat->r_type == PPC_RELOC_HA16_SECTDIFF || scat->r_type == PPC_RELOC_HA16)
4855                     {
4856                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
4857                             + ((word & (1<<15)) ? 1 : 0);
4858                     }
4859 #endif
4860                 }
4861                 else
4862                 {
4863                     barf("Can't handle Mach-O scattered relocation entry "
4864                          "with this r_length tag: "
4865                          "object file %s; entry type %ld; "
4866                          "r_length tag %ld; address %#lx\n",
4867                          OC_INFORMATIVE_FILENAME(oc),
4868                          scat->r_type,
4869                          scat->r_length,
4870                          scat->r_address);
4871                     return 0;
4872                 }
4873             }
4874             else /* scat->r_pcrel */
4875             {
4876                 barf("Don't know how to handle *PC-relative* Mach-O "
4877                      "scattered relocation entry: "
4878                      "object file %s; entry type %ld; address %#lx\n",
4879                      OC_INFORMATIVE_FILENAME(oc),
4880                      scat->r_type,
4881                      scat->r_address);
4882                return 0;
4883             }
4884
4885         }
4886         else /* !(relocs[i].r_address & R_SCATTERED) */
4887         {
4888             struct relocation_info *reloc = &relocs[i];
4889             if(reloc->r_pcrel && !reloc->r_extern)
4890                 continue;
4891
4892             if(reloc->r_length == 2)
4893             {
4894                 unsigned long word = 0;
4895 #ifdef powerpc_HOST_ARCH
4896                 unsigned long jumpIsland = 0;
4897                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
4898                                                       // to avoid warning and to catch
4899                                                       // bugs.
4900 #endif
4901
4902                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
4903                 checkProddableBlock(oc,wordPtr);
4904
4905                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
4906                 {
4907                     word = *wordPtr;
4908                 }
4909 #ifdef powerpc_HOST_ARCH
4910                 else if(reloc->r_type == PPC_RELOC_LO16)
4911                 {
4912                     word = ((unsigned short*) wordPtr)[1];
4913                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4914                 }
4915                 else if(reloc->r_type == PPC_RELOC_HI16)
4916                 {
4917                     word = ((unsigned short*) wordPtr)[1] << 16;
4918                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4919                 }
4920                 else if(reloc->r_type == PPC_RELOC_HA16)
4921                 {
4922                     word = ((unsigned short*) wordPtr)[1] << 16;
4923                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
4924                 }
4925                 else if(reloc->r_type == PPC_RELOC_BR24)
4926                 {
4927                     word = *wordPtr;
4928                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
4929                 }
4930 #endif
4931                 else
4932                 {
4933                     barf("Can't handle this Mach-O relocation entry "
4934                          "(not scattered): "
4935                          "object file %s; entry type %ld; address %#lx\n",
4936                          OC_INFORMATIVE_FILENAME(oc),
4937                          reloc->r_type,
4938                          reloc->r_address);
4939                     return 0;
4940                 }
4941
4942                 if(!reloc->r_extern)
4943                 {
4944                     long delta =
4945                         sections[reloc->r_symbolnum-1].offset
4946                         - sections[reloc->r_symbolnum-1].addr
4947                         + ((long) image);
4948
4949                     word += delta;
4950                 }
4951                 else
4952                 {
4953                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
4954                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4955                     void *symbolAddress = lookupSymbol(nm);
4956                     if(!symbolAddress)
4957                     {
4958                         errorBelch("\nunknown symbol `%s'", nm);
4959                         return 0;
4960                     }
4961
4962                     if(reloc->r_pcrel)
4963                     {
4964 #ifdef powerpc_HOST_ARCH
4965                             // In the .o file, this should be a relative jump to NULL
4966                             // and we'll change it to a relative jump to the symbol
4967                         ASSERT(word + reloc->r_address == 0);
4968                         jumpIsland = (unsigned long)
4969                                         &makeSymbolExtra(oc,
4970                                                          reloc->r_symbolnum,
4971                                                          (unsigned long) symbolAddress)
4972                                          -> jumpIsland;
4973                         if(jumpIsland != 0)
4974                         {
4975                             offsetToJumpIsland = word + jumpIsland
4976                                 - (((long)image) + sect->offset - sect->addr);
4977                         }
4978 #endif
4979                         word += (unsigned long) symbolAddress
4980                                 - (((long)image) + sect->offset - sect->addr);
4981                     }
4982                     else
4983                     {
4984                         word += (unsigned long) symbolAddress;
4985                     }
4986                 }
4987
4988                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
4989                 {
4990                     *wordPtr = word;
4991                     continue;
4992                 }
4993 #ifdef powerpc_HOST_ARCH
4994                 else if(reloc->r_type == PPC_RELOC_LO16)
4995                 {
4996                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
4997                     i++; continue;
4998                 }
4999                 else if(reloc->r_type == PPC_RELOC_HI16)
5000                 {
5001                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
5002                     i++; continue;
5003                 }
5004                 else if(reloc->r_type == PPC_RELOC_HA16)
5005                 {
5006                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
5007                         + ((word & (1<<15)) ? 1 : 0);
5008                     i++; continue;
5009                 }
5010                 else if(reloc->r_type == PPC_RELOC_BR24)
5011                 {
5012                     if((word & 0x03) != 0)
5013                         barf("%s: unconditional relative branch with a displacement "
5014                              "which isn't a multiple of 4 bytes: %#lx",
5015                              OC_INFORMATIVE_FILENAME(oc),
5016                              word);
5017
5018                     if((word & 0xFE000000) != 0xFE000000 &&
5019                        (word & 0xFE000000) != 0x00000000)
5020                     {
5021                         // The branch offset is too large.
5022                         // Therefore, we try to use a jump island.
5023                         if(jumpIsland == 0)
5024                         {
5025                             barf("%s: unconditional relative branch out of range: "
5026                                  "no jump island available: %#lx",
5027                                  OC_INFORMATIVE_FILENAME(oc),
5028                                  word);
5029                         }
5030
5031                         word = offsetToJumpIsland;
5032                         if((word & 0xFE000000) != 0xFE000000 &&
5033                            (word & 0xFE000000) != 0x00000000)
5034                             barf("%s: unconditional relative branch out of range: "
5035                                  "jump island out of range: %#lx",
5036                                  OC_INFORMATIVE_FILENAME(oc),
5037                                  word);
5038                     }
5039                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
5040                     continue;
5041                 }
5042 #endif
5043             }
5044             else
5045             {
5046                  barf("Can't handle Mach-O relocation entry (not scattered) "
5047                       "with this r_length tag: "
5048                       "object file %s; entry type %ld; "
5049                       "r_length tag %ld; address %#lx\n",
5050                       OC_INFORMATIVE_FILENAME(oc),
5051                       reloc->r_type,
5052                       reloc->r_length,
5053                       reloc->r_address);
5054                  return 0;
5055             }
5056         }
5057 #endif
5058     }
5059     IF_DEBUG(linker, debugBelch("relocateSection: done\n"));
5060     return 1;
5061 }
5062
5063 static int ocGetNames_MachO(ObjectCode* oc)
5064 {
5065     char *image = (char*) oc->image;
5066     struct mach_header *header = (struct mach_header*) image;
5067     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
5068     unsigned i,curSymbol = 0;
5069     struct segment_command *segLC = NULL;
5070     struct section *sections;
5071     struct symtab_command *symLC = NULL;
5072     struct nlist *nlist;
5073     unsigned long commonSize = 0;
5074     char    *commonStorage = NULL;
5075     unsigned long commonCounter;
5076
5077     IF_DEBUG(linker,debugBelch("ocGetNames_MachO: start\n"));
5078
5079     for(i=0;i<header->ncmds;i++)
5080     {
5081         if(lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64)
5082             segLC = (struct segment_command*) lc;
5083         else if(lc->cmd == LC_SYMTAB)
5084             symLC = (struct symtab_command*) lc;
5085         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
5086     }
5087
5088     sections = (struct section*) (segLC+1);
5089     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
5090                   : NULL;
5091
5092     if(!segLC)
5093         barf("ocGetNames_MachO: no segment load command");
5094
5095     for(i=0;i<segLC->nsects;i++)
5096     {
5097         IF_DEBUG(linker, debugBelch("ocGetNames_MachO: segment %d\n", i));
5098         if (sections[i].size == 0)
5099             continue;
5100
5101         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
5102         {
5103             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
5104                                       "ocGetNames_MachO(common symbols)");
5105             sections[i].offset = zeroFillArea - image;
5106         }
5107
5108         if(!strcmp(sections[i].sectname,"__text"))
5109             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
5110                 (void*) (image + sections[i].offset),
5111                 (void*) (image + sections[i].offset + sections[i].size));
5112         else if(!strcmp(sections[i].sectname,"__const"))
5113             addSection(oc, SECTIONKIND_RWDATA,
5114                 (void*) (image + sections[i].offset),
5115                 (void*) (image + sections[i].offset + sections[i].size));
5116         else if(!strcmp(sections[i].sectname,"__data"))
5117             addSection(oc, SECTIONKIND_RWDATA,
5118                 (void*) (image + sections[i].offset),
5119                 (void*) (image + sections[i].offset + sections[i].size));
5120         else if(!strcmp(sections[i].sectname,"__bss")
5121                 || !strcmp(sections[i].sectname,"__common"))
5122             addSection(oc, SECTIONKIND_RWDATA,
5123                 (void*) (image + sections[i].offset),
5124                 (void*) (image + sections[i].offset + sections[i].size));
5125
5126         addProddableBlock(oc, (void*) (image + sections[i].offset),
5127                                         sections[i].size);
5128     }
5129
5130         // count external symbols defined here
5131     oc->n_symbols = 0;
5132     if(symLC)
5133     {
5134         for(i=0;i<symLC->nsyms;i++)
5135         {
5136             if(nlist[i].n_type & N_STAB)
5137                 ;
5138             else if(nlist[i].n_type & N_EXT)
5139             {
5140                 if((nlist[i].n_type & N_TYPE) == N_UNDF
5141                     && (nlist[i].n_value != 0))
5142                 {
5143                     commonSize += nlist[i].n_value;
5144                     oc->n_symbols++;
5145                 }
5146                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
5147                     oc->n_symbols++;
5148             }
5149         }
5150     }
5151     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: %d external symbols\n", oc->n_symbols));
5152     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
5153                                    "ocGetNames_MachO(oc->symbols)");
5154
5155     if(symLC)
5156     {
5157         for(i=0;i<symLC->nsyms;i++)
5158         {
5159             if(nlist[i].n_type & N_STAB)
5160                 ;
5161             else if((nlist[i].n_type & N_TYPE) == N_SECT)
5162             {
5163                 if(nlist[i].n_type & N_EXT)
5164                 {
5165                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
5166                     if ((nlist[i].n_desc & N_WEAK_DEF) && lookupSymbol(nm)) {
5167                         // weak definition, and we already have a definition
5168                         IF_DEBUG(linker, debugBelch("    weak: %s\n", nm));
5169                     }
5170                     else
5171                     {
5172                             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: inserting %s\n", nm));
5173                             ghciInsertStrHashTable(oc->fileName, symhash, nm,
5174                                                     image
5175                                                     + sections[nlist[i].n_sect-1].offset
5176                                                     - sections[nlist[i].n_sect-1].addr
5177                                                     + nlist[i].n_value);
5178                             oc->symbols[curSymbol++] = nm;
5179                     }
5180                 }
5181             }
5182         }
5183     }
5184
5185     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
5186     commonCounter = (unsigned long)commonStorage;
5187     if(symLC)
5188     {
5189         for(i=0;i<symLC->nsyms;i++)
5190         {
5191             if((nlist[i].n_type & N_TYPE) == N_UNDF
5192                     && (nlist[i].n_type & N_EXT) && (nlist[i].n_value != 0))
5193             {
5194                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
5195                 unsigned long sz = nlist[i].n_value;
5196
5197                 nlist[i].n_value = commonCounter;
5198
5199                 IF_DEBUG(linker, debugBelch("ocGetNames_MachO: inserting common symbol: %s\n", nm));
5200                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
5201                                        (void*)commonCounter);
5202                 oc->symbols[curSymbol++] = nm;
5203
5204                 commonCounter += sz;
5205             }
5206         }
5207     }
5208     return 1;
5209 }
5210
5211 static int ocResolve_MachO(ObjectCode* oc)
5212 {
5213     char *image = (char*) oc->image;
5214     struct mach_header *header = (struct mach_header*) image;
5215     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
5216     unsigned i;
5217     struct segment_command *segLC = NULL;
5218     struct section *sections;
5219     struct symtab_command *symLC = NULL;
5220     struct dysymtab_command *dsymLC = NULL;
5221     struct nlist *nlist;
5222
5223     IF_DEBUG(linker, debugBelch("ocResolve_MachO: start\n"));
5224     for (i = 0; i < header->ncmds; i++)
5225     {
5226         if(lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64)
5227             segLC = (struct segment_command*) lc;
5228         else if(lc->cmd == LC_SYMTAB)
5229             symLC = (struct symtab_command*) lc;
5230         else if(lc->cmd == LC_DYSYMTAB)
5231             dsymLC = (struct dysymtab_command*) lc;
5232         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
5233     }
5234
5235     sections = (struct section*) (segLC+1);
5236     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
5237                   : NULL;
5238
5239     if(dsymLC)
5240     {
5241         unsigned long *indirectSyms
5242             = (unsigned long*) (image + dsymLC->indirectsymoff);
5243
5244         IF_DEBUG(linker, debugBelch("ocResolve_MachO: resolving dsymLC\n"));
5245         for (i = 0; i < segLC->nsects; i++)
5246         {
5247             if(    !strcmp(sections[i].sectname,"__la_symbol_ptr")
5248                 || !strcmp(sections[i].sectname,"__la_sym_ptr2")
5249                 || !strcmp(sections[i].sectname,"__la_sym_ptr3"))
5250             {
5251                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5252                     return 0;
5253             }
5254             else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr")
5255                 ||  !strcmp(sections[i].sectname,"__pointers"))
5256             {
5257                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5258                     return 0;
5259             }
5260             else if(!strcmp(sections[i].sectname,"__jump_table"))
5261             {
5262                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5263                     return 0;
5264             }
5265             else
5266             {
5267                 IF_DEBUG(linker, debugBelch("ocResolve_MachO: unknown section\n"));
5268             }
5269         }
5270     }
5271
5272     for(i=0;i<segLC->nsects;i++)
5273     {
5274             IF_DEBUG(linker, debugBelch("ocResolve_MachO: relocating section %d\n", i));
5275
5276         if (!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
5277             return 0;
5278     }
5279
5280 #if defined (powerpc_HOST_ARCH)
5281     ocFlushInstructionCache( oc );
5282 #endif
5283
5284     return 1;
5285 }
5286
5287 #ifdef powerpc_HOST_ARCH
5288 /*
5289  * The Mach-O object format uses leading underscores. But not everywhere.
5290  * There is a small number of runtime support functions defined in
5291  * libcc_dynamic.a whose name does not have a leading underscore.
5292  * As a consequence, we can't get their address from C code.
5293  * We have to use inline assembler just to take the address of a function.
5294  * Yuck.
5295  */
5296
5297 extern void* symbolsWithoutUnderscore[];
5298
5299 static void machoInitSymbolsWithoutUnderscore()
5300 {
5301     void **p = symbolsWithoutUnderscore;
5302     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
5303
5304 #undef SymI_NeedsProto
5305 #define SymI_NeedsProto(x)  \
5306     __asm__ volatile(".long " # x);
5307
5308     RTS_MACHO_NOUNDERLINE_SYMBOLS
5309
5310     __asm__ volatile(".text");
5311
5312 #undef SymI_NeedsProto
5313 #define SymI_NeedsProto(x)  \
5314     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
5315
5316     RTS_MACHO_NOUNDERLINE_SYMBOLS
5317
5318 #undef SymI_NeedsProto
5319 }
5320 #endif
5321
5322 #ifndef USE_MMAP
5323 /*
5324  * Figure out by how much to shift the entire Mach-O file in memory
5325  * when loading so that its single segment ends up 16-byte-aligned
5326  */
5327 static int machoGetMisalignment( FILE * f )
5328 {
5329     struct mach_header header;
5330     int misalignment;
5331
5332     {
5333         int n = fread(&header, sizeof(header), 1, f);
5334         if (n != 1) {
5335             barf("machoGetMisalignment: can't read the Mach-O header");
5336         }
5337     }
5338     fseek(f, -sizeof(header), SEEK_CUR);
5339
5340 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
5341     if(header.magic != MH_MAGIC_64) {
5342         barf("Bad magic. Expected: %08x, got: %08x.",
5343              MH_MAGIC_64, header.magic);
5344     }
5345 #else
5346     if(header.magic != MH_MAGIC) {
5347         barf("Bad magic. Expected: %08x, got: %08x.",
5348              MH_MAGIC, header.magic);
5349     }
5350 #endif
5351
5352     misalignment = (header.sizeofcmds + sizeof(header))
5353                     & 0xF;
5354
5355     return misalignment ? (16 - misalignment) : 0;
5356 }
5357 #endif
5358
5359 #endif