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 #if defined(__linux__)  || defined(__GLIBC__)
  17 #define _GNU_SOURCE 1
  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 <inttypes.h>
  44 #include <stdlib.h>
  45 #include <string.h>
  46 #include <stdio.h>
  47 #include <assert.h>
  48
  49 #ifdef HAVE_SYS_STAT_H
  50 #include <sys/stat.h>
  51 #endif
  52
  53 #if defined(HAVE_DLFCN_H)
  54 #include <dlfcn.h>
  55 #endif
  56
  57 #if defined(cygwin32_HOST_OS)
  58 #ifdef HAVE_DIRENT_H
  59 #include <dirent.h>
  60 #endif
  61
  62 #ifdef HAVE_SYS_TIME_H
  63 #include <sys/time.h>
  64 #endif
  65 #include <regex.h>
  66 #include <sys/fcntl.h>
  67 #include <sys/termios.h>
  68 #include <sys/utime.h>
  69 #include <sys/utsname.h>
  70 #include <sys/wait.h>
  71 #endif
  72
  73 #if !defined(powerpc_HOST_ARCH) && \
  74     (   defined(linux_HOST_OS    ) || defined(freebsd_HOST_OS) || \
  75         defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS ) || \
  76         defined(openbsd_HOST_OS  ) || defined(darwin_HOST_OS ) || \
  77         defined(kfreebsdgnu_HOST_OS) )
  78 /* Don't use mmap on powerpc_HOST_ARCH as mmap doesn't support
  79  * reallocating but we need to allocate jump islands just after each
  80  * object images. Otherwise relative branches to jump islands can fail
  81  * due to 24-bits displacement overflow.
  82  */
  83 #define USE_MMAP
  84 #include <fcntl.h>
  85 #include <sys/mman.h>
  86
  87 #ifdef HAVE_UNISTD_H
  88 #include <unistd.h>
  89 #endif
  90
  91 #endif
  92
  93 #if defined(linux_HOST_OS) || defined(solaris2_HOST_OS) || defined(freebsd_HOST_OS) || defined(kfreebsdgnu_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  94 #  define OBJFORMAT_ELF
  95 #  include <regex.h>    // regex is already used by dlopen() so this is OK
  96                         // to use here without requiring an additional lib
  97 #elif defined(cygwin32_HOST_OS) || defined (mingw32_HOST_OS)
  98 #  define OBJFORMAT_PEi386
  99 #  include <windows.h>
 100 #  include <math.h>
 101 #elif defined(darwin_HOST_OS)
 102 #  define OBJFORMAT_MACHO
 103 #  include <regex.h>
 104 #  include <mach/machine.h>
 105 #  include <mach-o/fat.h>
 106 #  include <mach-o/loader.h>
 107 #  include <mach-o/nlist.h>
 108 #  include <mach-o/reloc.h>
 109 #if !defined(HAVE_DLFCN_H)
 110 #  include <mach-o/dyld.h>
 111 #endif
 112 #if defined(powerpc_HOST_ARCH)
 113 #  include <mach-o/ppc/reloc.h>
 114 #endif
 115 #if defined(x86_64_HOST_ARCH)
 116 #  include <mach-o/x86_64/reloc.h>
 117 #endif
 118 #endif
 119
 120 #if defined(x86_64_HOST_ARCH) && defined(darwin_HOST_OS)
 121 #define ALWAYS_PIC
 122 #endif
 123
 124 /* Hash table mapping symbol names to Symbol */
 125 static /*Str*/HashTable *symhash;
 126
 127 /* Hash table mapping symbol names to StgStablePtr */
 128 static /*Str*/HashTable *stablehash;
 129
 130 /* List of currently loaded objects */
 131 ObjectCode *objects = NULL;     /* initially empty */
 132
 133 static HsInt loadOc( ObjectCode* oc );
 134 static ObjectCode* mkOc( char *path, char *image, int imageSize,
 135                          char *archiveMemberName
 136 #ifndef USE_MMAP
 137 #ifdef darwin_HOST_OS
 138                        , int misalignment
 139 #endif
 140 #endif
 141                        );
 142
 143 #if defined(OBJFORMAT_ELF)
 144 static int ocVerifyImage_ELF    ( ObjectCode* oc );
 145 static int ocGetNames_ELF       ( ObjectCode* oc );
 146 static int ocResolve_ELF        ( ObjectCode* oc );
 147 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 148 static int ocAllocateSymbolExtras_ELF ( ObjectCode* oc );
 149 #endif
 150 #elif defined(OBJFORMAT_PEi386)
 151 static int ocVerifyImage_PEi386 ( ObjectCode* oc );
 152 static int ocGetNames_PEi386    ( ObjectCode* oc );
 153 static int ocResolve_PEi386     ( ObjectCode* oc );
 154 static void *lookupSymbolInDLLs ( unsigned char *lbl );
 155 static void zapTrailingAtSign   ( unsigned char *sym );
 156 #elif defined(OBJFORMAT_MACHO)
 157 static int ocVerifyImage_MachO    ( ObjectCode* oc );
 158 static int ocGetNames_MachO       ( ObjectCode* oc );
 159 static int ocResolve_MachO        ( ObjectCode* oc );
 160
 161 #ifndef USE_MMAP
 162 static int machoGetMisalignment( FILE * );
 163 #endif
 164 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 165 static int ocAllocateSymbolExtras_MachO ( ObjectCode* oc );
 166 #endif
 167 #ifdef powerpc_HOST_ARCH
 168 static void machoInitSymbolsWithoutUnderscore( void );
 169 #endif
 170 #endif
 171
 172 /* on x86_64 we have a problem with relocating symbol references in
 173  * code that was compiled without -fPIC.  By default, the small memory
 174  * model is used, which assumes that symbol references can fit in a
 175  * 32-bit slot.  The system dynamic linker makes this work for
 176  * references to shared libraries by either (a) allocating a jump
 177  * table slot for code references, or (b) moving the symbol at load
 178  * time (and copying its contents, if necessary) for data references.
 179  *
 180  * We unfortunately can't tell whether symbol references are to code
 181  * or data.  So for now we assume they are code (the vast majority
 182  * are), and allocate jump-table slots.  Unfortunately this will
 183  * SILENTLY generate crashing code for data references.  This hack is
 184  * enabled by X86_64_ELF_NONPIC_HACK.
 185  *
 186  * One workaround is to use shared Haskell libraries.  This is
 187  * coming.  Another workaround is to keep the static libraries but
 188  * compile them with -fPIC, because that will generate PIC references
 189  * to data which can be relocated.  The PIC code is still too green to
 190  * do this systematically, though.
 191  *
 192  * See bug #781
 193  * See thread http://www.haskell.org/pipermail/cvs-ghc/2007-September/038458.html
 194  *
 195  * Naming Scheme for Symbol Macros
 196  *
 197  * SymI_*: symbol is internal to the RTS. It resides in an object
 198  *         file/library that is statically.
 199  * SymE_*: symbol is external to the RTS library. It might be linked
 200  *         dynamically.
 201  *
 202  * Sym*_HasProto  : the symbol prototype is imported in an include file
 203  *                  or defined explicitly
 204  * Sym*_NeedsProto: the symbol is undefined and we add a dummy
 205  *                  default proto extern void sym(void);
 206  */
 207 #define X86_64_ELF_NONPIC_HACK 1
 208
 209 /* Link objects into the lower 2Gb on x86_64.  GHC assumes the
 210  * small memory model on this architecture (see gcc docs,
 211  * -mcmodel=small).
 212  *
 213  * MAP_32BIT not available on OpenBSD/amd64
 214  */
 215 #if defined(x86_64_HOST_ARCH) && defined(MAP_32BIT)
 216 #define TRY_MAP_32BIT MAP_32BIT
 217 #else
 218 #define TRY_MAP_32BIT 0
 219 #endif
 220
 221 /*
 222  * Due to the small memory model (see above), on x86_64 we have to map
 223  * all our non-PIC object files into the low 2Gb of the address space
 224  * (why 2Gb and not 4Gb?  Because all addresses must be reachable
 225  * using a 32-bit signed PC-relative offset). On Linux we can do this
 226  * using the MAP_32BIT flag to mmap(), however on other OSs
 227  * (e.g. *BSD, see #2063, and also on Linux inside Xen, see #2512), we
 228  * can't do this.  So on these systems, we have to pick a base address
 229  * in the low 2Gb of the address space and try to allocate memory from
 230  * there.
 231  *
 232  * We pick a default address based on the OS, but also make this
 233  * configurable via an RTS flag (+RTS -xm)
 234  */
 235 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
 236
 237 #if defined(MAP_32BIT)
 238 // Try to use MAP_32BIT
 239 #define MMAP_32BIT_BASE_DEFAULT 0
 240 #else
 241 // A guess: 1Gb.
 242 #define MMAP_32BIT_BASE_DEFAULT 0x40000000
 243 #endif
 244
 245 static void *mmap_32bit_base = (void *)MMAP_32BIT_BASE_DEFAULT;
 246 #endif
 247
 248 /* MAP_ANONYMOUS is MAP_ANON on some systems, e.g. OpenBSD */
 249 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
 250 #define MAP_ANONYMOUS MAP_ANON
 251 #endif
 252
 253 /* -----------------------------------------------------------------------------
 254  * Built-in symbols from the RTS
 255  */
 256
 257 typedef struct _RtsSymbolVal {
 258     char   *lbl;
 259     void   *addr;
 260 } RtsSymbolVal;
 261
 262 #define Maybe_Stable_Names      SymI_HasProto(stg_mkWeakzh)                     \
 263                                 SymI_HasProto(stg_mkWeakForeignEnvzh)           \
 264                                 SymI_HasProto(stg_makeStableNamezh)             \
 265                                 SymI_HasProto(stg_finalizzeWeakzh)
 266
 267 #if !defined (mingw32_HOST_OS)
 268 #define RTS_POSIX_ONLY_SYMBOLS                  \
 269       SymI_HasProto(__hscore_get_saved_termios) \
 270       SymI_HasProto(__hscore_set_saved_termios) \
 271       SymI_HasProto(shutdownHaskellAndSignal)   \
 272       SymI_HasProto(lockFile)                   \
 273       SymI_HasProto(unlockFile)                 \
 274       SymI_HasProto(signal_handlers)            \
 275       SymI_HasProto(stg_sig_install)            \
 276       SymI_HasProto(rtsTimerSignal)             \
 277       SymI_HasProto(atexit)                     \
 278       SymI_NeedsProto(nocldstop)
 279 #endif
 280
 281 #if defined (cygwin32_HOST_OS)
 282 #define RTS_MINGW_ONLY_SYMBOLS /**/
 283 /* Don't have the ability to read import libs / archives, so
 284  * we have to stupidly list a lot of what libcygwin.a
 285  * exports; sigh.
 286  */
 287 #define RTS_CYGWIN_ONLY_SYMBOLS                          \
 288       SymI_HasProto(regfree)                             \
 289       SymI_HasProto(regexec)                             \
 290       SymI_HasProto(regerror)                            \
 291       SymI_HasProto(regcomp)                             \
 292       SymI_HasProto(__errno)                             \
 293       SymI_HasProto(access)                              \
 294       SymI_HasProto(chmod)                               \
 295       SymI_HasProto(chdir)                               \
 296       SymI_HasProto(close)                               \
 297       SymI_HasProto(creat)                               \
 298       SymI_HasProto(dup)                                 \
 299       SymI_HasProto(dup2)                                \
 300       SymI_HasProto(fstat)                               \
 301       SymI_HasProto(fcntl)                               \
 302       SymI_HasProto(getcwd)                              \
 303       SymI_HasProto(getenv)                              \
 304       SymI_HasProto(lseek)                               \
 305       SymI_HasProto(open)                                \
 306       SymI_HasProto(fpathconf)                           \
 307       SymI_HasProto(pathconf)                            \
 308       SymI_HasProto(stat)                                \
 309       SymI_HasProto(pow)                                 \
 310       SymI_HasProto(tanh)                                \
 311       SymI_HasProto(cosh)                                \
 312       SymI_HasProto(sinh)                                \
 313       SymI_HasProto(atan)                                \
 314       SymI_HasProto(acos)                                \
 315       SymI_HasProto(asin)                                \
 316       SymI_HasProto(tan)                                 \
 317       SymI_HasProto(cos)                                 \
 318       SymI_HasProto(sin)                                 \
 319       SymI_HasProto(exp)                                 \
 320       SymI_HasProto(log)                                 \
 321       SymI_HasProto(sqrt)                                \
 322       SymI_HasProto(localtime_r)                         \
 323       SymI_HasProto(gmtime_r)                            \
 324       SymI_HasProto(mktime)                              \
 325       SymI_NeedsProto(_imp___tzname)                     \
 326       SymI_HasProto(gettimeofday)                        \
 327       SymI_HasProto(timezone)                            \
 328       SymI_HasProto(tcgetattr)                           \
 329       SymI_HasProto(tcsetattr)                           \
 330       SymI_HasProto(memcpy)                              \
 331       SymI_HasProto(memmove)                             \
 332       SymI_HasProto(realloc)                             \
 333       SymI_HasProto(malloc)                              \
 334       SymI_HasProto(free)                                \
 335       SymI_HasProto(fork)                                \
 336       SymI_HasProto(lstat)                               \
 337       SymI_HasProto(isatty)                              \
 338       SymI_HasProto(mkdir)                               \
 339       SymI_HasProto(opendir)                             \
 340       SymI_HasProto(readdir)                             \
 341       SymI_HasProto(rewinddir)                           \
 342       SymI_HasProto(closedir)                            \
 343       SymI_HasProto(link)                                \
 344       SymI_HasProto(mkfifo)                              \
 345       SymI_HasProto(pipe)                                \
 346       SymI_HasProto(read)                                \
 347       SymI_HasProto(rename)                              \
 348       SymI_HasProto(rmdir)                               \
 349       SymI_HasProto(select)                              \
 350       SymI_HasProto(system)                              \
 351       SymI_HasProto(write)                               \
 352       SymI_HasProto(strcmp)                              \
 353       SymI_HasProto(strcpy)                              \
 354       SymI_HasProto(strncpy)                             \
 355       SymI_HasProto(strerror)                            \
 356       SymI_HasProto(sigaddset)                           \
 357       SymI_HasProto(sigemptyset)                         \
 358       SymI_HasProto(sigprocmask)                         \
 359       SymI_HasProto(umask)                               \
 360       SymI_HasProto(uname)                               \
 361       SymI_HasProto(unlink)                              \
 362       SymI_HasProto(utime)                               \
 363       SymI_HasProto(waitpid)
 364
 365 #elif !defined(mingw32_HOST_OS)
 366 #define RTS_MINGW_ONLY_SYMBOLS /**/
 367 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 368 #else /* defined(mingw32_HOST_OS) */
 369 #define RTS_POSIX_ONLY_SYMBOLS  /**/
 370 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 371
 372 #if HAVE_GETTIMEOFDAY
 373 #define RTS_MINGW_GETTIMEOFDAY_SYM SymI_NeedsProto(gettimeofday)
 374 #else
 375 #define RTS_MINGW_GETTIMEOFDAY_SYM /**/
 376 #endif
 377
 378 #if HAVE___MINGW_VFPRINTF
 379 #define RTS___MINGW_VFPRINTF_SYM SymI_HasProto(__mingw_vfprintf)
 380 #else
 381 #define RTS___MINGW_VFPRINTF_SYM /**/
 382 #endif
 383
 384 /* These are statically linked from the mingw libraries into the ghc
 385    executable, so we have to employ this hack. */
 386 #define RTS_MINGW_ONLY_SYMBOLS                           \
 387       SymI_HasProto(stg_asyncReadzh)                     \
 388       SymI_HasProto(stg_asyncWritezh)                    \
 389       SymI_HasProto(stg_asyncDoProczh)                   \
 390       SymI_HasProto(memset)                              \
 391       SymI_HasProto(inet_ntoa)                           \
 392       SymI_HasProto(inet_addr)                           \
 393       SymI_HasProto(htonl)                               \
 394       SymI_HasProto(recvfrom)                            \
 395       SymI_HasProto(listen)                              \
 396       SymI_HasProto(bind)                                \
 397       SymI_HasProto(shutdown)                            \
 398       SymI_HasProto(connect)                             \
 399       SymI_HasProto(htons)                               \
 400       SymI_HasProto(ntohs)                               \
 401       SymI_HasProto(getservbyname)                       \
 402       SymI_HasProto(getservbyport)                       \
 403       SymI_HasProto(getprotobynumber)                    \
 404       SymI_HasProto(getprotobyname)                      \
 405       SymI_HasProto(gethostbyname)                       \
 406       SymI_HasProto(gethostbyaddr)                       \
 407       SymI_HasProto(gethostname)                         \
 408       SymI_HasProto(strcpy)                              \
 409       SymI_HasProto(strncpy)                             \
 410       SymI_HasProto(abort)                               \
 411       SymI_NeedsProto(_alloca)                           \
 412       SymI_HasProto(isxdigit)                            \
 413       SymI_HasProto(isupper)                             \
 414       SymI_HasProto(ispunct)                             \
 415       SymI_HasProto(islower)                             \
 416       SymI_HasProto(isspace)                             \
 417       SymI_HasProto(isprint)                             \
 418       SymI_HasProto(isdigit)                             \
 419       SymI_HasProto(iscntrl)                             \
 420       SymI_HasProto(isalpha)                             \
 421       SymI_HasProto(isalnum)                             \
 422       SymI_HasProto(isascii)                             \
 423       RTS___MINGW_VFPRINTF_SYM                           \
 424       SymI_HasProto(strcmp)                              \
 425       SymI_HasProto(memmove)                             \
 426       SymI_HasProto(realloc)                             \
 427       SymI_HasProto(malloc)                              \
 428       SymI_HasProto(pow)                                 \
 429       SymI_HasProto(tanh)                                \
 430       SymI_HasProto(cosh)                                \
 431       SymI_HasProto(sinh)                                \
 432       SymI_HasProto(atan)                                \
 433       SymI_HasProto(acos)                                \
 434       SymI_HasProto(asin)                                \
 435       SymI_HasProto(tan)                                 \
 436       SymI_HasProto(cos)                                 \
 437       SymI_HasProto(sin)                                 \
 438       SymI_HasProto(exp)                                 \
 439       SymI_HasProto(log)                                 \
 440       SymI_HasProto(sqrt)                                \
 441       SymI_HasProto(powf)                                \
 442       SymI_HasProto(tanhf)                               \
 443       SymI_HasProto(coshf)                               \
 444       SymI_HasProto(sinhf)                               \
 445       SymI_HasProto(atanf)                               \
 446       SymI_HasProto(acosf)                               \
 447       SymI_HasProto(asinf)                               \
 448       SymI_HasProto(tanf)                                \
 449       SymI_HasProto(cosf)                                \
 450       SymI_HasProto(sinf)                                \
 451       SymI_HasProto(expf)                                \
 452       SymI_HasProto(logf)                                \
 453       SymI_HasProto(sqrtf)                               \
 454       SymI_HasProto(erf)                                 \
 455       SymI_HasProto(erfc)                                \
 456       SymI_HasProto(erff)                                \
 457       SymI_HasProto(erfcf)                               \
 458       SymI_HasProto(memcpy)                              \
 459       SymI_HasProto(rts_InstallConsoleEvent)             \
 460       SymI_HasProto(rts_ConsoleHandlerDone)              \
 461       SymI_NeedsProto(mktime)                            \
 462       SymI_NeedsProto(_imp___timezone)                   \
 463       SymI_NeedsProto(_imp___tzname)                     \
 464       SymI_NeedsProto(_imp__tzname)                      \
 465       SymI_NeedsProto(_imp___iob)                        \
 466       SymI_NeedsProto(_imp___osver)                      \
 467       SymI_NeedsProto(localtime)                         \
 468       SymI_NeedsProto(gmtime)                            \
 469       SymI_NeedsProto(opendir)                           \
 470       SymI_NeedsProto(readdir)                           \
 471       SymI_NeedsProto(rewinddir)                         \
 472       SymI_NeedsProto(_imp____mb_cur_max)                \
 473       SymI_NeedsProto(_imp___pctype)                     \
 474       SymI_NeedsProto(__chkstk)                          \
 475       RTS_MINGW_GETTIMEOFDAY_SYM                         \
 476       SymI_NeedsProto(closedir)
 477 #endif
 478
 479
 480 #if defined(darwin_HOST_OS) && HAVE_PRINTF_LDBLSTUB
 481 #define RTS_DARWIN_ONLY_SYMBOLS                             \
 482      SymI_NeedsProto(asprintf$LDBLStub)                     \
 483      SymI_NeedsProto(err$LDBLStub)                          \
 484      SymI_NeedsProto(errc$LDBLStub)                         \
 485      SymI_NeedsProto(errx$LDBLStub)                         \
 486      SymI_NeedsProto(fprintf$LDBLStub)                      \
 487      SymI_NeedsProto(fscanf$LDBLStub)                       \
 488      SymI_NeedsProto(fwprintf$LDBLStub)                     \
 489      SymI_NeedsProto(fwscanf$LDBLStub)                      \
 490      SymI_NeedsProto(printf$LDBLStub)                       \
 491      SymI_NeedsProto(scanf$LDBLStub)                        \
 492      SymI_NeedsProto(snprintf$LDBLStub)                     \
 493      SymI_NeedsProto(sprintf$LDBLStub)                      \
 494      SymI_NeedsProto(sscanf$LDBLStub)                       \
 495      SymI_NeedsProto(strtold$LDBLStub)                      \
 496      SymI_NeedsProto(swprintf$LDBLStub)                     \
 497      SymI_NeedsProto(swscanf$LDBLStub)                      \
 498      SymI_NeedsProto(syslog$LDBLStub)                       \
 499      SymI_NeedsProto(vasprintf$LDBLStub)                    \
 500      SymI_NeedsProto(verr$LDBLStub)                         \
 501      SymI_NeedsProto(verrc$LDBLStub)                        \
 502      SymI_NeedsProto(verrx$LDBLStub)                        \
 503      SymI_NeedsProto(vfprintf$LDBLStub)                     \
 504      SymI_NeedsProto(vfscanf$LDBLStub)                      \
 505      SymI_NeedsProto(vfwprintf$LDBLStub)                    \
 506      SymI_NeedsProto(vfwscanf$LDBLStub)                     \
 507      SymI_NeedsProto(vprintf$LDBLStub)                      \
 508      SymI_NeedsProto(vscanf$LDBLStub)                       \
 509      SymI_NeedsProto(vsnprintf$LDBLStub)                    \
 510      SymI_NeedsProto(vsprintf$LDBLStub)                     \
 511      SymI_NeedsProto(vsscanf$LDBLStub)                      \
 512      SymI_NeedsProto(vswprintf$LDBLStub)                    \
 513      SymI_NeedsProto(vswscanf$LDBLStub)                     \
 514      SymI_NeedsProto(vsyslog$LDBLStub)                      \
 515      SymI_NeedsProto(vwarn$LDBLStub)                        \
 516      SymI_NeedsProto(vwarnc$LDBLStub)                       \
 517      SymI_NeedsProto(vwarnx$LDBLStub)                       \
 518      SymI_NeedsProto(vwprintf$LDBLStub)                     \
 519      SymI_NeedsProto(vwscanf$LDBLStub)                      \
 520      SymI_NeedsProto(warn$LDBLStub)                         \
 521      SymI_NeedsProto(warnc$LDBLStub)                        \
 522      SymI_NeedsProto(warnx$LDBLStub)                        \
 523      SymI_NeedsProto(wcstold$LDBLStub)                      \
 524      SymI_NeedsProto(wprintf$LDBLStub)                      \
 525      SymI_NeedsProto(wscanf$LDBLStub)
 526 #else
 527 #define RTS_DARWIN_ONLY_SYMBOLS
 528 #endif
 529
 530 #ifndef SMP
 531 # define MAIN_CAP_SYM SymI_HasProto(MainCapability)
 532 #else
 533 # define MAIN_CAP_SYM
 534 #endif
 535
 536 #if !defined(mingw32_HOST_OS)
 537 #define RTS_USER_SIGNALS_SYMBOLS        \
 538    SymI_HasProto(setIOManagerControlFd) \
 539    SymI_HasProto(setIOManagerWakeupFd)  \
 540    SymI_HasProto(ioManagerWakeup)       \
 541    SymI_HasProto(blockUserSignals)      \
 542    SymI_HasProto(unblockUserSignals)
 543 #else
 544 #define RTS_USER_SIGNALS_SYMBOLS        \
 545    SymI_HasProto(ioManagerWakeup)       \
 546    SymI_HasProto(sendIOManagerEvent)    \
 547    SymI_HasProto(readIOManagerEvent)    \
 548    SymI_HasProto(getIOManagerEvent)     \
 549    SymI_HasProto(console_handler)
 550 #endif
 551
 552 #define RTS_LIBFFI_SYMBOLS                                  \
 553      SymE_NeedsProto(ffi_prep_cif)                          \
 554      SymE_NeedsProto(ffi_call)                              \
 555      SymE_NeedsProto(ffi_type_void)                         \
 556      SymE_NeedsProto(ffi_type_float)                        \
 557      SymE_NeedsProto(ffi_type_double)                       \
 558      SymE_NeedsProto(ffi_type_sint64)                       \
 559      SymE_NeedsProto(ffi_type_uint64)                       \
 560      SymE_NeedsProto(ffi_type_sint32)                       \
 561      SymE_NeedsProto(ffi_type_uint32)                       \
 562      SymE_NeedsProto(ffi_type_sint16)                       \
 563      SymE_NeedsProto(ffi_type_uint16)                       \
 564      SymE_NeedsProto(ffi_type_sint8)                        \
 565      SymE_NeedsProto(ffi_type_uint8)                        \
 566      SymE_NeedsProto(ffi_type_pointer)
 567
 568 #ifdef TABLES_NEXT_TO_CODE
 569 #define RTS_RET_SYMBOLS /* nothing */
 570 #else
 571 #define RTS_RET_SYMBOLS                                 \
 572       SymI_HasProto(stg_enter_ret)                      \
 573       SymI_HasProto(stg_gc_fun_ret)                     \
 574       SymI_HasProto(stg_ap_v_ret)                       \
 575       SymI_HasProto(stg_ap_f_ret)                       \
 576       SymI_HasProto(stg_ap_d_ret)                       \
 577       SymI_HasProto(stg_ap_l_ret)                       \
 578       SymI_HasProto(stg_ap_n_ret)                       \
 579       SymI_HasProto(stg_ap_p_ret)                       \
 580       SymI_HasProto(stg_ap_pv_ret)                      \
 581       SymI_HasProto(stg_ap_pp_ret)                      \
 582       SymI_HasProto(stg_ap_ppv_ret)                     \
 583       SymI_HasProto(stg_ap_ppp_ret)                     \
 584       SymI_HasProto(stg_ap_pppv_ret)                    \
 585       SymI_HasProto(stg_ap_pppp_ret)                    \
 586       SymI_HasProto(stg_ap_ppppp_ret)                   \
 587       SymI_HasProto(stg_ap_pppppp_ret)
 588 #endif
 589
 590 /* Modules compiled with -ticky may mention ticky counters */
 591 /* This list should marry up with the one in $(TOP)/includes/stg/Ticky.h */
 592 #define RTS_TICKY_SYMBOLS                               \
 593       SymI_NeedsProto(ticky_entry_ctrs)                 \
 594       SymI_NeedsProto(top_ct)                           \
 595                                                         \
 596       SymI_HasProto(ENT_VIA_NODE_ctr)                   \
 597       SymI_HasProto(ENT_STATIC_THK_ctr)                 \
 598       SymI_HasProto(ENT_DYN_THK_ctr)                    \
 599       SymI_HasProto(ENT_STATIC_FUN_DIRECT_ctr)          \
 600       SymI_HasProto(ENT_DYN_FUN_DIRECT_ctr)             \
 601       SymI_HasProto(ENT_STATIC_CON_ctr)                 \
 602       SymI_HasProto(ENT_DYN_CON_ctr)                    \
 603       SymI_HasProto(ENT_STATIC_IND_ctr)                 \
 604       SymI_HasProto(ENT_DYN_IND_ctr)                    \
 605       SymI_HasProto(ENT_PERM_IND_ctr)                   \
 606       SymI_HasProto(ENT_PAP_ctr)                        \
 607       SymI_HasProto(ENT_AP_ctr)                         \
 608       SymI_HasProto(ENT_AP_STACK_ctr)                   \
 609       SymI_HasProto(ENT_BH_ctr)                         \
 610       SymI_HasProto(UNKNOWN_CALL_ctr)                   \
 611       SymI_HasProto(SLOW_CALL_v_ctr)                    \
 612       SymI_HasProto(SLOW_CALL_f_ctr)                    \
 613       SymI_HasProto(SLOW_CALL_d_ctr)                    \
 614       SymI_HasProto(SLOW_CALL_l_ctr)                    \
 615       SymI_HasProto(SLOW_CALL_n_ctr)                    \
 616       SymI_HasProto(SLOW_CALL_p_ctr)                    \
 617       SymI_HasProto(SLOW_CALL_pv_ctr)                   \
 618       SymI_HasProto(SLOW_CALL_pp_ctr)                   \
 619       SymI_HasProto(SLOW_CALL_ppv_ctr)                  \
 620       SymI_HasProto(SLOW_CALL_ppp_ctr)                  \
 621       SymI_HasProto(SLOW_CALL_pppv_ctr)                 \
 622       SymI_HasProto(SLOW_CALL_pppp_ctr)                 \
 623       SymI_HasProto(SLOW_CALL_ppppp_ctr)                \
 624       SymI_HasProto(SLOW_CALL_pppppp_ctr)               \
 625       SymI_HasProto(SLOW_CALL_OTHER_ctr)                \
 626       SymI_HasProto(ticky_slow_call_unevald)            \
 627       SymI_HasProto(SLOW_CALL_ctr)                      \
 628       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_ctr)          \
 629       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_CHUNKS_ctr)   \
 630       SymI_HasProto(KNOWN_CALL_ctr)                     \
 631       SymI_HasProto(KNOWN_CALL_TOO_FEW_ARGS_ctr)        \
 632       SymI_HasProto(KNOWN_CALL_EXTRA_ARGS_ctr)          \
 633       SymI_HasProto(SLOW_CALL_FUN_TOO_FEW_ctr)          \
 634       SymI_HasProto(SLOW_CALL_FUN_CORRECT_ctr)          \
 635       SymI_HasProto(SLOW_CALL_FUN_TOO_MANY_ctr)         \
 636       SymI_HasProto(SLOW_CALL_PAP_TOO_FEW_ctr)          \
 637       SymI_HasProto(SLOW_CALL_PAP_CORRECT_ctr)          \
 638       SymI_HasProto(SLOW_CALL_PAP_TOO_MANY_ctr)         \
 639       SymI_HasProto(SLOW_CALL_UNEVALD_ctr)              \
 640       SymI_HasProto(UPDF_OMITTED_ctr)                   \
 641       SymI_HasProto(UPDF_PUSHED_ctr)                    \
 642       SymI_HasProto(CATCHF_PUSHED_ctr)                  \
 643       SymI_HasProto(UPDF_RCC_PUSHED_ctr)                \
 644       SymI_HasProto(UPDF_RCC_OMITTED_ctr)               \
 645       SymI_HasProto(UPD_SQUEEZED_ctr)                   \
 646       SymI_HasProto(UPD_CON_IN_NEW_ctr)                 \
 647       SymI_HasProto(UPD_CON_IN_PLACE_ctr)               \
 648       SymI_HasProto(UPD_PAP_IN_NEW_ctr)                 \
 649       SymI_HasProto(UPD_PAP_IN_PLACE_ctr)               \
 650       SymI_HasProto(ALLOC_HEAP_ctr)                     \
 651       SymI_HasProto(ALLOC_HEAP_tot)                     \
 652       SymI_HasProto(ALLOC_FUN_ctr)                      \
 653       SymI_HasProto(ALLOC_FUN_adm)                      \
 654       SymI_HasProto(ALLOC_FUN_gds)                      \
 655       SymI_HasProto(ALLOC_FUN_slp)                      \
 656       SymI_HasProto(UPD_NEW_IND_ctr)                    \
 657       SymI_HasProto(UPD_NEW_PERM_IND_ctr)               \
 658       SymI_HasProto(UPD_OLD_IND_ctr)                    \
 659       SymI_HasProto(UPD_OLD_PERM_IND_ctr)               \
 660       SymI_HasProto(UPD_BH_UPDATABLE_ctr)               \
 661       SymI_HasProto(UPD_BH_SINGLE_ENTRY_ctr)            \
 662       SymI_HasProto(UPD_CAF_BH_UPDATABLE_ctr)           \
 663       SymI_HasProto(UPD_CAF_BH_SINGLE_ENTRY_ctr)        \
 664       SymI_HasProto(GC_SEL_ABANDONED_ctr)               \
 665       SymI_HasProto(GC_SEL_MINOR_ctr)                   \
 666       SymI_HasProto(GC_SEL_MAJOR_ctr)                   \
 667       SymI_HasProto(GC_FAILED_PROMOTION_ctr)            \
 668       SymI_HasProto(ALLOC_UP_THK_ctr)                   \
 669       SymI_HasProto(ALLOC_SE_THK_ctr)                   \
 670       SymI_HasProto(ALLOC_THK_adm)                      \
 671       SymI_HasProto(ALLOC_THK_gds)                      \
 672       SymI_HasProto(ALLOC_THK_slp)                      \
 673       SymI_HasProto(ALLOC_CON_ctr)                      \
 674       SymI_HasProto(ALLOC_CON_adm)                      \
 675       SymI_HasProto(ALLOC_CON_gds)                      \
 676       SymI_HasProto(ALLOC_CON_slp)                      \
 677       SymI_HasProto(ALLOC_TUP_ctr)                      \
 678       SymI_HasProto(ALLOC_TUP_adm)                      \
 679       SymI_HasProto(ALLOC_TUP_gds)                      \
 680       SymI_HasProto(ALLOC_TUP_slp)                      \
 681       SymI_HasProto(ALLOC_BH_ctr)                       \
 682       SymI_HasProto(ALLOC_BH_adm)                       \
 683       SymI_HasProto(ALLOC_BH_gds)                       \
 684       SymI_HasProto(ALLOC_BH_slp)                       \
 685       SymI_HasProto(ALLOC_PRIM_ctr)                     \
 686       SymI_HasProto(ALLOC_PRIM_adm)                     \
 687       SymI_HasProto(ALLOC_PRIM_gds)                     \
 688       SymI_HasProto(ALLOC_PRIM_slp)                     \
 689       SymI_HasProto(ALLOC_PAP_ctr)                      \
 690       SymI_HasProto(ALLOC_PAP_adm)                      \
 691       SymI_HasProto(ALLOC_PAP_gds)                      \
 692       SymI_HasProto(ALLOC_PAP_slp)                      \
 693       SymI_HasProto(ALLOC_TSO_ctr)                      \
 694       SymI_HasProto(ALLOC_TSO_adm)                      \
 695       SymI_HasProto(ALLOC_TSO_gds)                      \
 696       SymI_HasProto(ALLOC_TSO_slp)                      \
 697       SymI_HasProto(RET_NEW_ctr)                        \
 698       SymI_HasProto(RET_OLD_ctr)                        \
 699       SymI_HasProto(RET_UNBOXED_TUP_ctr)                \
 700       SymI_HasProto(RET_SEMI_loads_avoided)
 701
 702
 703 // On most platforms, the garbage collector rewrites references
 704 //      to small integer and char objects to a set of common, shared ones.
 705 //
 706 // We don't do this when compiling to Windows DLLs at the moment because
 707 //      it doesn't support cross package data references well.
 708 //
 709 #if defined(__PIC__) && defined(mingw32_HOST_OS)
 710 #define RTS_INTCHAR_SYMBOLS
 711 #else
 712 #define RTS_INTCHAR_SYMBOLS                             \
 713       SymI_HasProto(stg_CHARLIKE_closure)               \
 714       SymI_HasProto(stg_INTLIKE_closure)
 715 #endif
 716
 717
 718 #define RTS_SYMBOLS                                     \
 719       Maybe_Stable_Names                                \
 720       RTS_TICKY_SYMBOLS                                 \
 721       SymI_HasProto(StgReturn)                          \
 722       SymI_HasProto(stg_enter_info)                     \
 723       SymI_HasProto(stg_gc_void_info)                   \
 724       SymI_HasProto(__stg_gc_enter_1)                   \
 725       SymI_HasProto(stg_gc_noregs)                      \
 726       SymI_HasProto(stg_gc_unpt_r1_info)                \
 727       SymI_HasProto(stg_gc_unpt_r1)                     \
 728       SymI_HasProto(stg_gc_unbx_r1_info)                \
 729       SymI_HasProto(stg_gc_unbx_r1)                     \
 730       SymI_HasProto(stg_gc_f1_info)                     \
 731       SymI_HasProto(stg_gc_f1)                          \
 732       SymI_HasProto(stg_gc_d1_info)                     \
 733       SymI_HasProto(stg_gc_d1)                          \
 734       SymI_HasProto(stg_gc_l1_info)                     \
 735       SymI_HasProto(stg_gc_l1)                          \
 736       SymI_HasProto(__stg_gc_fun)                       \
 737       SymI_HasProto(stg_gc_fun_info)                    \
 738       SymI_HasProto(stg_gc_gen)                         \
 739       SymI_HasProto(stg_gc_gen_info)                    \
 740       SymI_HasProto(stg_gc_gen_hp)                      \
 741       SymI_HasProto(stg_gc_ut)                          \
 742       SymI_HasProto(stg_gen_yield)                      \
 743       SymI_HasProto(stg_yield_noregs)                   \
 744       SymI_HasProto(stg_yield_to_interpreter)           \
 745       SymI_HasProto(stg_gen_block)                      \
 746       SymI_HasProto(stg_block_noregs)                   \
 747       SymI_HasProto(stg_block_1)                        \
 748       SymI_HasProto(stg_block_takemvar)                 \
 749       SymI_HasProto(stg_block_putmvar)                  \
 750       MAIN_CAP_SYM                                      \
 751       SymI_HasProto(MallocFailHook)                     \
 752       SymI_HasProto(OnExitHook)                         \
 753       SymI_HasProto(OutOfHeapHook)                      \
 754       SymI_HasProto(StackOverflowHook)                  \
 755       SymI_HasProto(addDLL)                             \
 756       SymI_HasProto(__int_encodeDouble)                 \
 757       SymI_HasProto(__word_encodeDouble)                \
 758       SymI_HasProto(__2Int_encodeDouble)                \
 759       SymI_HasProto(__int_encodeFloat)                  \
 760       SymI_HasProto(__word_encodeFloat)                 \
 761       SymI_HasProto(stg_atomicallyzh)                   \
 762       SymI_HasProto(barf)                               \
 763       SymI_HasProto(debugBelch)                         \
 764       SymI_HasProto(errorBelch)                         \
 765       SymI_HasProto(sysErrorBelch)                      \
 766       SymI_HasProto(stg_getMaskingStatezh)              \
 767       SymI_HasProto(stg_maskAsyncExceptionszh)          \
 768       SymI_HasProto(stg_maskUninterruptiblezh)          \
 769       SymI_HasProto(stg_catchzh)                        \
 770       SymI_HasProto(stg_catchRetryzh)                   \
 771       SymI_HasProto(stg_catchSTMzh)                     \
 772       SymI_HasProto(stg_checkzh)                        \
 773       SymI_HasProto(closure_flags)                      \
 774       SymI_HasProto(cmp_thread)                         \
 775       SymI_HasProto(createAdjustor)                     \
 776       SymI_HasProto(stg_decodeDoublezu2Intzh)           \
 777       SymI_HasProto(stg_decodeFloatzuIntzh)             \
 778       SymI_HasProto(defaultsHook)                       \
 779       SymI_HasProto(stg_delayzh)                        \
 780       SymI_HasProto(stg_deRefWeakzh)                    \
 781       SymI_HasProto(stg_deRefStablePtrzh)               \
 782       SymI_HasProto(dirty_MUT_VAR)                      \
 783       SymI_HasProto(stg_forkzh)                         \
 784       SymI_HasProto(stg_forkOnzh)                       \
 785       SymI_HasProto(forkProcess)                        \
 786       SymI_HasProto(forkOS_createThread)                \
 787       SymI_HasProto(freeHaskellFunctionPtr)             \
 788       SymI_HasProto(getOrSetTypeableStore)              \
 789       SymI_HasProto(getOrSetGHCConcSignalSignalHandlerStore)            \
 790       SymI_HasProto(getOrSetGHCConcWindowsPendingDelaysStore)           \
 791       SymI_HasProto(getOrSetGHCConcWindowsIOManagerThreadStore)         \
 792       SymI_HasProto(getOrSetGHCConcWindowsProddingStore)                \
 793       SymI_HasProto(getOrSetSystemEventThreadEventManagerStore)         \
 794       SymI_HasProto(getOrSetSystemEventThreadIOManagerThreadStore)      \
 795       SymI_HasProto(genSymZh)                           \
 796       SymI_HasProto(genericRaise)                       \
 797       SymI_HasProto(getProgArgv)                        \
 798       SymI_HasProto(getFullProgArgv)                    \
 799       SymI_HasProto(getStablePtr)                       \
 800       SymI_HasProto(hs_init)                            \
 801       SymI_HasProto(hs_exit)                            \
 802       SymI_HasProto(hs_set_argv)                        \
 803       SymI_HasProto(hs_add_root)                        \
 804       SymI_HasProto(hs_perform_gc)                      \
 805       SymI_HasProto(hs_free_stable_ptr)                 \
 806       SymI_HasProto(hs_free_fun_ptr)                    \
 807       SymI_HasProto(hs_hpc_rootModule)                  \
 808       SymI_HasProto(hs_hpc_module)                      \
 809       SymI_HasProto(initLinker)                         \
 810       SymI_HasProto(stg_unpackClosurezh)                \
 811       SymI_HasProto(stg_getApStackValzh)                \
 812       SymI_HasProto(stg_getSparkzh)                     \
 813       SymI_HasProto(stg_numSparkszh)                    \
 814       SymI_HasProto(stg_isCurrentThreadBoundzh)         \
 815       SymI_HasProto(stg_isEmptyMVarzh)                  \
 816       SymI_HasProto(stg_killThreadzh)                   \
 817       SymI_HasProto(loadArchive)                        \
 818       SymI_HasProto(loadObj)                            \
 819       SymI_HasProto(insertStableSymbol)                 \
 820       SymI_HasProto(insertSymbol)                       \
 821       SymI_HasProto(lookupSymbol)                       \
 822       SymI_HasProto(stg_makeStablePtrzh)                \
 823       SymI_HasProto(stg_mkApUpd0zh)                     \
 824       SymI_HasProto(stg_myThreadIdzh)                   \
 825       SymI_HasProto(stg_labelThreadzh)                  \
 826       SymI_HasProto(stg_newArrayzh)                     \
 827       SymI_HasProto(stg_newBCOzh)                       \
 828       SymI_HasProto(stg_newByteArrayzh)                 \
 829       SymI_HasProto_redirect(newCAF, newDynCAF)         \
 830       SymI_HasProto(stg_newMVarzh)                      \
 831       SymI_HasProto(stg_newMutVarzh)                    \
 832       SymI_HasProto(stg_newTVarzh)                      \
 833       SymI_HasProto(stg_noDuplicatezh)                  \
 834       SymI_HasProto(stg_atomicModifyMutVarzh)           \
 835       SymI_HasProto(stg_casMutVarzh)                    \
 836       SymI_HasProto(stg_newPinnedByteArrayzh)           \
 837       SymI_HasProto(stg_newAlignedPinnedByteArrayzh)    \
 838       SymI_HasProto(newSpark)                           \
 839       SymI_HasProto(performGC)                          \
 840       SymI_HasProto(performMajorGC)                     \
 841       SymI_HasProto(prog_argc)                          \
 842       SymI_HasProto(prog_argv)                          \
 843       SymI_HasProto(stg_putMVarzh)                      \
 844       SymI_HasProto(stg_raisezh)                        \
 845       SymI_HasProto(stg_raiseIOzh)                      \
 846       SymI_HasProto(stg_readTVarzh)                     \
 847       SymI_HasProto(stg_readTVarIOzh)                   \
 848       SymI_HasProto(resumeThread)                       \
 849       SymI_HasProto(resolveObjs)                        \
 850       SymI_HasProto(stg_retryzh)                        \
 851       SymI_HasProto(rts_apply)                          \
 852       SymI_HasProto(rts_checkSchedStatus)               \
 853       SymI_HasProto(rts_eval)                           \
 854       SymI_HasProto(rts_evalIO)                         \
 855       SymI_HasProto(rts_evalLazyIO)                     \
 856       SymI_HasProto(rts_evalStableIO)                   \
 857       SymI_HasProto(rts_eval_)                          \
 858       SymI_HasProto(rts_getBool)                        \
 859       SymI_HasProto(rts_getChar)                        \
 860       SymI_HasProto(rts_getDouble)                      \
 861       SymI_HasProto(rts_getFloat)                       \
 862       SymI_HasProto(rts_getInt)                         \
 863       SymI_HasProto(rts_getInt8)                        \
 864       SymI_HasProto(rts_getInt16)                       \
 865       SymI_HasProto(rts_getInt32)                       \
 866       SymI_HasProto(rts_getInt64)                       \
 867       SymI_HasProto(rts_getPtr)                         \
 868       SymI_HasProto(rts_getFunPtr)                      \
 869       SymI_HasProto(rts_getStablePtr)                   \
 870       SymI_HasProto(rts_getThreadId)                    \
 871       SymI_HasProto(rts_getWord)                        \
 872       SymI_HasProto(rts_getWord8)                       \
 873       SymI_HasProto(rts_getWord16)                      \
 874       SymI_HasProto(rts_getWord32)                      \
 875       SymI_HasProto(rts_getWord64)                      \
 876       SymI_HasProto(rts_lock)                           \
 877       SymI_HasProto(rts_mkBool)                         \
 878       SymI_HasProto(rts_mkChar)                         \
 879       SymI_HasProto(rts_mkDouble)                       \
 880       SymI_HasProto(rts_mkFloat)                        \
 881       SymI_HasProto(rts_mkInt)                          \
 882       SymI_HasProto(rts_mkInt8)                         \
 883       SymI_HasProto(rts_mkInt16)                        \
 884       SymI_HasProto(rts_mkInt32)                        \
 885       SymI_HasProto(rts_mkInt64)                        \
 886       SymI_HasProto(rts_mkPtr)                          \
 887       SymI_HasProto(rts_mkFunPtr)                       \
 888       SymI_HasProto(rts_mkStablePtr)                    \
 889       SymI_HasProto(rts_mkString)                       \
 890       SymI_HasProto(rts_mkWord)                         \
 891       SymI_HasProto(rts_mkWord8)                        \
 892       SymI_HasProto(rts_mkWord16)                       \
 893       SymI_HasProto(rts_mkWord32)                       \
 894       SymI_HasProto(rts_mkWord64)                       \
 895       SymI_HasProto(rts_unlock)                         \
 896       SymI_HasProto(rts_unsafeGetMyCapability)          \
 897       SymI_HasProto(rtsSupportsBoundThreads)            \
 898       SymI_HasProto(rts_isProfiled)                     \
 899       SymI_HasProto(setProgArgv)                        \
 900       SymI_HasProto(startupHaskell)                     \
 901       SymI_HasProto(shutdownHaskell)                    \
 902       SymI_HasProto(shutdownHaskellAndExit)             \
 903       SymI_HasProto(stable_ptr_table)                   \
 904       SymI_HasProto(stackOverflow)                      \
 905       SymI_HasProto(stg_CAF_BLACKHOLE_info)             \
 906       SymI_HasProto(stg_BLACKHOLE_info)                 \
 907       SymI_HasProto(__stg_EAGER_BLACKHOLE_info)         \
 908       SymI_HasProto(stg_BLOCKING_QUEUE_CLEAN_info)      \
 909       SymI_HasProto(stg_BLOCKING_QUEUE_DIRTY_info)      \
 910       SymI_HasProto(startTimer)                         \
 911       SymI_HasProto(stg_MVAR_CLEAN_info)                \
 912       SymI_HasProto(stg_MVAR_DIRTY_info)                \
 913       SymI_HasProto(stg_IND_STATIC_info)                \
 914       SymI_HasProto(stg_ARR_WORDS_info)                 \
 915       SymI_HasProto(stg_MUT_ARR_PTRS_DIRTY_info)        \
 916       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN_info)       \
 917       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN0_info)      \
 918       SymI_HasProto(stg_WEAK_info)                      \
 919       SymI_HasProto(stg_ap_v_info)                      \
 920       SymI_HasProto(stg_ap_f_info)                      \
 921       SymI_HasProto(stg_ap_d_info)                      \
 922       SymI_HasProto(stg_ap_l_info)                      \
 923       SymI_HasProto(stg_ap_n_info)                      \
 924       SymI_HasProto(stg_ap_p_info)                      \
 925       SymI_HasProto(stg_ap_pv_info)                     \
 926       SymI_HasProto(stg_ap_pp_info)                     \
 927       SymI_HasProto(stg_ap_ppv_info)                    \
 928       SymI_HasProto(stg_ap_ppp_info)                    \
 929       SymI_HasProto(stg_ap_pppv_info)                   \
 930       SymI_HasProto(stg_ap_pppp_info)                   \
 931       SymI_HasProto(stg_ap_ppppp_info)                  \
 932       SymI_HasProto(stg_ap_pppppp_info)                 \
 933       SymI_HasProto(stg_ap_0_fast)                      \
 934       SymI_HasProto(stg_ap_v_fast)                      \
 935       SymI_HasProto(stg_ap_f_fast)                      \
 936       SymI_HasProto(stg_ap_d_fast)                      \
 937       SymI_HasProto(stg_ap_l_fast)                      \
 938       SymI_HasProto(stg_ap_n_fast)                      \
 939       SymI_HasProto(stg_ap_p_fast)                      \
 940       SymI_HasProto(stg_ap_pv_fast)                     \
 941       SymI_HasProto(stg_ap_pp_fast)                     \
 942       SymI_HasProto(stg_ap_ppv_fast)                    \
 943       SymI_HasProto(stg_ap_ppp_fast)                    \
 944       SymI_HasProto(stg_ap_pppv_fast)                   \
 945       SymI_HasProto(stg_ap_pppp_fast)                   \
 946       SymI_HasProto(stg_ap_ppppp_fast)                  \
 947       SymI_HasProto(stg_ap_pppppp_fast)                 \
 948       SymI_HasProto(stg_ap_1_upd_info)                  \
 949       SymI_HasProto(stg_ap_2_upd_info)                  \
 950       SymI_HasProto(stg_ap_3_upd_info)                  \
 951       SymI_HasProto(stg_ap_4_upd_info)                  \
 952       SymI_HasProto(stg_ap_5_upd_info)                  \
 953       SymI_HasProto(stg_ap_6_upd_info)                  \
 954       SymI_HasProto(stg_ap_7_upd_info)                  \
 955       SymI_HasProto(stg_exit)                           \
 956       SymI_HasProto(stg_sel_0_upd_info)                 \
 957       SymI_HasProto(stg_sel_10_upd_info)                \
 958       SymI_HasProto(stg_sel_11_upd_info)                \
 959       SymI_HasProto(stg_sel_12_upd_info)                \
 960       SymI_HasProto(stg_sel_13_upd_info)                \
 961       SymI_HasProto(stg_sel_14_upd_info)                \
 962       SymI_HasProto(stg_sel_15_upd_info)                \
 963       SymI_HasProto(stg_sel_1_upd_info)                 \
 964       SymI_HasProto(stg_sel_2_upd_info)                 \
 965       SymI_HasProto(stg_sel_3_upd_info)                 \
 966       SymI_HasProto(stg_sel_4_upd_info)                 \
 967       SymI_HasProto(stg_sel_5_upd_info)                 \
 968       SymI_HasProto(stg_sel_6_upd_info)                 \
 969       SymI_HasProto(stg_sel_7_upd_info)                 \
 970       SymI_HasProto(stg_sel_8_upd_info)                 \
 971       SymI_HasProto(stg_sel_9_upd_info)                 \
 972       SymI_HasProto(stg_upd_frame_info)                 \
 973       SymI_HasProto(stg_bh_upd_frame_info)              \
 974       SymI_HasProto(suspendThread)                      \
 975       SymI_HasProto(stg_takeMVarzh)                     \
 976       SymI_HasProto(stg_threadStatuszh)                 \
 977       SymI_HasProto(stg_tryPutMVarzh)                   \
 978       SymI_HasProto(stg_tryTakeMVarzh)                  \
 979       SymI_HasProto(stg_unmaskAsyncExceptionszh)        \
 980       SymI_HasProto(unloadObj)                          \
 981       SymI_HasProto(stg_unsafeThawArrayzh)              \
 982       SymI_HasProto(stg_waitReadzh)                     \
 983       SymI_HasProto(stg_waitWritezh)                    \
 984       SymI_HasProto(stg_writeTVarzh)                    \
 985       SymI_HasProto(stg_yieldzh)                        \
 986       SymI_NeedsProto(stg_interp_constr_entry)          \
 987       SymI_HasProto(stg_arg_bitmaps)                    \
 988       SymI_HasProto(large_alloc_lim)                    \
 989       SymI_HasProto(g0)                                 \
 990       SymI_HasProto(allocate)                           \
 991       SymI_HasProto(allocateExec)                       \
 992       SymI_HasProto(freeExec)                           \
 993       SymI_HasProto(getAllocations)                     \
 994       SymI_HasProto(revertCAFs)                         \
 995       SymI_HasProto(RtsFlags)                           \
 996       SymI_NeedsProto(rts_breakpoint_io_action)         \
 997       SymI_NeedsProto(rts_stop_next_breakpoint)         \
 998       SymI_NeedsProto(rts_stop_on_exception)            \
 999       SymI_HasProto(stopTimer)                          \
1000       SymI_HasProto(n_capabilities)                     \
1001       SymI_HasProto(stg_traceCcszh)                     \
1002       SymI_HasProto(stg_traceEventzh)                   \
1003       RTS_USER_SIGNALS_SYMBOLS                          \
1004       RTS_INTCHAR_SYMBOLS
1005
1006
1007 // 64-bit support functions in libgcc.a
1008 #if defined(__GNUC__) && SIZEOF_VOID_P <= 4
1009 #define RTS_LIBGCC_SYMBOLS                             \
1010       SymI_NeedsProto(__divdi3)                        \
1011       SymI_NeedsProto(__udivdi3)                       \
1012       SymI_NeedsProto(__moddi3)                        \
1013       SymI_NeedsProto(__umoddi3)                       \
1014       SymI_NeedsProto(__muldi3)                        \
1015       SymI_NeedsProto(__ashldi3)                       \
1016       SymI_NeedsProto(__ashrdi3)                       \
1017       SymI_NeedsProto(__lshrdi3)
1018 #else
1019 #define RTS_LIBGCC_SYMBOLS
1020 #endif
1021
1022 #if defined(darwin_HOST_OS) && defined(powerpc_HOST_ARCH)
1023       // Symbols that don't have a leading underscore
1024       // on Mac OS X. They have to receive special treatment,
1025       // see machoInitSymbolsWithoutUnderscore()
1026 #define RTS_MACHO_NOUNDERLINE_SYMBOLS                   \
1027       SymI_NeedsProto(saveFP)                           \
1028       SymI_NeedsProto(restFP)
1029 #endif
1030
1031 /* entirely bogus claims about types of these symbols */
1032 #define SymI_NeedsProto(vvv)  extern void vvv(void);
1033 #if defined(__PIC__) && defined(mingw32_HOST_OS)
1034 #define SymE_HasProto(vvv)    SymE_HasProto(vvv);
1035 #define SymE_NeedsProto(vvv)    extern void _imp__ ## vvv (void);
1036 #else
1037 #define SymE_NeedsProto(vvv)  SymI_NeedsProto(vvv);
1038 #define SymE_HasProto(vvv)    SymI_HasProto(vvv)
1039 #endif
1040 #define SymI_HasProto(vvv) /**/
1041 #define SymI_HasProto_redirect(vvv,xxx) /**/
1042 RTS_SYMBOLS
1043 RTS_RET_SYMBOLS
1044 RTS_POSIX_ONLY_SYMBOLS
1045 RTS_MINGW_ONLY_SYMBOLS
1046 RTS_CYGWIN_ONLY_SYMBOLS
1047 RTS_DARWIN_ONLY_SYMBOLS
1048 RTS_LIBGCC_SYMBOLS
1049 RTS_LIBFFI_SYMBOLS
1050 #undef SymI_NeedsProto
1051 #undef SymI_HasProto
1052 #undef SymI_HasProto_redirect
1053 #undef SymE_HasProto
1054 #undef SymE_NeedsProto
1055
1056 #ifdef LEADING_UNDERSCORE
1057 #define MAYBE_LEADING_UNDERSCORE_STR(s) ("_" s)
1058 #else
1059 #define MAYBE_LEADING_UNDERSCORE_STR(s) (s)
1060 #endif
1061
1062 #define SymI_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1063                     (void*)(&(vvv)) },
1064 #define SymE_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1065             (void*)DLL_IMPORT_DATA_REF(vvv) },
1066
1067 #define SymI_NeedsProto(vvv) SymI_HasProto(vvv)
1068 #define SymE_NeedsProto(vvv) SymE_HasProto(vvv)
1069
1070 // SymI_HasProto_redirect allows us to redirect references to one symbol to
1071 // another symbol.  See newCAF/newDynCAF for an example.
1072 #define SymI_HasProto_redirect(vvv,xxx)   \
1073     { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1074       (void*)(&(xxx)) },
1075
1076 static RtsSymbolVal rtsSyms[] = {
1077       RTS_SYMBOLS
1078       RTS_RET_SYMBOLS
1079       RTS_POSIX_ONLY_SYMBOLS
1080       RTS_MINGW_ONLY_SYMBOLS
1081       RTS_CYGWIN_ONLY_SYMBOLS
1082       RTS_DARWIN_ONLY_SYMBOLS
1083       RTS_LIBGCC_SYMBOLS
1084       RTS_LIBFFI_SYMBOLS
1085 #if defined(darwin_HOST_OS) && defined(i386_HOST_ARCH)
1086       // dyld stub code contains references to this,
1087       // but it should never be called because we treat
1088       // lazy pointers as nonlazy.
1089       { "dyld_stub_binding_helper", (void*)0xDEADBEEF },
1090 #endif
1091       { 0, 0 } /* sentinel */
1092 };
1093
1094
1095
1096 /* -----------------------------------------------------------------------------
1097  * Insert symbols into hash tables, checking for duplicates.
1098  */
1099
1100 static void ghciInsertStrHashTable ( char* obj_name,
1101                                      HashTable *table,
1102                                      char* key,
1103                                      void *data
1104                                    )
1105 {
1106    if (lookupHashTable(table, (StgWord)key) == NULL)
1107    {
1108       insertStrHashTable(table, (StgWord)key, data);
1109       return;
1110    }
1111    debugBelch(
1112       "\n\n"
1113       "GHCi runtime linker: fatal error: I found a duplicate definition for symbol\n"
1114       "   %s\n"
1115       "whilst processing object file\n"
1116       "   %s\n"
1117       "This could be caused by:\n"
1118       "   * Loading two different object files which export the same symbol\n"
1119       "   * Specifying the same object file twice on the GHCi command line\n"
1120       "   * An incorrect `package.conf' entry, causing some object to be\n"
1121       "     loaded twice.\n"
1122       "GHCi cannot safely continue in this situation.  Exiting now.  Sorry.\n"
1123       "\n",
1124       (char*)key,
1125       obj_name
1126    );
1127    stg_exit(1);
1128 }
1129 /* -----------------------------------------------------------------------------
1130  * initialize the object linker
1131  */
1132
1133
1134 static int linker_init_done = 0 ;
1135
1136 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1137 static void *dl_prog_handle;
1138 static regex_t re_invalid;
1139 static regex_t re_realso;
1140 #ifdef THREADED_RTS
1141 static Mutex dl_mutex; // mutex to protect dlopen/dlerror critical section
1142 #endif
1143 #endif
1144
1145 void
1146 initLinker( void )
1147 {
1148     RtsSymbolVal *sym;
1149 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1150     int compileResult;
1151 #endif
1152
1153     IF_DEBUG(linker, debugBelch("initLinker: start\n"));
1154
1155     /* Make initLinker idempotent, so we can call it
1156        before evey relevant operation; that means we
1157        don't need to initialise the linker separately */
1158     if (linker_init_done == 1) {
1159         IF_DEBUG(linker, debugBelch("initLinker: idempotent return\n"));
1160         return;
1161     } else {
1162         linker_init_done = 1;
1163     }
1164
1165 #if defined(THREADED_RTS) && (defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO))
1166     initMutex(&dl_mutex);
1167 #endif
1168     stablehash = allocStrHashTable();
1169     symhash = allocStrHashTable();
1170
1171     /* populate the symbol table with stuff from the RTS */
1172     for (sym = rtsSyms; sym->lbl != NULL; sym++) {
1173         ghciInsertStrHashTable("(GHCi built-in symbols)",
1174                                symhash, sym->lbl, sym->addr);
1175         IF_DEBUG(linker, debugBelch("initLinker: inserting rts symbol %s, %p\n", sym->lbl, sym->addr));
1176     }
1177 #   if defined(OBJFORMAT_MACHO) && defined(powerpc_HOST_ARCH)
1178     machoInitSymbolsWithoutUnderscore();
1179 #   endif
1180
1181 #   if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1182 #   if defined(RTLD_DEFAULT)
1183     dl_prog_handle = RTLD_DEFAULT;
1184 #   else
1185     dl_prog_handle = dlopen(NULL, RTLD_LAZY);
1186 #   endif /* RTLD_DEFAULT */
1187
1188     compileResult = regcomp(&re_invalid,
1189            "(([^ \t()])+\\.so([^ \t:()])*):([ \t])*(invalid ELF header|file too short)",
1190            REG_EXTENDED);
1191     ASSERT( compileResult == 0 );
1192     compileResult = regcomp(&re_realso,
1193            "(GROUP|INPUT) *\\( *(([^ )])+)",
1194            REG_EXTENDED);
1195     ASSERT( compileResult == 0 );
1196 #   endif
1197
1198 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1199     if (RtsFlags.MiscFlags.linkerMemBase != 0) {
1200         // User-override for mmap_32bit_base
1201         mmap_32bit_base = (void*)RtsFlags.MiscFlags.linkerMemBase;
1202     }
1203 #endif
1204
1205 #if defined(mingw32_HOST_OS)
1206     /*
1207      * These two libraries cause problems when added to the static link,
1208      * but are necessary for resolving symbols in GHCi, hence we load
1209      * them manually here.
1210      */
1211     addDLL("msvcrt");
1212     addDLL("kernel32");
1213 #endif
1214
1215     IF_DEBUG(linker, debugBelch("initLinker: done\n"));
1216     return;
1217 }
1218
1219 void
1220 exitLinker( void ) {
1221 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1222    if (linker_init_done == 1) {
1223       regfree(&re_invalid);
1224       regfree(&re_realso);
1225 #ifdef THREADED_RTS
1226       closeMutex(&dl_mutex);
1227 #endif
1228    }
1229 #endif
1230 }
1231
1232 /* -----------------------------------------------------------------------------
1233  *                  Loading DLL or .so dynamic libraries
1234  * -----------------------------------------------------------------------------
1235  *
1236  * Add a DLL from which symbols may be found.  In the ELF case, just
1237  * do RTLD_GLOBAL-style add, so no further messing around needs to
1238  * happen in order that symbols in the loaded .so are findable --
1239  * lookupSymbol() will subsequently see them by dlsym on the program's
1240  * dl-handle.  Returns NULL if success, otherwise ptr to an err msg.
1241  *
1242  * In the PEi386 case, open the DLLs and put handles to them in a
1243  * linked list.  When looking for a symbol, try all handles in the
1244  * list.  This means that we need to load even DLLs that are guaranteed
1245  * to be in the ghc.exe image already, just so we can get a handle
1246  * to give to loadSymbol, so that we can find the symbols.  For such
1247  * libraries, the LoadLibrary call should be a no-op except for returning
1248  * the handle.
1249  *
1250  */
1251
1252 #if defined(OBJFORMAT_PEi386)
1253 /* A record for storing handles into DLLs. */
1254
1255 typedef
1256    struct _OpenedDLL {
1257       char*              name;
1258       struct _OpenedDLL* next;
1259       HINSTANCE instance;
1260    }
1261    OpenedDLL;
1262
1263 /* A list thereof. */
1264 static OpenedDLL* opened_dlls = NULL;
1265 #endif
1266
1267 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1268
1269 static const char *
1270 internal_dlopen(const char *dll_name)
1271 {
1272    void *hdl;
1273    const char *errmsg;
1274    char *errmsg_copy;
1275
1276    // omitted: RTLD_NOW
1277    // see http://www.haskell.org/pipermail/cvs-ghc/2007-September/038570.html
1278    IF_DEBUG(linker,
1279       debugBelch("internal_dlopen: dll_name = '%s'\n", dll_name));
1280
1281    //-------------- Begin critical section ------------------
1282    // This critical section is necessary because dlerror() is not
1283    // required to be reentrant (see POSIX -- IEEE Std 1003.1-2008)
1284    // Also, the error message returned must be copied to preserve it
1285    // (see POSIX also)
1286
1287    ACQUIRE_LOCK(&dl_mutex);
1288    hdl = dlopen(dll_name, RTLD_LAZY | RTLD_GLOBAL);
1289
1290    errmsg = NULL;
1291    if (hdl == NULL) {
1292       /* dlopen failed; return a ptr to the error msg. */
1293       errmsg = dlerror();
1294       if (errmsg == NULL) errmsg = "addDLL: unknown error";
1295       errmsg_copy = stgMallocBytes(strlen(errmsg)+1, "addDLL");
1296       strcpy(errmsg_copy, errmsg);
1297       errmsg = errmsg_copy;
1298    }
1299    RELEASE_LOCK(&dl_mutex);
1300    //--------------- End critical section -------------------
1301
1302    return errmsg;
1303 }
1304 #  endif
1305
1306 const char *
1307 addDLL( char *dll_name )
1308 {
1309 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1310    /* ------------------- ELF DLL loader ------------------- */
1311
1312 #define NMATCH 5
1313    regmatch_t match[NMATCH];
1314    const char *errmsg;
1315    FILE* fp;
1316    size_t match_length;
1317 #define MAXLINE 1000
1318    char line[MAXLINE];
1319    int result;
1320
1321    initLinker();
1322
1323    IF_DEBUG(linker, debugBelch("addDLL: dll_name = '%s'\n", dll_name));
1324    errmsg = internal_dlopen(dll_name);
1325
1326    if (errmsg == NULL) {
1327       return NULL;
1328    }
1329
1330    // GHC Trac ticket #2615
1331    // On some systems (e.g., Gentoo Linux) dynamic files (e.g. libc.so)
1332    // contain linker scripts rather than ELF-format object code. This
1333    // code handles the situation by recognizing the real object code
1334    // file name given in the linker script.
1335    //
1336    // If an "invalid ELF header" error occurs, it is assumed that the
1337    // .so file contains a linker script instead of ELF object code.
1338    // In this case, the code looks for the GROUP ( ... ) linker
1339    // directive. If one is found, the first file name inside the
1340    // parentheses is treated as the name of a dynamic library and the
1341    // code attempts to dlopen that file. If this is also unsuccessful,
1342    // an error message is returned.
1343
1344    // see if the error message is due to an invalid ELF header
1345    IF_DEBUG(linker, debugBelch("errmsg = '%s'\n", errmsg));
1346    result = regexec(&re_invalid, errmsg, (size_t) NMATCH, match, 0);
1347    IF_DEBUG(linker, debugBelch("result = %i\n", result));
1348    if (result == 0) {
1349       // success -- try to read the named file as a linker script
1350       match_length = (size_t) stg_min((match[1].rm_eo - match[1].rm_so),
1351                                  MAXLINE-1);
1352       strncpy(line, (errmsg+(match[1].rm_so)),match_length);
1353       line[match_length] = '\0'; // make sure string is null-terminated
1354       IF_DEBUG(linker, debugBelch ("file name = '%s'\n", line));
1355       if ((fp = fopen(line, "r")) == NULL) {
1356          return errmsg; // return original error if open fails
1357       }
1358       // try to find a GROUP ( ... ) command
1359       while (fgets(line, MAXLINE, fp) != NULL) {
1360          IF_DEBUG(linker, debugBelch("input line = %s", line));
1361          if (regexec(&re_realso, line, (size_t) NMATCH, match, 0) == 0) {
1362             // success -- try to dlopen the first named file
1363             IF_DEBUG(linker, debugBelch("match%s\n",""));
1364             line[match[2].rm_eo] = '\0';
1365             errmsg = internal_dlopen(line+match[2].rm_so);
1366             break;
1367          }
1368          // if control reaches here, no GROUP ( ... ) directive was found
1369          // and the original error message is returned to the caller
1370       }
1371       fclose(fp);
1372    }
1373    return errmsg;
1374
1375 #  elif defined(OBJFORMAT_PEi386)
1376    /* ------------------- Win32 DLL loader ------------------- */
1377
1378    char*      buf;
1379    OpenedDLL* o_dll;
1380    HINSTANCE  instance;
1381
1382    initLinker();
1383
1384    /* debugBelch("\naddDLL; dll_name = `%s'\n", dll_name); */
1385
1386    /* See if we've already got it, and ignore if so. */
1387    for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
1388       if (0 == strcmp(o_dll->name, dll_name))
1389          return NULL;
1390    }
1391
1392    /* The file name has no suffix (yet) so that we can try
1393       both foo.dll and foo.drv
1394
1395       The documentation for LoadLibrary says:
1396         If no file name extension is specified in the lpFileName
1397         parameter, the default library extension .dll is
1398         appended. However, the file name string can include a trailing
1399         point character (.) to indicate that the module name has no
1400         extension. */
1401
1402    buf = stgMallocBytes(strlen(dll_name) + 10, "addDLL");
1403    sprintf(buf, "%s.DLL", dll_name);
1404    instance = LoadLibrary(buf);
1405    if (instance == NULL) {
1406        if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1407        // KAA: allow loading of drivers (like winspool.drv)
1408        sprintf(buf, "%s.DRV", dll_name);
1409        instance = LoadLibrary(buf);
1410        if (instance == NULL) {
1411            if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1412            // #1883: allow loading of unix-style libfoo.dll DLLs
1413            sprintf(buf, "lib%s.DLL", dll_name);
1414            instance = LoadLibrary(buf);
1415            if (instance == NULL) {
1416                goto error;
1417            }
1418        }
1419    }
1420    stgFree(buf);
1421
1422    /* Add this DLL to the list of DLLs in which to search for symbols. */
1423    o_dll = stgMallocBytes( sizeof(OpenedDLL), "addDLL" );
1424    o_dll->name     = stgMallocBytes(1+strlen(dll_name), "addDLL");
1425    strcpy(o_dll->name, dll_name);
1426    o_dll->instance = instance;
1427    o_dll->next     = opened_dlls;
1428    opened_dlls     = o_dll;
1429
1430    return NULL;
1431
1432 error:
1433    stgFree(buf);
1434    sysErrorBelch(dll_name);
1435
1436    /* LoadLibrary failed; return a ptr to the error msg. */
1437    return "addDLL: could not load DLL";
1438
1439 #  else
1440    barf("addDLL: not implemented on this platform");
1441 #  endif
1442 }
1443
1444 /* -----------------------------------------------------------------------------
1445  * insert a stable symbol in the hash table
1446  */
1447
1448 void
1449 insertStableSymbol(char* obj_name, char* key, StgPtr p)
1450 {
1451   ghciInsertStrHashTable(obj_name, stablehash, key, getStablePtr(p));
1452 }
1453
1454
1455 /* -----------------------------------------------------------------------------
1456  * insert a symbol in the hash table
1457  */
1458 void
1459 insertSymbol(char* obj_name, char* key, void* data)
1460 {
1461   ghciInsertStrHashTable(obj_name, symhash, key, data);
1462 }
1463
1464 /* -----------------------------------------------------------------------------
1465  * lookup a symbol in the hash table
1466  */
1467 void *
1468 lookupSymbol( char *lbl )
1469 {
1470     void *val;
1471     IF_DEBUG(linker, debugBelch("lookupSymbol: looking up %s\n", lbl));
1472     initLinker() ;
1473     ASSERT(symhash != NULL);
1474     val = lookupStrHashTable(symhash, lbl);
1475
1476     if (val == NULL) {
1477         IF_DEBUG(linker, debugBelch("lookupSymbol: symbol not found\n"));
1478 #       if defined(OBJFORMAT_ELF)
1479         return dlsym(dl_prog_handle, lbl);
1480 #       elif defined(OBJFORMAT_MACHO)
1481 #       if HAVE_DLFCN_H
1482         /* On OS X 10.3 and later, we use dlsym instead of the old legacy
1483            interface.
1484
1485            HACK: On OS X, global symbols are prefixed with an underscore.
1486                  However, dlsym wants us to omit the leading underscore from the
1487                  symbol name. For now, we simply strip it off here (and ONLY
1488                  here).
1489         */
1490         IF_DEBUG(linker, debugBelch("lookupSymbol: looking up %s with dlsym\n", lbl));
1491         ASSERT(lbl[0] == '_');
1492         return dlsym(dl_prog_handle, lbl+1);
1493 #       else
1494         if(NSIsSymbolNameDefined(lbl)) {
1495             NSSymbol symbol = NSLookupAndBindSymbol(lbl);
1496             return NSAddressOfSymbol(symbol);
1497         } else {
1498             return NULL;
1499         }
1500 #       endif /* HAVE_DLFCN_H */
1501 #       elif defined(OBJFORMAT_PEi386)
1502         void* sym;
1503
1504         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1505         if (sym != NULL) { return sym; };
1506
1507         // Also try looking up the symbol without the @N suffix.  Some
1508         // DLLs have the suffixes on their symbols, some don't.
1509         zapTrailingAtSign ( (unsigned char*)lbl );
1510         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1511         if (sym != NULL) { return sym; };
1512         return NULL;
1513
1514 #       else
1515         ASSERT(2+2 == 5);
1516         return NULL;
1517 #       endif
1518     } else {
1519         IF_DEBUG(linker, debugBelch("lookupSymbol: value of %s is %p\n", lbl, val));
1520         return val;
1521     }
1522 }
1523
1524 /* -----------------------------------------------------------------------------
1525  * Debugging aid: look in GHCi's object symbol tables for symbols
1526  * within DELTA bytes of the specified address, and show their names.
1527  */
1528 #ifdef DEBUG
1529 void ghci_enquire ( char* addr );
1530
1531 void ghci_enquire ( char* addr )
1532 {
1533    int   i;
1534    char* sym;
1535    char* a;
1536    const int DELTA = 64;
1537    ObjectCode* oc;
1538
1539    initLinker();
1540
1541    for (oc = objects; oc; oc = oc->next) {
1542       for (i = 0; i < oc->n_symbols; i++) {
1543          sym = oc->symbols[i];
1544          if (sym == NULL) continue;
1545          a = NULL;
1546          if (a == NULL) {
1547             a = lookupStrHashTable(symhash, sym);
1548          }
1549          if (a == NULL) {
1550              // debugBelch("ghci_enquire: can't find %s\n", sym);
1551          }
1552          else if (addr-DELTA <= a && a <= addr+DELTA) {
1553             debugBelch("%p + %3d  ==  `%s'\n", addr, (int)(a - addr), sym);
1554          }
1555       }
1556    }
1557 }
1558 #endif
1559
1560 #ifdef USE_MMAP
1561 #define ROUND_UP(x,size) ((x + size - 1) & ~(size - 1))
1562
1563 static void *
1564 mmapForLinker (size_t bytes, nat flags, int fd)
1565 {
1566    void *map_addr = NULL;
1567    void *result;
1568    int pagesize, size;
1569    static nat fixed = 0;
1570
1571    IF_DEBUG(linker, debugBelch("mmapForLinker: start\n"));
1572    pagesize = getpagesize();
1573    size = ROUND_UP(bytes, pagesize);
1574
1575 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1576 mmap_again:
1577
1578    if (mmap_32bit_base != 0) {
1579        map_addr = mmap_32bit_base;
1580    }
1581 #endif
1582
1583    IF_DEBUG(linker, debugBelch("mmapForLinker: \tprotection %#0x\n", PROT_EXEC | PROT_READ | PROT_WRITE));
1584    IF_DEBUG(linker, debugBelch("mmapForLinker: \tflags      %#0x\n", MAP_PRIVATE | TRY_MAP_32BIT | fixed | flags));
1585    result = mmap(map_addr, size, PROT_EXEC|PROT_READ|PROT_WRITE,
1586                     MAP_PRIVATE|TRY_MAP_32BIT|fixed|flags, fd, 0);
1587
1588    if (result == MAP_FAILED) {
1589        sysErrorBelch("mmap %lu bytes at %p",(lnat)size,map_addr);
1590        errorBelch("Try specifying an address with +RTS -xm<addr> -RTS");
1591        stg_exit(EXIT_FAILURE);
1592    }
1593
1594 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1595    if (mmap_32bit_base != 0) {
1596        if (result == map_addr) {
1597            mmap_32bit_base = (StgWord8*)map_addr + size;
1598        } else {
1599            if ((W_)result > 0x80000000) {
1600                // oops, we were given memory over 2Gb
1601 #if defined(freebsd_HOST_OS)  || defined(kfreebsdgnu_HOST_OS) || defined(dragonfly_HOST_OS)
1602                // Some platforms require MAP_FIXED.  This is normally
1603                // a bad idea, because MAP_FIXED will overwrite
1604                // existing mappings.
1605                munmap(result,size);
1606                fixed = MAP_FIXED;
1607                goto mmap_again;
1608 #else
1609                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);
1610 #endif
1611            } else {
1612                // hmm, we were given memory somewhere else, but it's
1613                // still under 2Gb so we can use it.  Next time, ask
1614                // for memory right after the place we just got some
1615                mmap_32bit_base = (StgWord8*)result + size;
1616            }
1617        }
1618    } else {
1619        if ((W_)result > 0x80000000) {
1620            // oops, we were given memory over 2Gb
1621            // ... try allocating memory somewhere else?;
1622            debugTrace(DEBUG_linker,"MAP_32BIT didn't work; gave us %lu bytes at 0x%p", bytes, result);
1623            munmap(result, size);
1624
1625            // Set a base address and try again... (guess: 1Gb)
1626            mmap_32bit_base = (void*)0x40000000;
1627            goto mmap_again;
1628        }
1629    }
1630 #endif
1631
1632    IF_DEBUG(linker, debugBelch("mmapForLinker: mapped %lu bytes starting at %p\n", (lnat)size, result));
1633    IF_DEBUG(linker, debugBelch("mmapForLinker: done\n"));
1634    return result;
1635 }
1636 #endif // USE_MMAP
1637
1638 static ObjectCode*
1639 mkOc( char *path, char *image, int imageSize,
1640       char *archiveMemberName
1641 #ifndef USE_MMAP
1642 #ifdef darwin_HOST_OS
1643     , int misalignment
1644 #endif
1645 #endif
1646     ) {
1647    ObjectCode* oc;
1648
1649    IF_DEBUG(linker, debugBelch("mkOc: start\n"));
1650    oc = stgMallocBytes(sizeof(ObjectCode), "loadArchive(oc)");
1651
1652 #  if defined(OBJFORMAT_ELF)
1653    oc->formatName = "ELF";
1654 #  elif defined(OBJFORMAT_PEi386)
1655    oc->formatName = "PEi386";
1656 #  elif defined(OBJFORMAT_MACHO)
1657    oc->formatName = "Mach-O";
1658 #  else
1659    stgFree(oc);
1660    barf("loadObj: not implemented on this platform");
1661 #  endif
1662
1663    oc->image = image;
1664    /* sigh, strdup() isn't a POSIX function, so do it the long way */
1665    oc->fileName = stgMallocBytes( strlen(path)+1, "loadObj" );
1666    strcpy(oc->fileName, path);
1667
1668    if (archiveMemberName) {
1669        oc->archiveMemberName = stgMallocBytes( strlen(archiveMemberName)+1, "loadObj" );
1670        strcpy(oc->archiveMemberName, archiveMemberName);
1671    }
1672    else {
1673        oc->archiveMemberName = NULL;
1674    }
1675
1676    oc->fileSize          = imageSize;
1677    oc->symbols           = NULL;
1678    oc->sections          = NULL;
1679    oc->proddables        = NULL;
1680
1681 #ifndef USE_MMAP
1682 #ifdef darwin_HOST_OS
1683    oc->misalignment = misalignment;
1684 #endif
1685 #endif
1686
1687    /* chain it onto the list of objects */
1688    oc->next              = objects;
1689    objects               = oc;
1690
1691    IF_DEBUG(linker, debugBelch("mkOc: done\n"));
1692    return oc;
1693 }
1694
1695 HsInt
1696 loadArchive( char *path )
1697 {
1698     ObjectCode* oc;
1699     char *image;
1700     int memberSize;
1701     FILE *f;
1702     int n;
1703     size_t thisFileNameSize;
1704     char *fileName;
1705     size_t fileNameSize;
1706     int isObject, isGnuIndex;
1707     char tmp[20];
1708     char *gnuFileIndex;
1709     int gnuFileIndexSize;
1710 #if defined(darwin_HOST_OS)
1711     int i;
1712     uint32_t nfat_arch, nfat_offset, cputype, cpusubtype;
1713 #if defined(i386_HOST_ARCH)
1714     const uint32_t mycputype = CPU_TYPE_X86;
1715     const uint32_t mycpusubtype = CPU_SUBTYPE_X86_ALL;
1716 #elif defined(x86_64_HOST_ARCH)
1717     const uint32_t mycputype = CPU_TYPE_X86_64;
1718     const uint32_t mycpusubtype = CPU_SUBTYPE_X86_64_ALL;
1719 #elif defined(powerpc_HOST_ARCH)
1720     const uint32_t mycputype = CPU_TYPE_POWERPC;
1721     const uint32_t mycpusubtype = CPU_SUBTYPE_POWERPC_ALL;
1722 #elif defined(powerpc64_HOST_ARCH)
1723     const uint32_t mycputype = CPU_TYPE_POWERPC64;
1724     const uint32_t mycpusubtype = CPU_SUBTYPE_POWERPC_ALL;
1725 #else
1726 #error Unknown Darwin architecture
1727 #endif
1728 #if !defined(USE_MMAP)
1729     int misalignment;
1730 #endif
1731 #endif
1732
1733     IF_DEBUG(linker, debugBelch("loadArchive: start\n"));
1734     IF_DEBUG(linker, debugBelch("loadArchive: Loading archive `%s'\n", path));
1735
1736     gnuFileIndex = NULL;
1737     gnuFileIndexSize = 0;
1738
1739     fileNameSize = 32;
1740     fileName = stgMallocBytes(fileNameSize, "loadArchive(fileName)");
1741
1742     f = fopen(path, "rb");
1743     if (!f)
1744         barf("loadObj: can't read `%s'", path);
1745
1746     /* Check if this is an archive by looking for the magic "!<arch>\n"
1747      * string.  Usually, if this fails, we barf and quit.  On Darwin however,
1748      * we may have a fat archive, which contains archives for more than
1749      * one architecture.  Fat archives start with the magic number 0xcafebabe,
1750      * always stored big endian.  If we find a fat_header, we scan through
1751      * the fat_arch structs, searching through for one for our host
1752      * architecture.  If a matching struct is found, we read the offset
1753      * of our archive data (nfat_offset) and seek forward nfat_offset bytes
1754      * from the start of the file.
1755      *
1756      * A subtlety is that all of the members of the fat_header and fat_arch
1757      * structs are stored big endian, so we need to call byte order
1758      * conversion functions.
1759      *
1760      * If we find the appropriate architecture in a fat archive, we gobble
1761      * its magic "!<arch>\n" string and continue processing just as if
1762      * we had a single architecture archive.
1763      */
1764
1765     n = fread ( tmp, 1, 8, f );
1766     if (n != 8)
1767         barf("loadArchive: Failed reading header from `%s'", path);
1768     if (strncmp(tmp, "!<arch>\n", 8) != 0) {
1769
1770 #if defined(darwin_HOST_OS)
1771         /* Not a standard archive, look for a fat archive magic number: */
1772         if (ntohl(*(uint32_t *)tmp) == FAT_MAGIC) {
1773             nfat_arch = ntohl(*(uint32_t *)(tmp + 4));
1774             IF_DEBUG(linker, debugBelch("loadArchive: found a fat archive containing %d architectures\n", nfat_arch));
1775             nfat_offset = 0;
1776
1777             for (i = 0; i < (int)nfat_arch; i++) {
1778                 /* search for the right arch */
1779                 n = fread( tmp, 1, 20, f );
1780                 if (n != 8)
1781                     barf("loadArchive: Failed reading arch from `%s'", path);
1782                 cputype = ntohl(*(uint32_t *)tmp);
1783                 cpusubtype = ntohl(*(uint32_t *)(tmp + 4));
1784
1785                 if (cputype == mycputype && cpusubtype == mycpusubtype) {
1786                     IF_DEBUG(linker, debugBelch("loadArchive: found my archive in a fat archive\n"));
1787                     nfat_offset = ntohl(*(uint32_t *)(tmp + 8));
1788                     break;
1789                 }
1790             }
1791
1792             if (nfat_offset == 0) {
1793                barf ("loadArchive: searched %d architectures, but no host arch found", (int)nfat_arch);
1794             }
1795             else {
1796                 n = fseek( f, nfat_offset, SEEK_SET );
1797                 if (n != 0)
1798                     barf("loadArchive: Failed to seek to arch in `%s'", path);
1799                 n = fread ( tmp, 1, 8, f );
1800                 if (n != 8)
1801                     barf("loadArchive: Failed reading header from `%s'", path);
1802                 if (strncmp(tmp, "!<arch>\n", 8) != 0) {
1803                     barf("loadArchive: couldn't find archive in `%s' at offset %d", path, nfat_offset);
1804                 }
1805             }
1806         }
1807         else {
1808             barf("loadArchive: Neither an archive, nor a fat archive: `%s'", path);
1809         }
1810
1811 #else
1812         barf("loadArchive: Not an archive: `%s'", path);
1813 #endif
1814     }
1815
1816     IF_DEBUG(linker, debugBelch("loadArchive: loading archive contents\n"));
1817
1818     while(1) {
1819         n = fread ( fileName, 1, 16, f );
1820         if (n != 16) {
1821             if (feof(f)) {
1822                 IF_DEBUG(linker, debugBelch("loadArchive: EOF while reading from '%s'\n", path));
1823                 break;
1824             }
1825             else {
1826                 barf("loadArchive: Failed reading file name from `%s'", path);
1827             }
1828         }
1829
1830 #if defined(darwin_HOST_OS)
1831         if (strncmp(fileName, "!<arch>\n", 8) == 0) {
1832             IF_DEBUG(linker, debugBelch("loadArchive: found the start of another archive, breaking\n"));
1833             break;
1834         }
1835 #endif
1836
1837         n = fread ( tmp, 1, 12, f );
1838         if (n != 12)
1839             barf("loadArchive: Failed reading mod time from `%s'", path);
1840         n = fread ( tmp, 1, 6, f );
1841         if (n != 6)
1842             barf("loadArchive: Failed reading owner from `%s'", path);
1843         n = fread ( tmp, 1, 6, f );
1844         if (n != 6)
1845             barf("loadArchive: Failed reading group from `%s'", path);
1846         n = fread ( tmp, 1, 8, f );
1847         if (n != 8)
1848             barf("loadArchive: Failed reading mode from `%s'", path);
1849         n = fread ( tmp, 1, 10, f );
1850         if (n != 10)
1851             barf("loadArchive: Failed reading size from `%s'", path);
1852         tmp[10] = '\0';
1853         for (n = 0; isdigit(tmp[n]); n++);
1854         tmp[n] = '\0';
1855         memberSize = atoi(tmp);
1856
1857         IF_DEBUG(linker, debugBelch("loadArchive: size of this archive member is %d\n", memberSize));
1858         n = fread ( tmp, 1, 2, f );
1859         if (n != 2)
1860             barf("loadArchive: Failed reading magic from `%s'", path);
1861         if (strncmp(tmp, "\x60\x0A", 2) != 0)
1862             barf("loadArchive: Failed reading magic from `%s' at %ld. Got %c%c",
1863                  path, ftell(f), tmp[0], tmp[1]);
1864
1865         isGnuIndex = 0;
1866         /* Check for BSD-variant large filenames */
1867         if (0 == strncmp(fileName, "#1/", 3)) {
1868             fileName[16] = '\0';
1869             if (isdigit(fileName[3])) {
1870                 for (n = 4; isdigit(fileName[n]); n++);
1871                 fileName[n] = '\0';
1872                 thisFileNameSize = atoi(fileName + 3);
1873                 memberSize -= thisFileNameSize;
1874                 if (thisFileNameSize >= fileNameSize) {
1875                     /* Double it to avoid potentially continually
1876                        increasing it by 1 */
1877                     fileNameSize = thisFileNameSize * 2;
1878                     fileName = stgReallocBytes(fileName, fileNameSize, "loadArchive(fileName)");
1879                 }
1880                 n = fread ( fileName, 1, thisFileNameSize, f );
1881                 if (n != (int)thisFileNameSize) {
1882                     barf("loadArchive: Failed reading filename from `%s'",
1883                          path);
1884                 }
1885                 fileName[thisFileNameSize] = 0;
1886
1887                 /* On OS X at least, thisFileNameSize is the size of the
1888                    fileName field, not the length of the fileName
1889                    itself. */
1890                 thisFileNameSize = strlen(fileName);
1891             }
1892             else {
1893                 barf("loadArchive: BSD-variant filename size not found while reading filename from `%s'", path);
1894             }
1895         }
1896         /* Check for GNU file index file */
1897         else if (0 == strncmp(fileName, "//", 2)) {
1898             fileName[0] = '\0';
1899             thisFileNameSize = 0;
1900             isGnuIndex = 1;
1901         }
1902         /* Check for a file in the GNU file index */
1903         else if (fileName[0] == '/') {
1904             if (isdigit(fileName[1])) {
1905                 int i;
1906
1907                 for (n = 2; isdigit(fileName[n]); n++);
1908                 fileName[n] = '\0';
1909                 n = atoi(fileName + 1);
1910
1911                 if (gnuFileIndex == NULL) {
1912                     barf("loadArchive: GNU-variant filename without an index while reading from `%s'", path);
1913                 }
1914                 if (n < 0 || n > gnuFileIndexSize) {
1915                     barf("loadArchive: GNU-variant filename offset %d out of range [0..%d] while reading filename from `%s'", n, gnuFileIndexSize, path);
1916                 }
1917                 if (n != 0 && gnuFileIndex[n - 1] != '\n') {
1918                     barf("loadArchive: GNU-variant filename offset %d invalid (range [0..%d]) while reading filename from `%s'", n, gnuFileIndexSize, path);
1919                 }
1920                 for (i = n; gnuFileIndex[i] != '/'; i++);
1921                 thisFileNameSize = i - n;
1922                 if (thisFileNameSize >= fileNameSize) {
1923                     /* Double it to avoid potentially continually
1924                        increasing it by 1 */
1925                     fileNameSize = thisFileNameSize * 2;
1926                     fileName = stgReallocBytes(fileName, fileNameSize, "loadArchive(fileName)");
1927                 }
1928                 memcpy(fileName, gnuFileIndex + n, thisFileNameSize);
1929                 fileName[thisFileNameSize] = '\0';
1930             }
1931             else if (fileName[1] == ' ') {
1932                 fileName[0] = '\0';
1933                 thisFileNameSize = 0;
1934             }
1935             else {
1936                 barf("loadArchive: GNU-variant filename offset not found while reading filename from `%s'", path);
1937             }
1938         }
1939         /* Finally, the case where the filename field actually contains
1940            the filename */
1941         else {
1942             /* GNU ar terminates filenames with a '/', this allowing
1943                spaces in filenames. So first look to see if there is a
1944                terminating '/'. */
1945             for (thisFileNameSize = 0;
1946                  thisFileNameSize < 16;
1947                  thisFileNameSize++) {
1948                 if (fileName[thisFileNameSize] == '/') {
1949                     fileName[thisFileNameSize] = '\0';
1950                     break;
1951                 }
1952             }
1953             /* If we didn't find a '/', then a space teminates the
1954                filename. Note that if we don't find one, then
1955                thisFileNameSize ends up as 16, and we already have the
1956                '\0' at the end. */
1957             if (thisFileNameSize == 16) {
1958                 for (thisFileNameSize = 0;
1959                      thisFileNameSize < 16;
1960                      thisFileNameSize++) {
1961                     if (fileName[thisFileNameSize] == ' ') {
1962                         fileName[thisFileNameSize] = '\0';
1963                         break;
1964                     }
1965                 }
1966             }
1967         }
1968
1969         IF_DEBUG(linker,
1970                  debugBelch("loadArchive: Found member file `%s'\n", fileName));
1971
1972         isObject = thisFileNameSize >= 2
1973                 && fileName[thisFileNameSize - 2] == '.'
1974                 && fileName[thisFileNameSize - 1] == 'o';
1975
1976         IF_DEBUG(linker, debugBelch("loadArchive: \tthisFileNameSize = %d\n", (int)thisFileNameSize));
1977         IF_DEBUG(linker, debugBelch("loadArchive: \tisObject = %d\n", isObject));
1978
1979         if (isObject) {
1980             char *archiveMemberName;
1981
1982             IF_DEBUG(linker, debugBelch("loadArchive: Member is an object file...loading...\n"));
1983
1984             /* We can't mmap from the archive directly, as object
1985                files need to be 8-byte aligned but files in .ar
1986                archives are 2-byte aligned. When possible we use mmap
1987                to get some anonymous memory, as on 64-bit platforms if
1988                we use malloc then we can be given memory above 2^32.
1989                In the mmap case we're probably wasting lots of space;
1990                we could do better. */
1991 #if defined(USE_MMAP)
1992             image = mmapForLinker(memberSize, MAP_ANONYMOUS, -1);
1993 #elif defined(darwin_HOST_OS)
1994             /* See loadObj() */
1995             misalignment = machoGetMisalignment(f);
1996             image = stgMallocBytes(memberSize + misalignment, "loadArchive(image)");
1997             image += misalignment;
1998 #else
1999             image = stgMallocBytes(memberSize, "loadArchive(image)");
2000 #endif
2001             n = fread ( image, 1, memberSize, f );
2002             if (n != memberSize) {
2003                 barf("loadArchive: error whilst reading `%s'", path);
2004             }
2005
2006             archiveMemberName = stgMallocBytes(strlen(path) + thisFileNameSize + 3,
2007                                                "loadArchive(file)");
2008             sprintf(archiveMemberName, "%s(%.*s)",
2009                     path, (int)thisFileNameSize, fileName);
2010
2011             oc = mkOc(path, image, memberSize, archiveMemberName
2012 #ifndef USE_MMAP
2013 #ifdef darwin_HOST_OS
2014                      , misalignment
2015 #endif
2016 #endif
2017                      );
2018
2019             stgFree(archiveMemberName);
2020
2021             if (0 == loadOc(oc)) {
2022                 stgFree(fileName);
2023                 return 0;
2024             }
2025         }
2026         else if (isGnuIndex) {
2027             if (gnuFileIndex != NULL) {
2028                 barf("loadArchive: GNU-variant index found, but already have an index, while reading filename from `%s'", path);
2029             }
2030             IF_DEBUG(linker, debugBelch("loadArchive: Found GNU-variant file index\n"));
2031 #ifdef USE_MMAP
2032             gnuFileIndex = mmapForLinker(memberSize + 1, MAP_ANONYMOUS, -1);
2033 #else
2034             gnuFileIndex = stgMallocBytes(memberSize + 1, "loadArchive(image)");
2035 #endif
2036             n = fread ( gnuFileIndex, 1, memberSize, f );
2037             if (n != memberSize) {
2038                 barf("loadArchive: error whilst reading `%s'", path);
2039             }
2040             gnuFileIndex[memberSize] = '/';
2041             gnuFileIndexSize = memberSize;
2042         }
2043         else {
2044             IF_DEBUG(linker, debugBelch("loadArchive: '%s' does not appear to be an object file\n", fileName));
2045             n = fseek(f, memberSize, SEEK_CUR);
2046             if (n != 0)
2047                 barf("loadArchive: error whilst seeking by %d in `%s'",
2048                      memberSize, path);
2049         }
2050
2051         /* .ar files are 2-byte aligned */
2052         if (memberSize % 2) {
2053             IF_DEBUG(linker, debugBelch("loadArchive: trying to read one pad byte\n"));
2054             n = fread ( tmp, 1, 1, f );
2055             if (n != 1) {
2056                 if (feof(f)) {
2057                     IF_DEBUG(linker, debugBelch("loadArchive: found EOF while reading one pad byte\n"));
2058                     break;
2059                 }
2060                 else {
2061                     barf("loadArchive: Failed reading padding from `%s'", path);
2062                 }
2063             }
2064             IF_DEBUG(linker, debugBelch("loadArchive: successfully read one pad byte\n"));
2065         }
2066         IF_DEBUG(linker, debugBelch("loadArchive: reached end of archive loading while loop\n"));
2067     }
2068
2069     fclose(f);
2070
2071     stgFree(fileName);
2072     if (gnuFileIndex != NULL) {
2073 #ifdef USE_MMAP
2074         munmap(gnuFileIndex, gnuFileIndexSize + 1);
2075 #else
2076         stgFree(gnuFileIndex);
2077 #endif
2078     }
2079
2080     IF_DEBUG(linker, debugBelch("loadArchive: done\n"));
2081     return 1;
2082 }
2083
2084 /* -----------------------------------------------------------------------------
2085  * Load an obj (populate the global symbol table, but don't resolve yet)
2086  *
2087  * Returns: 1 if ok, 0 on error.
2088  */
2089 HsInt
2090 loadObj( char *path )
2091 {
2092    ObjectCode* oc;
2093    char *image;
2094    int fileSize;
2095    struct stat st;
2096    int r;
2097 #ifdef USE_MMAP
2098    int fd;
2099 #else
2100    FILE *f;
2101 #  if defined(darwin_HOST_OS)
2102    int misalignment;
2103 #  endif
2104 #endif
2105    IF_DEBUG(linker, debugBelch("loadObj %s\n", path));
2106
2107    initLinker();
2108
2109    /* debugBelch("loadObj %s\n", path ); */
2110
2111    /* Check that we haven't already loaded this object.
2112       Ignore requests to load multiple times */
2113    {
2114        ObjectCode *o;
2115        int is_dup = 0;
2116        for (o = objects; o; o = o->next) {
2117           if (0 == strcmp(o->fileName, path)) {
2118              is_dup = 1;
2119              break; /* don't need to search further */
2120           }
2121        }
2122        if (is_dup) {
2123           IF_DEBUG(linker, debugBelch(
2124             "GHCi runtime linker: warning: looks like you're trying to load the\n"
2125             "same object file twice:\n"
2126             "   %s\n"
2127             "GHCi will ignore this, but be warned.\n"
2128             , path));
2129           return 1; /* success */
2130        }
2131    }
2132
2133    r = stat(path, &st);
2134    if (r == -1) {
2135        IF_DEBUG(linker, debugBelch("File doesn't exist\n"));
2136        return 0;
2137    }
2138
2139    fileSize = st.st_size;
2140
2141 #ifdef USE_MMAP
2142    /* On many architectures malloc'd memory isn't executable, so we need to use mmap. */
2143
2144 #if defined(openbsd_HOST_OS)
2145    fd = open(path, O_RDONLY, S_IRUSR);
2146 #else
2147    fd = open(path, O_RDONLY);
2148 #endif
2149    if (fd == -1)
2150       barf("loadObj: can't open `%s'", path);
2151
2152    image = mmapForLinker(fileSize, 0, fd);
2153
2154    close(fd);
2155
2156 #else /* !USE_MMAP */
2157    /* load the image into memory */
2158    f = fopen(path, "rb");
2159    if (!f)
2160        barf("loadObj: can't read `%s'", path);
2161
2162 #   if defined(mingw32_HOST_OS)
2163         // TODO: We would like to use allocateExec here, but allocateExec
2164         //       cannot currently allocate blocks large enough.
2165     image = VirtualAlloc(NULL, fileSize, MEM_RESERVE | MEM_COMMIT,
2166                              PAGE_EXECUTE_READWRITE);
2167 #   elif defined(darwin_HOST_OS)
2168     // In a Mach-O .o file, all sections can and will be misaligned
2169     // if the total size of the headers is not a multiple of the
2170     // desired alignment. This is fine for .o files that only serve
2171     // as input for the static linker, but it's not fine for us,
2172     // as SSE (used by gcc for floating point) and Altivec require
2173     // 16-byte alignment.
2174     // We calculate the correct alignment from the header before
2175     // reading the file, and then we misalign image on purpose so
2176     // that the actual sections end up aligned again.
2177    misalignment = machoGetMisalignment(f);
2178    image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
2179    image += misalignment;
2180 #  else
2181    image = stgMallocBytes(fileSize, "loadObj(image)");
2182 #  endif
2183
2184    {
2185        int n;
2186        n = fread ( image, 1, fileSize, f );
2187        if (n != fileSize)
2188            barf("loadObj: error whilst reading `%s'", path);
2189    }
2190    fclose(f);
2191 #endif /* USE_MMAP */
2192
2193    oc = mkOc(path, image, fileSize, NULL
2194 #ifndef USE_MMAP
2195 #ifdef darwin_HOST_OS
2196             , misalignment
2197 #endif
2198 #endif
2199             );
2200
2201    return loadOc(oc);
2202 }
2203
2204 static HsInt
2205 loadOc( ObjectCode* oc ) {
2206    int r;
2207
2208    IF_DEBUG(linker, debugBelch("loadOc: start\n"));
2209
2210 #  if defined(OBJFORMAT_MACHO) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
2211    r = ocAllocateSymbolExtras_MachO ( oc );
2212    if (!r) {
2213        IF_DEBUG(linker, debugBelch("loadOc: ocAllocateSymbolExtras_MachO failed\n"));
2214        return r;
2215    }
2216 #  elif defined(OBJFORMAT_ELF) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
2217    r = ocAllocateSymbolExtras_ELF ( oc );
2218    if (!r) {
2219        IF_DEBUG(linker, debugBelch("loadOc: ocAllocateSymbolExtras_ELF failed\n"));
2220        return r;
2221    }
2222 #endif
2223
2224    /* verify the in-memory image */
2225 #  if defined(OBJFORMAT_ELF)
2226    r = ocVerifyImage_ELF ( oc );
2227 #  elif defined(OBJFORMAT_PEi386)
2228    r = ocVerifyImage_PEi386 ( oc );
2229 #  elif defined(OBJFORMAT_MACHO)
2230    r = ocVerifyImage_MachO ( oc );
2231 #  else
2232    barf("loadObj: no verify method");
2233 #  endif
2234    if (!r) {
2235        IF_DEBUG(linker, debugBelch("loadOc: ocVerifyImage_* failed\n"));
2236        return r;
2237    }
2238
2239    /* build the symbol list for this image */
2240 #  if defined(OBJFORMAT_ELF)
2241    r = ocGetNames_ELF ( oc );
2242 #  elif defined(OBJFORMAT_PEi386)
2243    r = ocGetNames_PEi386 ( oc );
2244 #  elif defined(OBJFORMAT_MACHO)
2245    r = ocGetNames_MachO ( oc );
2246 #  else
2247    barf("loadObj: no getNames method");
2248 #  endif
2249    if (!r) {
2250        IF_DEBUG(linker, debugBelch("loadOc: ocGetNames_* failed\n"));
2251        return r;
2252    }
2253
2254    /* loaded, but not resolved yet */
2255    oc->status = OBJECT_LOADED;
2256    IF_DEBUG(linker, debugBelch("loadOc: done.\n"));
2257
2258    return 1;
2259 }
2260
2261 /* -----------------------------------------------------------------------------
2262  * resolve all the currently unlinked objects in memory
2263  *
2264  * Returns: 1 if ok, 0 on error.
2265  */
2266 HsInt
2267 resolveObjs( void )
2268 {
2269     ObjectCode *oc;
2270     int r;
2271
2272     IF_DEBUG(linker, debugBelch("resolveObjs: start\n"));
2273     initLinker();
2274
2275     for (oc = objects; oc; oc = oc->next) {
2276         if (oc->status != OBJECT_RESOLVED) {
2277 #           if defined(OBJFORMAT_ELF)
2278             r = ocResolve_ELF ( oc );
2279 #           elif defined(OBJFORMAT_PEi386)
2280             r = ocResolve_PEi386 ( oc );
2281 #           elif defined(OBJFORMAT_MACHO)
2282             r = ocResolve_MachO ( oc );
2283 #           else
2284             barf("resolveObjs: not implemented on this platform");
2285 #           endif
2286             if (!r) { return r; }
2287             oc->status = OBJECT_RESOLVED;
2288         }
2289     }
2290     IF_DEBUG(linker, debugBelch("resolveObjs: done\n"));
2291     return 1;
2292 }
2293
2294 /* -----------------------------------------------------------------------------
2295  * delete an object from the pool
2296  */
2297 HsInt
2298 unloadObj( char *path )
2299 {
2300     ObjectCode *oc, *prev;
2301     HsBool unloadedAnyObj = HS_BOOL_FALSE;
2302
2303     ASSERT(symhash != NULL);
2304     ASSERT(objects != NULL);
2305
2306     initLinker();
2307
2308     prev = NULL;
2309     for (oc = objects; oc; prev = oc, oc = oc->next) {
2310         if (!strcmp(oc->fileName,path)) {
2311
2312             /* Remove all the mappings for the symbols within this
2313              * object..
2314              */
2315             {
2316                 int i;
2317                 for (i = 0; i < oc->n_symbols; i++) {
2318                    if (oc->symbols[i] != NULL) {
2319                        removeStrHashTable(symhash, oc->symbols[i], NULL);
2320                    }
2321                 }
2322             }
2323
2324             if (prev == NULL) {
2325                 objects = oc->next;
2326             } else {
2327                 prev->next = oc->next;
2328             }
2329
2330             // We're going to leave this in place, in case there are
2331             // any pointers from the heap into it:
2332                 // #ifdef mingw32_HOST_OS
2333                 //  VirtualFree(oc->image);
2334                 // #else
2335             //  stgFree(oc->image);
2336             // #endif
2337             stgFree(oc->fileName);
2338             stgFree(oc->symbols);
2339             stgFree(oc->sections);
2340             stgFree(oc);
2341
2342             /* This could be a member of an archive so continue
2343              * unloading other members. */
2344             unloadedAnyObj = HS_BOOL_TRUE;
2345         }
2346     }
2347
2348     if (unloadedAnyObj) {
2349         return 1;
2350     }
2351     else {
2352         errorBelch("unloadObj: can't find `%s' to unload", path);
2353         return 0;
2354     }
2355 }
2356
2357 /* -----------------------------------------------------------------------------
2358  * Sanity checking.  For each ObjectCode, maintain a list of address ranges
2359  * which may be prodded during relocation, and abort if we try and write
2360  * outside any of these.
2361  */
2362 static void
2363 addProddableBlock ( ObjectCode* oc, void* start, int size )
2364 {
2365    ProddableBlock* pb
2366       = stgMallocBytes(sizeof(ProddableBlock), "addProddableBlock");
2367
2368    IF_DEBUG(linker, debugBelch("addProddableBlock: %p %p %d\n", oc, start, size));
2369    ASSERT(size > 0);
2370    pb->start      = start;
2371    pb->size       = size;
2372    pb->next       = oc->proddables;
2373    oc->proddables = pb;
2374 }
2375
2376 static void
2377 checkProddableBlock (ObjectCode *oc, void *addr )
2378 {
2379    ProddableBlock* pb;
2380
2381    for (pb = oc->proddables; pb != NULL; pb = pb->next) {
2382       char* s = (char*)(pb->start);
2383       char* e = s + pb->size - 1;
2384       char* a = (char*)addr;
2385       /* Assumes that the biggest fixup involves a 4-byte write.  This
2386          probably needs to be changed to 8 (ie, +7) on 64-bit
2387          plats. */
2388       if (a >= s && (a+3) <= e) return;
2389    }
2390    barf("checkProddableBlock: invalid fixup in runtime linker");
2391 }
2392
2393 /* -----------------------------------------------------------------------------
2394  * Section management.
2395  */
2396 static void
2397 addSection ( ObjectCode* oc, SectionKind kind,
2398                          void* start, void* end )
2399 {
2400    Section* s   = stgMallocBytes(sizeof(Section), "addSection");
2401    s->start     = start;
2402    s->end       = end;
2403    s->kind      = kind;
2404    s->next      = oc->sections;
2405    oc->sections = s;
2406
2407    IF_DEBUG(linker, debugBelch("addSection: %p-%p (size %ld), kind %d\n",
2408                                start, ((char*)end)-1, (long)end - (long)start + 1, kind ));
2409 }
2410
2411
2412 /* --------------------------------------------------------------------------
2413  * Symbol Extras.
2414  * This is about allocating a small chunk of memory for every symbol in the
2415  * object file. We make sure that the SymboLExtras are always "in range" of
2416  * limited-range PC-relative instructions on various platforms by allocating
2417  * them right next to the object code itself.
2418  */
2419
2420 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
2421
2422 /*
2423   ocAllocateSymbolExtras
2424
2425   Allocate additional space at the end of the object file image to make room
2426   for jump islands (powerpc, x86_64) and GOT entries (x86_64).
2427
2428   PowerPC relative branch instructions have a 24 bit displacement field.
2429   As PPC code is always 4-byte-aligned, this yields a +-32MB range.
2430   If a particular imported symbol is outside this range, we have to redirect
2431   the jump to a short piece of new code that just loads the 32bit absolute
2432   address and jumps there.
2433   On x86_64, PC-relative jumps and PC-relative accesses to the GOT are limited
2434   to 32 bits (+-2GB).
2435
2436   This function just allocates space for one SymbolExtra for every
2437   undefined symbol in the object file. The code for the jump islands is
2438   filled in by makeSymbolExtra below.
2439 */
2440
2441 static int ocAllocateSymbolExtras( ObjectCode* oc, int count, int first )
2442 {
2443 #ifdef USE_MMAP
2444   int pagesize, n, m;
2445 #endif
2446   int aligned;
2447 #ifndef USE_MMAP
2448   int misalignment = 0;
2449 #ifdef darwin_HOST_OS
2450   misalignment = oc->misalignment;
2451 #endif
2452 #endif
2453
2454   if( count > 0 )
2455   {
2456     // round up to the nearest 4
2457     aligned = (oc->fileSize + 3) & ~3;
2458
2459 #ifdef USE_MMAP
2460     pagesize = getpagesize();
2461     n = ROUND_UP( oc->fileSize, pagesize );
2462     m = ROUND_UP( aligned + sizeof (SymbolExtra) * count, pagesize );
2463
2464     /* we try to use spare space at the end of the last page of the
2465      * image for the jump islands, but if there isn't enough space
2466      * then we have to map some (anonymously, remembering MAP_32BIT).
2467      */
2468     if( m > n ) // we need to allocate more pages
2469     {
2470         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
2471                                           MAP_ANONYMOUS, -1);
2472     }
2473     else
2474     {
2475         oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
2476     }
2477 #else
2478     oc->image -= misalignment;
2479     oc->image = stgReallocBytes( oc->image,
2480                                  misalignment +
2481                                  aligned + sizeof (SymbolExtra) * count,
2482                                  "ocAllocateSymbolExtras" );
2483     oc->image += misalignment;
2484
2485     oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
2486 #endif /* USE_MMAP */
2487
2488     memset( oc->symbol_extras, 0, sizeof (SymbolExtra) * count );
2489   }
2490   else
2491     oc->symbol_extras = NULL;
2492
2493   oc->first_symbol_extra = first;
2494   oc->n_symbol_extras = count;
2495
2496   return 1;
2497 }
2498
2499 static SymbolExtra* makeSymbolExtra( ObjectCode* oc,
2500                                      unsigned long symbolNumber,
2501                                      unsigned long target )
2502 {
2503   SymbolExtra *extra;
2504
2505   ASSERT( symbolNumber >= oc->first_symbol_extra
2506         && symbolNumber - oc->first_symbol_extra < oc->n_symbol_extras);
2507
2508   extra = &oc->symbol_extras[symbolNumber - oc->first_symbol_extra];
2509
2510 #ifdef powerpc_HOST_ARCH
2511   // lis r12, hi16(target)
2512   extra->jumpIsland.lis_r12     = 0x3d80;
2513   extra->jumpIsland.hi_addr     = target >> 16;
2514
2515   // ori r12, r12, lo16(target)
2516   extra->jumpIsland.ori_r12_r12 = 0x618c;
2517   extra->jumpIsland.lo_addr     = target & 0xffff;
2518
2519   // mtctr r12
2520   extra->jumpIsland.mtctr_r12   = 0x7d8903a6;
2521
2522   // bctr
2523   extra->jumpIsland.bctr        = 0x4e800420;
2524 #endif
2525 #ifdef x86_64_HOST_ARCH
2526         // jmp *-14(%rip)
2527   static uint8_t jmp[] = { 0xFF, 0x25, 0xF2, 0xFF, 0xFF, 0xFF };
2528   extra->addr = target;
2529   memcpy(extra->jumpIsland, jmp, 6);
2530 #endif
2531
2532   return extra;
2533 }
2534
2535 #endif
2536
2537 /* --------------------------------------------------------------------------
2538  * PowerPC specifics (instruction cache flushing)
2539  * ------------------------------------------------------------------------*/
2540
2541 #ifdef powerpc_HOST_ARCH
2542 /*
2543    ocFlushInstructionCache
2544
2545    Flush the data & instruction caches.
2546    Because the PPC has split data/instruction caches, we have to
2547    do that whenever we modify code at runtime.
2548  */
2549
2550 static void
2551 ocFlushInstructionCacheFrom(void* begin, size_t length)
2552 {
2553     size_t         n = (length + 3) / 4;
2554     unsigned long* p = begin;
2555
2556     while (n--)
2557     {
2558         __asm__ volatile ( "dcbf 0,%0\n\t"
2559                            "sync\n\t"
2560                            "icbi 0,%0"
2561                            :
2562                            : "r" (p)
2563                          );
2564         p++;
2565     }
2566     __asm__ volatile ( "sync\n\t"
2567                        "isync"
2568                      );
2569 }
2570
2571 static void
2572 ocFlushInstructionCache( ObjectCode *oc )
2573 {
2574     /* The main object code */
2575     ocFlushInstructionCacheFrom(oc->image
2576 #ifdef darwin_HOST_OS
2577             + oc->misalignment
2578 #endif
2579             , oc->fileSize);
2580
2581     /* Jump Islands */
2582     ocFlushInstructionCacheFrom(oc->symbol_extras, sizeof(SymbolExtra) * oc->n_symbol_extras);
2583 }
2584 #endif /* powerpc_HOST_ARCH */
2585
2586
2587 /* --------------------------------------------------------------------------
2588  * PEi386 specifics (Win32 targets)
2589  * ------------------------------------------------------------------------*/
2590
2591 /* The information for this linker comes from
2592       Microsoft Portable Executable
2593       and Common Object File Format Specification
2594       revision 5.1 January 1998
2595    which SimonM says comes from the MS Developer Network CDs.
2596
2597    It can be found there (on older CDs), but can also be found
2598    online at:
2599
2600       http://www.microsoft.com/hwdev/hardware/PECOFF.asp
2601
2602    (this is Rev 6.0 from February 1999).
2603
2604    Things move, so if that fails, try searching for it via
2605
2606       http://www.google.com/search?q=PE+COFF+specification
2607
2608    The ultimate reference for the PE format is the Winnt.h
2609    header file that comes with the Platform SDKs; as always,
2610    implementations will drift wrt their documentation.
2611
2612    A good background article on the PE format is Matt Pietrek's
2613    March 1994 article in Microsoft System Journal (MSJ)
2614    (Vol.9, No. 3): "Peering Inside the PE: A Tour of the
2615    Win32 Portable Executable File Format." The info in there
2616    has recently been updated in a two part article in
2617    MSDN magazine, issues Feb and March 2002,
2618    "Inside Windows: An In-Depth Look into the Win32 Portable
2619    Executable File Format"
2620
2621    John Levine's book "Linkers and Loaders" contains useful
2622    info on PE too.
2623 */
2624
2625
2626 #if defined(OBJFORMAT_PEi386)
2627
2628
2629
2630 typedef unsigned char  UChar;
2631 typedef unsigned short UInt16;
2632 typedef unsigned int   UInt32;
2633 typedef          int   Int32;
2634
2635
2636 typedef
2637    struct {
2638       UInt16 Machine;
2639       UInt16 NumberOfSections;
2640       UInt32 TimeDateStamp;
2641       UInt32 PointerToSymbolTable;
2642       UInt32 NumberOfSymbols;
2643       UInt16 SizeOfOptionalHeader;
2644       UInt16 Characteristics;
2645    }
2646    COFF_header;
2647
2648 #define sizeof_COFF_header 20
2649
2650
2651 typedef
2652    struct {
2653       UChar  Name[8];
2654       UInt32 VirtualSize;
2655       UInt32 VirtualAddress;
2656       UInt32 SizeOfRawData;
2657       UInt32 PointerToRawData;
2658       UInt32 PointerToRelocations;
2659       UInt32 PointerToLinenumbers;
2660       UInt16 NumberOfRelocations;
2661       UInt16 NumberOfLineNumbers;
2662       UInt32 Characteristics;
2663    }
2664    COFF_section;
2665
2666 #define sizeof_COFF_section 40
2667
2668
2669 typedef
2670    struct {
2671       UChar  Name[8];
2672       UInt32 Value;
2673       UInt16 SectionNumber;
2674       UInt16 Type;
2675       UChar  StorageClass;
2676       UChar  NumberOfAuxSymbols;
2677    }
2678    COFF_symbol;
2679
2680 #define sizeof_COFF_symbol 18
2681
2682
2683 typedef
2684    struct {
2685       UInt32 VirtualAddress;
2686       UInt32 SymbolTableIndex;
2687       UInt16 Type;
2688    }
2689    COFF_reloc;
2690
2691 #define sizeof_COFF_reloc 10
2692
2693
2694 /* From PE spec doc, section 3.3.2 */
2695 /* Note use of MYIMAGE_* since IMAGE_* are already defined in
2696    windows.h -- for the same purpose, but I want to know what I'm
2697    getting, here. */
2698 #define MYIMAGE_FILE_RELOCS_STRIPPED     0x0001
2699 #define MYIMAGE_FILE_EXECUTABLE_IMAGE    0x0002
2700 #define MYIMAGE_FILE_DLL                 0x2000
2701 #define MYIMAGE_FILE_SYSTEM              0x1000
2702 #define MYIMAGE_FILE_BYTES_REVERSED_HI   0x8000
2703 #define MYIMAGE_FILE_BYTES_REVERSED_LO   0x0080
2704 #define MYIMAGE_FILE_32BIT_MACHINE       0x0100
2705
2706 /* From PE spec doc, section 5.4.2 and 5.4.4 */
2707 #define MYIMAGE_SYM_CLASS_EXTERNAL       2
2708 #define MYIMAGE_SYM_CLASS_STATIC         3
2709 #define MYIMAGE_SYM_UNDEFINED            0
2710
2711 /* From PE spec doc, section 4.1 */
2712 #define MYIMAGE_SCN_CNT_CODE             0x00000020
2713 #define MYIMAGE_SCN_CNT_INITIALIZED_DATA 0x00000040
2714 #define MYIMAGE_SCN_LNK_NRELOC_OVFL      0x01000000
2715
2716 /* From PE spec doc, section 5.2.1 */
2717 #define MYIMAGE_REL_I386_DIR32           0x0006
2718 #define MYIMAGE_REL_I386_REL32           0x0014
2719
2720
2721 /* We use myindex to calculate array addresses, rather than
2722    simply doing the normal subscript thing.  That's because
2723    some of the above structs have sizes which are not
2724    a whole number of words.  GCC rounds their sizes up to a
2725    whole number of words, which means that the address calcs
2726    arising from using normal C indexing or pointer arithmetic
2727    are just plain wrong.  Sigh.
2728 */
2729 static UChar *
2730 myindex ( int scale, void* base, int index )
2731 {
2732    return
2733       ((UChar*)base) + scale * index;
2734 }
2735
2736
2737 static void
2738 printName ( UChar* name, UChar* strtab )
2739 {
2740    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2741       UInt32 strtab_offset = * (UInt32*)(name+4);
2742       debugBelch("%s", strtab + strtab_offset );
2743    } else {
2744       int i;
2745       for (i = 0; i < 8; i++) {
2746          if (name[i] == 0) break;
2747          debugBelch("%c", name[i] );
2748       }
2749    }
2750 }
2751
2752
2753 static void
2754 copyName ( UChar* name, UChar* strtab, UChar* dst, int dstSize )
2755 {
2756    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2757       UInt32 strtab_offset = * (UInt32*)(name+4);
2758       strncpy ( (char*)dst, (char*)strtab+strtab_offset, dstSize );
2759       dst[dstSize-1] = 0;
2760    } else {
2761       int i = 0;
2762       while (1) {
2763          if (i >= 8) break;
2764          if (name[i] == 0) break;
2765          dst[i] = name[i];
2766          i++;
2767       }
2768       dst[i] = 0;
2769    }
2770 }
2771
2772
2773 static UChar *
2774 cstring_from_COFF_symbol_name ( UChar* name, UChar* strtab )
2775 {
2776    UChar* newstr;
2777    /* If the string is longer than 8 bytes, look in the
2778       string table for it -- this will be correctly zero terminated.
2779    */
2780    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2781       UInt32 strtab_offset = * (UInt32*)(name+4);
2782       return ((UChar*)strtab) + strtab_offset;
2783    }
2784    /* Otherwise, if shorter than 8 bytes, return the original,
2785       which by defn is correctly terminated.
2786    */
2787    if (name[7]==0) return name;
2788    /* The annoying case: 8 bytes.  Copy into a temporary
2789       (XXX which is never freed ...)
2790    */
2791    newstr = stgMallocBytes(9, "cstring_from_COFF_symbol_name");
2792    ASSERT(newstr);
2793    strncpy((char*)newstr,(char*)name,8);
2794    newstr[8] = 0;
2795    return newstr;
2796 }
2797
2798 /* Getting the name of a section is mildly tricky, so we make a
2799    function for it.  Sadly, in one case we have to copy the string
2800    (when it is exactly 8 bytes long there's no trailing '\0'), so for
2801    consistency we *always* copy the string; the caller must free it
2802 */
2803 static char *
2804 cstring_from_section_name (UChar* name, UChar* strtab)
2805 {
2806     char *newstr;
2807
2808     if (name[0]=='/') {
2809         int strtab_offset = strtol((char*)name+1,NULL,10);
2810         int len = strlen(((char*)strtab) + strtab_offset);
2811
2812         newstr = stgMallocBytes(len, "cstring_from_section_symbol_name");
2813         strcpy((char*)newstr, (char*)((UChar*)strtab) + strtab_offset);
2814         return newstr;
2815     }
2816     else
2817     {
2818         newstr = stgMallocBytes(9, "cstring_from_section_symbol_name");
2819         ASSERT(newstr);
2820         strncpy((char*)newstr,(char*)name,8);
2821         newstr[8] = 0;
2822         return newstr;
2823     }
2824 }
2825
2826 /* Just compares the short names (first 8 chars) */
2827 static COFF_section *
2828 findPEi386SectionCalled ( ObjectCode* oc,  UChar* name )
2829 {
2830    int i;
2831    COFF_header* hdr
2832       = (COFF_header*)(oc->image);
2833    COFF_section* sectab
2834       = (COFF_section*) (
2835            ((UChar*)(oc->image))
2836            + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2837         );
2838    for (i = 0; i < hdr->NumberOfSections; i++) {
2839       UChar* n1;
2840       UChar* n2;
2841       COFF_section* section_i
2842          = (COFF_section*)
2843            myindex ( sizeof_COFF_section, sectab, i );
2844       n1 = (UChar*) &(section_i->Name);
2845       n2 = name;
2846       if (n1[0]==n2[0] && n1[1]==n2[1] && n1[2]==n2[2] &&
2847           n1[3]==n2[3] && n1[4]==n2[4] && n1[5]==n2[5] &&
2848           n1[6]==n2[6] && n1[7]==n2[7])
2849          return section_i;
2850    }
2851
2852    return NULL;
2853 }
2854
2855
2856 static void
2857 zapTrailingAtSign ( UChar* sym )
2858 {
2859 #  define my_isdigit(c) ((c) >= '0' && (c) <= '9')
2860    int i, j;
2861    if (sym[0] == 0) return;
2862    i = 0;
2863    while (sym[i] != 0) i++;
2864    i--;
2865    j = i;
2866    while (j > 0 && my_isdigit(sym[j])) j--;
2867    if (j > 0 && sym[j] == '@' && j != i) sym[j] = 0;
2868 #  undef my_isdigit
2869 }
2870
2871 static void *
2872 lookupSymbolInDLLs ( UChar *lbl )
2873 {
2874     OpenedDLL* o_dll;
2875     void *sym;
2876
2877     for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
2878         /* debugBelch("look in %s for %s\n", o_dll->name, lbl); */
2879
2880         if (lbl[0] == '_') {
2881             /* HACK: if the name has an initial underscore, try stripping
2882                it off & look that up first. I've yet to verify whether there's
2883                a Rule that governs whether an initial '_' *should always* be
2884                stripped off when mapping from import lib name to the DLL name.
2885             */
2886             sym = GetProcAddress(o_dll->instance, (char*)(lbl+1));
2887             if (sym != NULL) {
2888                 /*debugBelch("found %s in %s\n", lbl+1,o_dll->name);*/
2889                 return sym;
2890             }
2891         }
2892         sym = GetProcAddress(o_dll->instance, (char*)lbl);
2893         if (sym != NULL) {
2894             /*debugBelch("found %s in %s\n", lbl,o_dll->name);*/
2895             return sym;
2896            }
2897     }
2898     return NULL;
2899 }
2900
2901
2902 static int
2903 ocVerifyImage_PEi386 ( ObjectCode* oc )
2904 {
2905    int i;
2906    UInt32 j, noRelocs;
2907    COFF_header*  hdr;
2908    COFF_section* sectab;
2909    COFF_symbol*  symtab;
2910    UChar*        strtab;
2911    /* debugBelch("\nLOADING %s\n", oc->fileName); */
2912    hdr = (COFF_header*)(oc->image);
2913    sectab = (COFF_section*) (
2914                ((UChar*)(oc->image))
2915                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2916             );
2917    symtab = (COFF_symbol*) (
2918                ((UChar*)(oc->image))
2919                + hdr->PointerToSymbolTable
2920             );
2921    strtab = ((UChar*)symtab)
2922             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2923
2924    if (hdr->Machine != 0x14c) {
2925       errorBelch("%s: Not x86 PEi386", oc->fileName);
2926       return 0;
2927    }
2928    if (hdr->SizeOfOptionalHeader != 0) {
2929       errorBelch("%s: PEi386 with nonempty optional header", oc->fileName);
2930       return 0;
2931    }
2932    if ( /* (hdr->Characteristics & MYIMAGE_FILE_RELOCS_STRIPPED) || */
2933         (hdr->Characteristics & MYIMAGE_FILE_EXECUTABLE_IMAGE) ||
2934         (hdr->Characteristics & MYIMAGE_FILE_DLL) ||
2935         (hdr->Characteristics & MYIMAGE_FILE_SYSTEM) ) {
2936       errorBelch("%s: Not a PEi386 object file", oc->fileName);
2937       return 0;
2938    }
2939    if ( (hdr->Characteristics & MYIMAGE_FILE_BYTES_REVERSED_HI)
2940         /* || !(hdr->Characteristics & MYIMAGE_FILE_32BIT_MACHINE) */ ) {
2941       errorBelch("%s: Invalid PEi386 word size or endiannness: %d",
2942                  oc->fileName,
2943                  (int)(hdr->Characteristics));
2944       return 0;
2945    }
2946    /* If the string table size is way crazy, this might indicate that
2947       there are more than 64k relocations, despite claims to the
2948       contrary.  Hence this test. */
2949    /* debugBelch("strtab size %d\n", * (UInt32*)strtab); */
2950 #if 0
2951    if ( (*(UInt32*)strtab) > 600000 ) {
2952       /* Note that 600k has no special significance other than being
2953          big enough to handle the almost-2MB-sized lumps that
2954          constitute HSwin32*.o. */
2955       debugBelch("PEi386 object has suspiciously large string table; > 64k relocs?");
2956       return 0;
2957    }
2958 #endif
2959
2960    /* No further verification after this point; only debug printing. */
2961    i = 0;
2962    IF_DEBUG(linker, i=1);
2963    if (i == 0) return 1;
2964
2965    debugBelch( "sectab offset = %d\n", ((UChar*)sectab) - ((UChar*)hdr) );
2966    debugBelch( "symtab offset = %d\n", ((UChar*)symtab) - ((UChar*)hdr) );
2967    debugBelch( "strtab offset = %d\n", ((UChar*)strtab) - ((UChar*)hdr) );
2968
2969    debugBelch("\n" );
2970    debugBelch( "Machine:           0x%x\n", (UInt32)(hdr->Machine) );
2971    debugBelch( "# sections:        %d\n",   (UInt32)(hdr->NumberOfSections) );
2972    debugBelch( "time/date:         0x%x\n", (UInt32)(hdr->TimeDateStamp) );
2973    debugBelch( "symtab offset:     %d\n",   (UInt32)(hdr->PointerToSymbolTable) );
2974    debugBelch( "# symbols:         %d\n",   (UInt32)(hdr->NumberOfSymbols) );
2975    debugBelch( "sz of opt hdr:     %d\n",   (UInt32)(hdr->SizeOfOptionalHeader) );
2976    debugBelch( "characteristics:   0x%x\n", (UInt32)(hdr->Characteristics) );
2977
2978    /* Print the section table. */
2979    debugBelch("\n" );
2980    for (i = 0; i < hdr->NumberOfSections; i++) {
2981       COFF_reloc* reltab;
2982       COFF_section* sectab_i
2983          = (COFF_section*)
2984            myindex ( sizeof_COFF_section, sectab, i );
2985       debugBelch(
2986                 "\n"
2987                 "section %d\n"
2988                 "     name `",
2989                 i
2990               );
2991       printName ( sectab_i->Name, strtab );
2992       debugBelch(
2993                 "'\n"
2994                 "    vsize %d\n"
2995                 "    vaddr %d\n"
2996                 "  data sz %d\n"
2997                 " data off %d\n"
2998                 "  num rel %d\n"
2999                 "  off rel %d\n"
3000                 "  ptr raw 0x%x\n",
3001                 sectab_i->VirtualSize,
3002                 sectab_i->VirtualAddress,
3003                 sectab_i->SizeOfRawData,
3004                 sectab_i->PointerToRawData,
3005                 sectab_i->NumberOfRelocations,
3006                 sectab_i->PointerToRelocations,
3007                 sectab_i->PointerToRawData
3008               );
3009       reltab = (COFF_reloc*) (
3010                   ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
3011                );
3012
3013       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
3014         /* If the relocation field (a short) has overflowed, the
3015          * real count can be found in the first reloc entry.
3016          *
3017          * See Section 4.1 (last para) of the PE spec (rev6.0).
3018          */
3019         COFF_reloc* rel = (COFF_reloc*)
3020                            myindex ( sizeof_COFF_reloc, reltab, 0 );
3021         noRelocs = rel->VirtualAddress;
3022         j = 1;
3023       } else {
3024         noRelocs = sectab_i->NumberOfRelocations;
3025         j = 0;
3026       }
3027
3028       for (; j < noRelocs; j++) {
3029          COFF_symbol* sym;
3030          COFF_reloc* rel = (COFF_reloc*)
3031                            myindex ( sizeof_COFF_reloc, reltab, j );
3032          debugBelch(
3033                    "        type 0x%-4x   vaddr 0x%-8x   name `",
3034                    (UInt32)rel->Type,
3035                    rel->VirtualAddress );
3036          sym = (COFF_symbol*)
3037                myindex ( sizeof_COFF_symbol, symtab, rel->SymbolTableIndex );
3038          /* Hmm..mysterious looking offset - what's it for? SOF */
3039          printName ( sym->Name, strtab -10 );
3040          debugBelch("'\n" );
3041       }
3042
3043       debugBelch("\n" );
3044    }
3045    debugBelch("\n" );
3046    debugBelch("string table has size 0x%x\n", * (UInt32*)strtab );
3047    debugBelch("---START of string table---\n");
3048    for (i = 4; i < *(Int32*)strtab; i++) {
3049       if (strtab[i] == 0)
3050          debugBelch("\n"); else
3051          debugBelch("%c", strtab[i] );
3052    }
3053    debugBelch("--- END  of string table---\n");
3054
3055    debugBelch("\n" );
3056    i = 0;
3057    while (1) {
3058       COFF_symbol* symtab_i;
3059       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
3060       symtab_i = (COFF_symbol*)
3061                  myindex ( sizeof_COFF_symbol, symtab, i );
3062       debugBelch(
3063                 "symbol %d\n"
3064                 "     name `",
3065                 i
3066               );
3067       printName ( symtab_i->Name, strtab );
3068       debugBelch(
3069                 "'\n"
3070                 "    value 0x%x\n"
3071                 "   1+sec# %d\n"
3072                 "     type 0x%x\n"
3073                 "   sclass 0x%x\n"
3074                 "     nAux %d\n",
3075                 symtab_i->Value,
3076                 (Int32)(symtab_i->SectionNumber),
3077                 (UInt32)symtab_i->Type,
3078                 (UInt32)symtab_i->StorageClass,
3079                 (UInt32)symtab_i->NumberOfAuxSymbols
3080               );
3081       i += symtab_i->NumberOfAuxSymbols;
3082       i++;
3083    }
3084
3085    debugBelch("\n" );
3086    return 1;
3087 }
3088
3089
3090 static int
3091 ocGetNames_PEi386 ( ObjectCode* oc )
3092 {
3093    COFF_header*  hdr;
3094    COFF_section* sectab;
3095    COFF_symbol*  symtab;
3096    UChar*        strtab;
3097
3098    UChar* sname;
3099    void*  addr;
3100    int    i;
3101
3102    hdr = (COFF_header*)(oc->image);
3103    sectab = (COFF_section*) (
3104                ((UChar*)(oc->image))
3105                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
3106             );
3107    symtab = (COFF_symbol*) (
3108                ((UChar*)(oc->image))
3109                + hdr->PointerToSymbolTable
3110             );
3111    strtab = ((UChar*)(oc->image))
3112             + hdr->PointerToSymbolTable
3113             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
3114
3115    /* Allocate space for any (local, anonymous) .bss sections. */
3116
3117    for (i = 0; i < hdr->NumberOfSections; i++) {
3118       UInt32 bss_sz;
3119       UChar* zspace;
3120       COFF_section* sectab_i
3121          = (COFF_section*)
3122            myindex ( sizeof_COFF_section, sectab, i );
3123
3124       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3125
3126       if (0 != strcmp(secname, ".bss")) {
3127           stgFree(secname);
3128           continue;
3129       }
3130
3131       stgFree(secname);
3132
3133       /* sof 10/05: the PE spec text isn't too clear regarding what
3134        * the SizeOfRawData field is supposed to hold for object
3135        * file sections containing just uninitialized data -- for executables,
3136        * it is supposed to be zero; unclear what it's supposed to be
3137        * for object files. However, VirtualSize is guaranteed to be
3138        * zero for object files, which definitely suggests that SizeOfRawData
3139        * will be non-zero (where else would the size of this .bss section be
3140        * stored?) Looking at the COFF_section info for incoming object files,
3141        * this certainly appears to be the case.
3142        *
3143        * => I suspect we've been incorrectly handling .bss sections in (relocatable)
3144        * object files up until now. This turned out to bite us with ghc-6.4.1's use
3145        * of gcc-3.4.x, which has started to emit initially-zeroed-out local 'static'
3146        * variable decls into to the .bss section. (The specific function in Q which
3147        * triggered this is libraries/base/cbits/dirUtils.c:__hscore_getFolderPath())
3148        */
3149       if (sectab_i->VirtualSize == 0 && sectab_i->SizeOfRawData == 0) continue;
3150       /* This is a non-empty .bss section.  Allocate zeroed space for
3151          it, and set its PointerToRawData field such that oc->image +
3152          PointerToRawData == addr_of_zeroed_space.  */
3153       bss_sz = sectab_i->VirtualSize;
3154       if ( bss_sz < sectab_i->SizeOfRawData) { bss_sz = sectab_i->SizeOfRawData; }
3155       zspace = stgCallocBytes(1, bss_sz, "ocGetNames_PEi386(anonymous bss)");
3156       sectab_i->PointerToRawData = ((UChar*)zspace) - ((UChar*)(oc->image));
3157       addProddableBlock(oc, zspace, bss_sz);
3158       /* debugBelch("BSS anon section at 0x%x\n", zspace); */
3159    }
3160
3161    /* Copy section information into the ObjectCode. */
3162
3163    for (i = 0; i < hdr->NumberOfSections; i++) {
3164       UChar* start;
3165       UChar* end;
3166       UInt32 sz;
3167
3168       SectionKind kind
3169          = SECTIONKIND_OTHER;
3170       COFF_section* sectab_i
3171          = (COFF_section*)
3172            myindex ( sizeof_COFF_section, sectab, i );
3173
3174       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3175
3176       IF_DEBUG(linker, debugBelch("section name = %s\n", secname ));
3177
3178 #     if 0
3179       /* I'm sure this is the Right Way to do it.  However, the
3180          alternative of testing the sectab_i->Name field seems to
3181          work ok with Cygwin.
3182       */
3183       if (sectab_i->Characteristics & MYIMAGE_SCN_CNT_CODE ||
3184           sectab_i->Characteristics & MYIMAGE_SCN_CNT_INITIALIZED_DATA)
3185          kind = SECTIONKIND_CODE_OR_RODATA;
3186 #     endif
3187
3188       if (0==strcmp(".text",(char*)secname) ||
3189           0==strcmp(".rdata",(char*)secname)||
3190           0==strcmp(".rodata",(char*)secname))
3191          kind = SECTIONKIND_CODE_OR_RODATA;
3192       if (0==strcmp(".data",(char*)secname) ||
3193           0==strcmp(".bss",(char*)secname))
3194          kind = SECTIONKIND_RWDATA;
3195
3196       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
3197       sz = sectab_i->SizeOfRawData;
3198       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
3199
3200       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
3201       end   = start + sz - 1;
3202
3203       if (kind == SECTIONKIND_OTHER
3204           /* Ignore sections called which contain stabs debugging
3205              information. */
3206           && 0 != strcmp(".stab", (char*)secname)
3207           && 0 != strcmp(".stabstr", (char*)secname)
3208           /* ignore constructor section for now */
3209           && 0 != strcmp(".ctors", (char*)secname)
3210           /* ignore section generated from .ident */
3211           && 0!= strncmp(".debug", (char*)secname, 6)
3212           /* ignore unknown section that appeared in gcc 3.4.5(?) */
3213           && 0!= strcmp(".reloc", (char*)secname)
3214           && 0 != strcmp(".rdata$zzz", (char*)secname)
3215          ) {
3216          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", secname, oc->fileName);
3217          stgFree(secname);
3218          return 0;
3219       }
3220
3221       if (kind != SECTIONKIND_OTHER && end >= start) {
3222          addSection(oc, kind, start, end);
3223          addProddableBlock(oc, start, end - start + 1);
3224       }
3225
3226       stgFree(secname);
3227    }
3228
3229    /* Copy exported symbols into the ObjectCode. */
3230
3231    oc->n_symbols = hdr->NumberOfSymbols;
3232    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
3233                                   "ocGetNames_PEi386(oc->symbols)");
3234    /* Call me paranoid; I don't care. */
3235    for (i = 0; i < oc->n_symbols; i++)
3236       oc->symbols[i] = NULL;
3237
3238    i = 0;
3239    while (1) {
3240       COFF_symbol* symtab_i;
3241       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
3242       symtab_i = (COFF_symbol*)
3243                  myindex ( sizeof_COFF_symbol, symtab, i );
3244
3245       addr  = NULL;
3246
3247       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
3248           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
3249          /* This symbol is global and defined, viz, exported */
3250          /* for MYIMAGE_SYMCLASS_EXTERNAL
3251                 && !MYIMAGE_SYM_UNDEFINED,
3252             the address of the symbol is:
3253                 address of relevant section + offset in section
3254          */
3255          COFF_section* sectabent
3256             = (COFF_section*) myindex ( sizeof_COFF_section,
3257                                         sectab,
3258                                         symtab_i->SectionNumber-1 );
3259          addr = ((UChar*)(oc->image))
3260                 + (sectabent->PointerToRawData
3261                    + symtab_i->Value);
3262       }
3263       else
3264       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
3265           && symtab_i->Value > 0) {
3266          /* This symbol isn't in any section at all, ie, global bss.
3267             Allocate zeroed space for it. */
3268          addr = stgCallocBytes(1, symtab_i->Value,
3269                                "ocGetNames_PEi386(non-anonymous bss)");
3270          addSection(oc, SECTIONKIND_RWDATA, addr,
3271                         ((UChar*)addr) + symtab_i->Value - 1);
3272          addProddableBlock(oc, addr, symtab_i->Value);
3273          /* debugBelch("BSS      section at 0x%x\n", addr); */
3274       }
3275
3276       if (addr != NULL ) {
3277          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
3278          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
3279          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
3280          ASSERT(i >= 0 && i < oc->n_symbols);
3281          /* cstring_from_COFF_symbol_name always succeeds. */
3282          oc->symbols[i] = (char*)sname;
3283          ghciInsertStrHashTable(oc->fileName, symhash, (char*)sname, addr);
3284       } else {
3285 #        if 0
3286          debugBelch(
3287                    "IGNORING symbol %d\n"
3288                    "     name `",
3289                    i
3290                  );
3291          printName ( symtab_i->Name, strtab );
3292          debugBelch(
3293                    "'\n"
3294                    "    value 0x%x\n"
3295                    "   1+sec# %d\n"
3296                    "     type 0x%x\n"
3297                    "   sclass 0x%x\n"
3298                    "     nAux %d\n",
3299                    symtab_i->Value,
3300                    (Int32)(symtab_i->SectionNumber),
3301                    (UInt32)symtab_i->Type,
3302                    (UInt32)symtab_i->StorageClass,
3303                    (UInt32)symtab_i->NumberOfAuxSymbols
3304                  );
3305 #        endif
3306       }
3307
3308       i += symtab_i->NumberOfAuxSymbols;
3309       i++;
3310    }
3311
3312    return 1;
3313 }
3314
3315
3316 static int
3317 ocResolve_PEi386 ( ObjectCode* oc )
3318 {
3319    COFF_header*  hdr;
3320    COFF_section* sectab;
3321    COFF_symbol*  symtab;
3322    UChar*        strtab;
3323
3324    UInt32        A;
3325    UInt32        S;
3326    UInt32*       pP;
3327
3328    int i;
3329    UInt32 j, noRelocs;
3330
3331    /* ToDo: should be variable-sized?  But is at least safe in the
3332       sense of buffer-overrun-proof. */
3333    UChar symbol[1000];
3334    /* debugBelch("resolving for %s\n", oc->fileName); */
3335
3336    hdr = (COFF_header*)(oc->image);
3337    sectab = (COFF_section*) (
3338                ((UChar*)(oc->image))
3339                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
3340             );
3341    symtab = (COFF_symbol*) (
3342                ((UChar*)(oc->image))
3343                + hdr->PointerToSymbolTable
3344             );
3345    strtab = ((UChar*)(oc->image))
3346             + hdr->PointerToSymbolTable
3347             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
3348
3349    for (i = 0; i < hdr->NumberOfSections; i++) {
3350       COFF_section* sectab_i
3351          = (COFF_section*)
3352            myindex ( sizeof_COFF_section, sectab, i );
3353       COFF_reloc* reltab
3354          = (COFF_reloc*) (
3355               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
3356            );
3357
3358       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3359
3360       /* Ignore sections called which contain stabs debugging
3361          information. */
3362       if (0 == strcmp(".stab", (char*)secname)
3363           || 0 == strcmp(".stabstr", (char*)secname)
3364           || 0 == strcmp(".ctors", (char*)secname)
3365           || 0 == strncmp(".debug", (char*)secname, 6)
3366           || 0 == strcmp(".rdata$zzz", (char*)secname)) {
3367           stgFree(secname);
3368           continue;
3369       }
3370
3371       stgFree(secname);
3372
3373       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
3374         /* If the relocation field (a short) has overflowed, the
3375          * real count can be found in the first reloc entry.
3376          *
3377          * See Section 4.1 (last para) of the PE spec (rev6.0).
3378          *
3379          * Nov2003 update: the GNU linker still doesn't correctly
3380          * handle the generation of relocatable object files with
3381          * overflown relocations. Hence the output to warn of potential
3382          * troubles.
3383          */
3384         COFF_reloc* rel = (COFF_reloc*)
3385                            myindex ( sizeof_COFF_reloc, reltab, 0 );
3386         noRelocs = rel->VirtualAddress;
3387
3388         /* 10/05: we now assume (and check for) a GNU ld that is capable
3389          * of handling object files with (>2^16) of relocs.
3390          */
3391 #if 0
3392         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
3393                    noRelocs);
3394 #endif
3395         j = 1;
3396       } else {
3397         noRelocs = sectab_i->NumberOfRelocations;
3398         j = 0;
3399       }
3400
3401
3402       for (; j < noRelocs; j++) {
3403          COFF_symbol* sym;
3404          COFF_reloc* reltab_j
3405             = (COFF_reloc*)
3406               myindex ( sizeof_COFF_reloc, reltab, j );
3407
3408          /* the location to patch */
3409          pP = (UInt32*)(
3410                  ((UChar*)(oc->image))
3411                  + (sectab_i->PointerToRawData
3412                     + reltab_j->VirtualAddress
3413                     - sectab_i->VirtualAddress )
3414               );
3415          /* the existing contents of pP */
3416          A = *pP;
3417          /* the symbol to connect to */
3418          sym = (COFF_symbol*)
3419                myindex ( sizeof_COFF_symbol,
3420                          symtab, reltab_j->SymbolTableIndex );
3421          IF_DEBUG(linker,
3422                   debugBelch(
3423                             "reloc sec %2d num %3d:  type 0x%-4x   "
3424                             "vaddr 0x%-8x   name `",
3425                             i, j,
3426                             (UInt32)reltab_j->Type,
3427                             reltab_j->VirtualAddress );
3428                             printName ( sym->Name, strtab );
3429                             debugBelch("'\n" ));
3430
3431          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
3432             COFF_section* section_sym
3433                = findPEi386SectionCalled ( oc, sym->Name );
3434             if (!section_sym) {
3435                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
3436                return 0;
3437             }
3438             S = ((UInt32)(oc->image))
3439                 + (section_sym->PointerToRawData
3440                    + sym->Value);
3441          } else {
3442             copyName ( sym->Name, strtab, symbol, 1000-1 );
3443             S = (UInt32) lookupSymbol( (char*)symbol );
3444             if ((void*)S != NULL) goto foundit;
3445             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3446             return 0;
3447            foundit:;
3448          }
3449          checkProddableBlock(oc, pP);
3450          switch (reltab_j->Type) {
3451             case MYIMAGE_REL_I386_DIR32:
3452                *pP = A + S;
3453                break;
3454             case MYIMAGE_REL_I386_REL32:
3455                /* Tricky.  We have to insert a displacement at
3456                   pP which, when added to the PC for the _next_
3457                   insn, gives the address of the target (S).
3458                   Problem is to know the address of the next insn
3459                   when we only know pP.  We assume that this
3460                   literal field is always the last in the insn,
3461                   so that the address of the next insn is pP+4
3462                   -- hence the constant 4.
3463                   Also I don't know if A should be added, but so
3464                   far it has always been zero.
3465
3466                   SOF 05/2005: 'A' (old contents of *pP) have been observed
3467                   to contain values other than zero (the 'wx' object file
3468                   that came with wxhaskell-0.9.4; dunno how it was compiled..).
3469                   So, add displacement to old value instead of asserting
3470                   A to be zero. Fixes wxhaskell-related crashes, and no other
3471                   ill effects have been observed.
3472
3473                   Update: the reason why we're seeing these more elaborate
3474                   relocations is due to a switch in how the NCG compiles SRTs
3475                   and offsets to them from info tables. SRTs live in .(ro)data,
3476                   while info tables live in .text, causing GAS to emit REL32/DISP32
3477                   relocations with non-zero values. Adding the displacement is
3478                   the right thing to do.
3479                */
3480                *pP = S - ((UInt32)pP) - 4 + A;
3481                break;
3482             default:
3483                debugBelch("%s: unhandled PEi386 relocation type %d",
3484                      oc->fileName, reltab_j->Type);
3485                return 0;
3486          }
3487
3488       }
3489    }
3490
3491    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
3492    return 1;
3493 }
3494
3495 #endif /* defined(OBJFORMAT_PEi386) */
3496
3497
3498 /* --------------------------------------------------------------------------
3499  * ELF specifics
3500  * ------------------------------------------------------------------------*/
3501
3502 #if defined(OBJFORMAT_ELF)
3503
3504 #define FALSE 0
3505 #define TRUE  1
3506
3507 #if defined(sparc_HOST_ARCH)
3508 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
3509 #elif defined(i386_HOST_ARCH)
3510 #  define ELF_TARGET_386    /* Used inside <elf.h> */
3511 #elif defined(x86_64_HOST_ARCH)
3512 #  define ELF_TARGET_X64_64
3513 #  define ELF_64BIT
3514 #endif
3515
3516 #if !defined(openbsd_HOST_OS)
3517 #  include <elf.h>
3518 #else
3519 /* openbsd elf has things in different places, with diff names */
3520 #  include <elf_abi.h>
3521 #  include <machine/reloc.h>
3522 #  define R_386_32    RELOC_32
3523 #  define R_386_PC32  RELOC_PC32
3524 #endif
3525
3526 /* If elf.h doesn't define it */
3527 #  ifndef R_X86_64_PC64
3528 #    define R_X86_64_PC64 24
3529 #  endif
3530
3531 /*
3532  * Define a set of types which can be used for both ELF32 and ELF64
3533  */
3534
3535 #ifdef ELF_64BIT
3536 #define ELFCLASS    ELFCLASS64
3537 #define Elf_Addr    Elf64_Addr
3538 #define Elf_Word    Elf64_Word
3539 #define Elf_Sword   Elf64_Sword
3540 #define Elf_Ehdr    Elf64_Ehdr
3541 #define Elf_Phdr    Elf64_Phdr
3542 #define Elf_Shdr    Elf64_Shdr
3543 #define Elf_Sym     Elf64_Sym
3544 #define Elf_Rel     Elf64_Rel
3545 #define Elf_Rela    Elf64_Rela
3546 #ifndef ELF_ST_TYPE
3547 #define ELF_ST_TYPE ELF64_ST_TYPE
3548 #endif
3549 #ifndef ELF_ST_BIND
3550 #define ELF_ST_BIND ELF64_ST_BIND
3551 #endif
3552 #ifndef ELF_R_TYPE
3553 #define ELF_R_TYPE  ELF64_R_TYPE
3554 #endif
3555 #ifndef ELF_R_SYM
3556 #define ELF_R_SYM   ELF64_R_SYM
3557 #endif
3558 #else
3559 #define ELFCLASS    ELFCLASS32
3560 #define Elf_Addr    Elf32_Addr
3561 #define Elf_Word    Elf32_Word
3562 #define Elf_Sword   Elf32_Sword
3563 #define Elf_Ehdr    Elf32_Ehdr
3564 #define Elf_Phdr    Elf32_Phdr
3565 #define Elf_Shdr    Elf32_Shdr
3566 #define Elf_Sym     Elf32_Sym
3567 #define Elf_Rel     Elf32_Rel
3568 #define Elf_Rela    Elf32_Rela
3569 #ifndef ELF_ST_TYPE
3570 #define ELF_ST_TYPE ELF32_ST_TYPE
3571 #endif
3572 #ifndef ELF_ST_BIND
3573 #define ELF_ST_BIND ELF32_ST_BIND
3574 #endif
3575 #ifndef ELF_R_TYPE
3576 #define ELF_R_TYPE  ELF32_R_TYPE
3577 #endif
3578 #ifndef ELF_R_SYM
3579 #define ELF_R_SYM   ELF32_R_SYM
3580 #endif
3581 #endif
3582
3583
3584 /*
3585  * Functions to allocate entries in dynamic sections.  Currently we simply
3586  * preallocate a large number, and we don't check if a entry for the given
3587  * target already exists (a linear search is too slow).  Ideally these
3588  * entries would be associated with symbols.
3589  */
3590
3591 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
3592 #define GOT_SIZE            0x20000
3593 #define FUNCTION_TABLE_SIZE 0x10000
3594 #define PLT_SIZE            0x08000
3595
3596 #ifdef ELF_NEED_GOT
3597 static Elf_Addr got[GOT_SIZE];
3598 static unsigned int gotIndex;
3599 static Elf_Addr gp_val = (Elf_Addr)got;
3600
3601 static Elf_Addr
3602 allocateGOTEntry(Elf_Addr target)
3603 {
3604    Elf_Addr *entry;
3605
3606    if (gotIndex >= GOT_SIZE)
3607       barf("Global offset table overflow");
3608
3609    entry = &got[gotIndex++];
3610    *entry = target;
3611    return (Elf_Addr)entry;
3612 }
3613 #endif
3614
3615 #ifdef ELF_FUNCTION_DESC
3616 typedef struct {
3617    Elf_Addr ip;
3618    Elf_Addr gp;
3619 } FunctionDesc;
3620
3621 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
3622 static unsigned int functionTableIndex;
3623
3624 static Elf_Addr
3625 allocateFunctionDesc(Elf_Addr target)
3626 {
3627    FunctionDesc *entry;
3628
3629    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
3630       barf("Function table overflow");
3631
3632    entry = &functionTable[functionTableIndex++];
3633    entry->ip = target;
3634    entry->gp = (Elf_Addr)gp_val;
3635    return (Elf_Addr)entry;
3636 }
3637
3638 static Elf_Addr
3639 copyFunctionDesc(Elf_Addr target)
3640 {
3641    FunctionDesc *olddesc = (FunctionDesc *)target;
3642    FunctionDesc *newdesc;
3643
3644    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
3645    newdesc->gp = olddesc->gp;
3646    return (Elf_Addr)newdesc;
3647 }
3648 #endif
3649
3650 #ifdef ELF_NEED_PLT
3651
3652 typedef struct {
3653    unsigned char code[sizeof(plt_code)];
3654 } PLTEntry;
3655
3656 static Elf_Addr
3657 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
3658 {
3659    PLTEntry *plt = (PLTEntry *)oc->plt;
3660    PLTEntry *entry;
3661
3662    if (oc->pltIndex >= PLT_SIZE)
3663       barf("Procedure table overflow");
3664
3665    entry = &plt[oc->pltIndex++];
3666    memcpy(entry->code, plt_code, sizeof(entry->code));
3667    PLT_RELOC(entry->code, target);
3668    return (Elf_Addr)entry;
3669 }
3670
3671 static unsigned int
3672 PLTSize(void)
3673 {
3674    return (PLT_SIZE * sizeof(PLTEntry));
3675 }
3676 #endif
3677
3678
3679 /*
3680  * Generic ELF functions
3681  */
3682
3683 static char *
3684 findElfSection ( void* objImage, Elf_Word sh_type )
3685 {
3686    char* ehdrC = (char*)objImage;
3687    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
3688    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
3689    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3690    char* ptr = NULL;
3691    int i;
3692
3693    for (i = 0; i < ehdr->e_shnum; i++) {
3694       if (shdr[i].sh_type == sh_type
3695           /* Ignore the section header's string table. */
3696           && i != ehdr->e_shstrndx
3697           /* Ignore string tables named .stabstr, as they contain
3698              debugging info. */
3699           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3700          ) {
3701          ptr = ehdrC + shdr[i].sh_offset;
3702          break;
3703       }
3704    }
3705    return ptr;
3706 }
3707
3708 static int
3709 ocVerifyImage_ELF ( ObjectCode* oc )
3710 {
3711    Elf_Shdr* shdr;
3712    Elf_Sym*  stab;
3713    int i, j, nent, nstrtab, nsymtabs;
3714    char* sh_strtab;
3715    char* strtab;
3716
3717    char*     ehdrC = (char*)(oc->image);
3718    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
3719
3720    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
3721        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
3722        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
3723        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
3724       errorBelch("%s: not an ELF object", oc->fileName);
3725       return 0;
3726    }
3727
3728    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
3729       errorBelch("%s: unsupported ELF format", oc->fileName);
3730       return 0;
3731    }
3732
3733    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
3734        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
3735    } else
3736    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
3737        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
3738    } else {
3739        errorBelch("%s: unknown endiannness", oc->fileName);
3740        return 0;
3741    }
3742
3743    if (ehdr->e_type != ET_REL) {
3744       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
3745       return 0;
3746    }
3747    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
3748
3749    IF_DEBUG(linker,debugBelch( "Architecture is " ));
3750    switch (ehdr->e_machine) {
3751       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
3752 #ifdef EM_SPARC32PLUS
3753       case EM_SPARC32PLUS:
3754 #endif
3755       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
3756 #ifdef EM_IA_64
3757       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
3758 #endif
3759       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
3760 #ifdef EM_X86_64
3761       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
3762 #elif defined(EM_AMD64)
3763       case EM_AMD64: IF_DEBUG(linker,debugBelch( "amd64" )); break;
3764 #endif
3765       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
3766                      errorBelch("%s: unknown architecture (e_machine == %d)"
3767                                 , oc->fileName, ehdr->e_machine);
3768                      return 0;
3769    }
3770
3771    IF_DEBUG(linker,debugBelch(
3772              "\nSection header table: start %ld, n_entries %d, ent_size %d\n",
3773              (long)ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
3774
3775    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
3776
3777    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3778
3779    if (ehdr->e_shstrndx == SHN_UNDEF) {
3780       errorBelch("%s: no section header string table", oc->fileName);
3781       return 0;
3782    } else {
3783       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
3784                           ehdr->e_shstrndx));
3785       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3786    }
3787
3788    for (i = 0; i < ehdr->e_shnum; i++) {
3789       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
3790       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
3791       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
3792       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
3793       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
3794                ehdrC + shdr[i].sh_offset,
3795                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
3796
3797       if (shdr[i].sh_type == SHT_REL) {
3798           IF_DEBUG(linker,debugBelch("Rel  " ));
3799       } else if (shdr[i].sh_type == SHT_RELA) {
3800           IF_DEBUG(linker,debugBelch("RelA " ));
3801       } else {
3802           IF_DEBUG(linker,debugBelch("     "));
3803       }
3804       if (sh_strtab) {
3805           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
3806       }
3807    }
3808
3809    IF_DEBUG(linker,debugBelch( "\nString tables" ));
3810    strtab = NULL;
3811    nstrtab = 0;
3812    for (i = 0; i < ehdr->e_shnum; i++) {
3813       if (shdr[i].sh_type == SHT_STRTAB
3814           /* Ignore the section header's string table. */
3815           && i != ehdr->e_shstrndx
3816           /* Ignore string tables named .stabstr, as they contain
3817              debugging info. */
3818           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3819          ) {
3820          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
3821          strtab = ehdrC + shdr[i].sh_offset;
3822          nstrtab++;
3823       }
3824    }
3825    if (nstrtab != 1) {
3826       errorBelch("%s: no string tables, or too many", oc->fileName);
3827       return 0;
3828    }
3829
3830    nsymtabs = 0;
3831    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
3832    for (i = 0; i < ehdr->e_shnum; i++) {
3833       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3834       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
3835       nsymtabs++;
3836       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3837       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3838       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%ld rem)\n",
3839                nent,
3840                (long)shdr[i].sh_size % sizeof(Elf_Sym)
3841              ));
3842       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
3843          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
3844          return 0;
3845       }
3846       for (j = 0; j < nent; j++) {
3847          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
3848          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
3849                              (int)stab[j].st_shndx,
3850                              (int)stab[j].st_size,
3851                              (char*)stab[j].st_value ));
3852
3853          IF_DEBUG(linker,debugBelch("type=" ));
3854          switch (ELF_ST_TYPE(stab[j].st_info)) {
3855             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
3856             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
3857             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
3858             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
3859             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
3860             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
3861          }
3862          IF_DEBUG(linker,debugBelch("  " ));
3863
3864          IF_DEBUG(linker,debugBelch("bind=" ));
3865          switch (ELF_ST_BIND(stab[j].st_info)) {
3866             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
3867             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
3868             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
3869             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
3870          }
3871          IF_DEBUG(linker,debugBelch("  " ));
3872
3873          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
3874       }
3875    }
3876
3877    if (nsymtabs == 0) {
3878       errorBelch("%s: didn't find any symbol tables", oc->fileName);
3879       return 0;
3880    }
3881
3882    return 1;
3883 }
3884
3885 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
3886 {
3887     *is_bss = FALSE;
3888
3889     if (hdr->sh_type == SHT_PROGBITS
3890         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
3891         /* .text-style section */
3892         return SECTIONKIND_CODE_OR_RODATA;
3893     }
3894
3895     if (hdr->sh_type == SHT_PROGBITS
3896             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3897             /* .data-style section */
3898             return SECTIONKIND_RWDATA;
3899     }
3900
3901     if (hdr->sh_type == SHT_PROGBITS
3902         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
3903         /* .rodata-style section */
3904         return SECTIONKIND_CODE_OR_RODATA;
3905     }
3906
3907     if (hdr->sh_type == SHT_NOBITS
3908         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3909         /* .bss-style section */
3910         *is_bss = TRUE;
3911         return SECTIONKIND_RWDATA;
3912     }
3913
3914     return SECTIONKIND_OTHER;
3915 }
3916
3917
3918 static int
3919 ocGetNames_ELF ( ObjectCode* oc )
3920 {
3921    int i, j, k, nent;
3922    Elf_Sym* stab;
3923
3924    char*     ehdrC    = (char*)(oc->image);
3925    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
3926    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
3927    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3928
3929    ASSERT(symhash != NULL);
3930
3931    if (!strtab) {
3932       errorBelch("%s: no strtab", oc->fileName);
3933       return 0;
3934    }
3935
3936    k = 0;
3937    for (i = 0; i < ehdr->e_shnum; i++) {
3938       /* Figure out what kind of section it is.  Logic derived from
3939          Figure 1.14 ("Special Sections") of the ELF document
3940          ("Portable Formats Specification, Version 1.1"). */
3941       int         is_bss = FALSE;
3942       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
3943
3944       if (is_bss && shdr[i].sh_size > 0) {
3945          /* This is a non-empty .bss section.  Allocate zeroed space for
3946             it, and set its .sh_offset field such that
3947             ehdrC + .sh_offset == addr_of_zeroed_space.  */
3948          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
3949                                        "ocGetNames_ELF(BSS)");
3950          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
3951          /*
3952          debugBelch("BSS section at 0x%x, size %d\n",
3953                          zspace, shdr[i].sh_size);
3954          */
3955       }
3956
3957       /* fill in the section info */
3958       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
3959          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
3960          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
3961                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
3962       }
3963
3964       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3965
3966       /* copy stuff into this module's object symbol table */
3967       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3968       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3969
3970       oc->n_symbols = nent;
3971       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3972                                    "ocGetNames_ELF(oc->symbols)");
3973
3974       for (j = 0; j < nent; j++) {
3975
3976          char  isLocal = FALSE; /* avoids uninit-var warning */
3977          char* ad      = NULL;
3978          char* nm      = strtab + stab[j].st_name;
3979          int   secno   = stab[j].st_shndx;
3980
3981          /* Figure out if we want to add it; if so, set ad to its
3982             address.  Otherwise leave ad == NULL. */
3983
3984          if (secno == SHN_COMMON) {
3985             isLocal = FALSE;
3986             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
3987             /*
3988             debugBelch("COMMON symbol, size %d name %s\n",
3989                             stab[j].st_size, nm);
3990             */
3991             /* Pointless to do addProddableBlock() for this area,
3992                since the linker should never poke around in it. */
3993          }
3994          else
3995          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
3996                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
3997               )
3998               /* and not an undefined symbol */
3999               && stab[j].st_shndx != SHN_UNDEF
4000               /* and not in a "special section" */
4001               && stab[j].st_shndx < SHN_LORESERVE
4002               &&
4003               /* and it's a not a section or string table or anything silly */
4004               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
4005                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
4006                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
4007               )
4008             ) {
4009             /* Section 0 is the undefined section, hence > and not >=. */
4010             ASSERT(secno > 0 && secno < ehdr->e_shnum);
4011             /*
4012             if (shdr[secno].sh_type == SHT_NOBITS) {
4013                debugBelch("   BSS symbol, size %d off %d name %s\n",
4014                                stab[j].st_size, stab[j].st_value, nm);
4015             }
4016             */
4017             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
4018             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
4019                isLocal = TRUE;
4020             } else {
4021 #ifdef ELF_FUNCTION_DESC
4022                /* dlsym() and the initialisation table both give us function
4023                 * descriptors, so to be consistent we store function descriptors
4024                 * in the symbol table */
4025                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
4026                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
4027 #endif
4028                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s\n",
4029                                       ad, oc->fileName, nm ));
4030                isLocal = FALSE;
4031             }
4032          }
4033
4034          /* And the decision is ... */
4035
4036          if (ad != NULL) {
4037             ASSERT(nm != NULL);
4038             oc->symbols[j] = nm;
4039             /* Acquire! */
4040             if (isLocal) {
4041                /* Ignore entirely. */
4042             } else {
4043                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
4044             }
4045          } else {
4046             /* Skip. */
4047             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
4048                                    strtab + stab[j].st_name ));
4049             /*
4050             debugBelch(
4051                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
4052                     (int)ELF_ST_BIND(stab[j].st_info),
4053                     (int)ELF_ST_TYPE(stab[j].st_info),
4054                     (int)stab[j].st_shndx,
4055                     strtab + stab[j].st_name
4056                    );
4057             */
4058             oc->symbols[j] = NULL;
4059          }
4060
4061       }
4062    }
4063
4064    return 1;
4065 }
4066
4067 /* Do ELF relocations which lack an explicit addend.  All x86-linux
4068    relocations appear to be of this form. */
4069 static int
4070 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
4071                          Elf_Shdr* shdr, int shnum,
4072                          Elf_Sym*  stab, char* strtab )
4073 {
4074    int j;
4075    char *symbol;
4076    Elf_Word* targ;
4077    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
4078    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
4079    int target_shndx = shdr[shnum].sh_info;
4080    int symtab_shndx = shdr[shnum].sh_link;
4081
4082    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
4083    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
4084    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
4085                           target_shndx, symtab_shndx ));
4086
4087    /* Skip sections that we're not interested in. */
4088    {
4089        int is_bss;
4090        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
4091        if (kind == SECTIONKIND_OTHER) {
4092            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
4093            return 1;
4094        }
4095    }
4096
4097    for (j = 0; j < nent; j++) {
4098       Elf_Addr offset = rtab[j].r_offset;
4099       Elf_Addr info   = rtab[j].r_info;
4100
4101       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
4102       Elf_Word* pP = (Elf_Word*)P;
4103       Elf_Addr  A  = *pP;
4104       Elf_Addr  S;
4105       void*     S_tmp;
4106       Elf_Addr  value;
4107       StgStablePtr stablePtr;
4108       StgPtr stableVal;
4109
4110       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
4111                              j, (void*)offset, (void*)info ));
4112       if (!info) {
4113          IF_DEBUG(linker,debugBelch( " ZERO" ));
4114          S = 0;
4115       } else {
4116          Elf_Sym sym = stab[ELF_R_SYM(info)];
4117          /* First see if it is a local symbol. */
4118          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
4119             /* Yes, so we can get the address directly from the ELF symbol
4120                table. */
4121             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
4122             S = (Elf_Addr)
4123                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
4124                        + stab[ELF_R_SYM(info)].st_value);
4125
4126          } else {
4127             symbol = strtab + sym.st_name;
4128             stablePtr = (StgStablePtr)lookupHashTable(stablehash, (StgWord)symbol);
4129             if (NULL == stablePtr) {
4130               /* No, so look up the name in our global table. */
4131               S_tmp = lookupSymbol( symbol );
4132               S = (Elf_Addr)S_tmp;
4133             } else {
4134               stableVal = deRefStablePtr( stablePtr );
4135               S_tmp = stableVal;
4136               S = (Elf_Addr)S_tmp;
4137             }
4138          }
4139          if (!S) {
4140             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
4141             return 0;
4142          }
4143          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
4144       }
4145
4146       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
4147                              (void*)P, (void*)S, (void*)A ));
4148       checkProddableBlock ( oc, pP );
4149
4150       value = S + A;
4151
4152       switch (ELF_R_TYPE(info)) {
4153 #        ifdef i386_HOST_ARCH
4154          case R_386_32:   *pP = value;     break;
4155          case R_386_PC32: *pP = value - P; break;
4156 #        endif
4157          default:
4158             errorBelch("%s: unhandled ELF relocation(Rel) type %lu\n",
4159                   oc->fileName, (lnat)ELF_R_TYPE(info));
4160             return 0;
4161       }
4162
4163    }
4164    return 1;
4165 }
4166
4167 /* Do ELF relocations for which explicit addends are supplied.
4168    sparc-solaris relocations appear to be of this form. */
4169 static int
4170 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
4171                           Elf_Shdr* shdr, int shnum,
4172                           Elf_Sym*  stab, char* strtab )
4173 {
4174    int j;
4175    char *symbol = NULL;
4176    Elf_Addr targ;
4177    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
4178    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
4179    int target_shndx = shdr[shnum].sh_info;
4180    int symtab_shndx = shdr[shnum].sh_link;
4181
4182    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
4183    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
4184    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
4185                           target_shndx, symtab_shndx ));
4186
4187    for (j = 0; j < nent; j++) {
4188 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
4189       /* This #ifdef only serves to avoid unused-var warnings. */
4190       Elf_Addr  offset = rtab[j].r_offset;
4191       Elf_Addr  P      = targ + offset;
4192 #endif
4193       Elf_Addr  info   = rtab[j].r_info;
4194       Elf_Addr  A      = rtab[j].r_addend;
4195       Elf_Addr  S;
4196       void*     S_tmp;
4197       Elf_Addr  value;
4198 #     if defined(sparc_HOST_ARCH)
4199       Elf_Word* pP = (Elf_Word*)P;
4200       Elf_Word  w1, w2;
4201 #     elif defined(powerpc_HOST_ARCH)
4202       Elf_Sword delta;
4203 #     endif
4204
4205       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
4206                              j, (void*)offset, (void*)info,
4207                                 (void*)A ));
4208       if (!info) {
4209          IF_DEBUG(linker,debugBelch( " ZERO" ));
4210          S = 0;
4211       } else {
4212          Elf_Sym sym = stab[ELF_R_SYM(info)];
4213          /* First see if it is a local symbol. */
4214          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
4215             /* Yes, so we can get the address directly from the ELF symbol
4216                table. */
4217             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
4218             S = (Elf_Addr)
4219                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
4220                        + stab[ELF_R_SYM(info)].st_value);
4221 #ifdef ELF_FUNCTION_DESC
4222             /* Make a function descriptor for this function */
4223             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
4224                S = allocateFunctionDesc(S + A);
4225                A = 0;
4226             }
4227 #endif
4228          } else {
4229             /* No, so look up the name in our global table. */
4230             symbol = strtab + sym.st_name;
4231             S_tmp = lookupSymbol( symbol );
4232             S = (Elf_Addr)S_tmp;
4233
4234 #ifdef ELF_FUNCTION_DESC
4235             /* If a function, already a function descriptor - we would
4236                have to copy it to add an offset. */
4237             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
4238                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
4239 #endif
4240          }
4241          if (!S) {
4242            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
4243            return 0;
4244          }
4245          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
4246       }
4247
4248       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
4249                                         (void*)P, (void*)S, (void*)A ));
4250       /* checkProddableBlock ( oc, (void*)P ); */
4251
4252       value = S + A;
4253
4254       switch (ELF_R_TYPE(info)) {
4255 #        if defined(sparc_HOST_ARCH)
4256          case R_SPARC_WDISP30:
4257             w1 = *pP & 0xC0000000;
4258             w2 = (Elf_Word)((value - P) >> 2);
4259             ASSERT((w2 & 0xC0000000) == 0);
4260             w1 |= w2;
4261             *pP = w1;
4262             break;
4263          case R_SPARC_HI22:
4264             w1 = *pP & 0xFFC00000;
4265             w2 = (Elf_Word)(value >> 10);
4266             ASSERT((w2 & 0xFFC00000) == 0);
4267             w1 |= w2;
4268             *pP = w1;
4269             break;
4270          case R_SPARC_LO10:
4271             w1 = *pP & ~0x3FF;
4272             w2 = (Elf_Word)(value & 0x3FF);
4273             ASSERT((w2 & ~0x3FF) == 0);
4274             w1 |= w2;
4275             *pP = w1;
4276             break;
4277
4278          /* According to the Sun documentation:
4279             R_SPARC_UA32
4280             This relocation type resembles R_SPARC_32, except it refers to an
4281             unaligned word. That is, the word to be relocated must be treated
4282             as four separate bytes with arbitrary alignment, not as a word
4283             aligned according to the architecture requirements.
4284          */
4285          case R_SPARC_UA32:
4286             w2  = (Elf_Word)value;
4287
4288             // SPARC doesn't do misaligned writes of 32 bit words,
4289             //       so we have to do this one byte-at-a-time.
4290             char *pPc   = (char*)pP;
4291             pPc[0]      = (char) ((Elf_Word)(w2 & 0xff000000) >> 24);
4292             pPc[1]      = (char) ((Elf_Word)(w2 & 0x00ff0000) >> 16);
4293             pPc[2]      = (char) ((Elf_Word)(w2 & 0x0000ff00) >> 8);
4294             pPc[3]      = (char) ((Elf_Word)(w2 & 0x000000ff));
4295             break;
4296
4297          case R_SPARC_32:
4298             w2 = (Elf_Word)value;
4299             *pP = w2;
4300             break;
4301 #        elif defined(powerpc_HOST_ARCH)
4302          case R_PPC_ADDR16_LO:
4303             *(Elf32_Half*) P = value;
4304             break;
4305
4306          case R_PPC_ADDR16_HI:
4307             *(Elf32_Half*) P = value >> 16;
4308             break;
4309
4310          case R_PPC_ADDR16_HA:
4311             *(Elf32_Half*) P = (value + 0x8000) >> 16;
4312             break;
4313
4314          case R_PPC_ADDR32:
4315             *(Elf32_Word *) P = value;
4316             break;
4317
4318          case R_PPC_REL32:
4319             *(Elf32_Word *) P = value - P;
4320             break;
4321
4322          case R_PPC_REL24:
4323             delta = value - P;
4324
4325             if( delta << 6 >> 6 != delta )
4326             {
4327                value = (Elf_Addr) (&makeSymbolExtra( oc, ELF_R_SYM(info), value )
4328                                         ->jumpIsland);
4329                delta = value - P;
4330
4331                if( value == 0 || delta << 6 >> 6 != delta )
4332                {
4333                   barf( "Unable to make SymbolExtra for #%d",
4334                         ELF_R_SYM(info) );
4335                   return 0;
4336                }
4337             }
4338
4339             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
4340                                           | (delta & 0x3fffffc);
4341             break;
4342 #        endif
4343
4344 #if x86_64_HOST_ARCH
4345       case R_X86_64_64:
4346           *(Elf64_Xword *)P = value;
4347           break;
4348
4349       case R_X86_64_PC32:
4350       {
4351 #if defined(ALWAYS_PIC)
4352           barf("R_X86_64_PC32 relocation, but ALWAYS_PIC.");
4353 #else
4354           StgInt64 off = value - P;
4355           if (off >= 0x7fffffffL || off < -0x80000000L) {
4356 #if X86_64_ELF_NONPIC_HACK
4357               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4358                                                 -> jumpIsland;
4359               off = pltAddress + A - P;
4360 #else
4361               barf("R_X86_64_PC32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4362                    symbol, off, oc->fileName );
4363 #endif
4364           }
4365           *(Elf64_Word *)P = (Elf64_Word)off;
4366 #endif
4367           break;
4368       }
4369
4370       case R_X86_64_PC64:
4371       {
4372           StgInt64 off = value - P;
4373           *(Elf64_Word *)P = (Elf64_Word)off;
4374           break;
4375       }
4376
4377       case R_X86_64_32:
4378 #if defined(ALWAYS_PIC)
4379           barf("R_X86_64_32 relocation, but ALWAYS_PIC.");
4380 #else
4381           if (value >= 0x7fffffffL) {
4382 #if X86_64_ELF_NONPIC_HACK
4383               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4384                                                 -> jumpIsland;
4385               value = pltAddress + A;
4386 #else
4387               barf("R_X86_64_32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4388                    symbol, value, oc->fileName );
4389 #endif
4390           }
4391           *(Elf64_Word *)P = (Elf64_Word)value;
4392 #endif
4393           break;
4394
4395       case R_X86_64_32S:
4396 #if defined(ALWAYS_PIC)
4397           barf("R_X86_64_32S relocation, but ALWAYS_PIC.");
4398 #else
4399           if ((StgInt64)value > 0x7fffffffL || (StgInt64)value < -0x80000000L) {
4400 #if X86_64_ELF_NONPIC_HACK
4401               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4402                                                 -> jumpIsland;
4403               value = pltAddress + A;
4404 #else
4405               barf("R_X86_64_32S relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4406                    symbol, value, oc->fileName );
4407 #endif
4408           }
4409           *(Elf64_Sword *)P = (Elf64_Sword)value;
4410 #endif
4411           break;
4412
4413       case R_X86_64_GOTPCREL:
4414       {
4415           StgInt64 gotAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)->addr;
4416           StgInt64 off = gotAddress + A - P;
4417           *(Elf64_Word *)P = (Elf64_Word)off;
4418           break;
4419       }
4420
4421       case R_X86_64_PLT32:
4422       {
4423 #if defined(ALWAYS_PIC)
4424           barf("R_X86_64_PLT32 relocation, but ALWAYS_PIC.");
4425 #else
4426           StgInt64 off = value - P;
4427           if (off >= 0x7fffffffL || off < -0x80000000L) {
4428               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4429                                                     -> jumpIsland;
4430               off = pltAddress + A - P;
4431           }
4432           *(Elf64_Word *)P = (Elf64_Word)off;
4433 #endif
4434           break;
4435       }
4436 #endif
4437
4438          default:
4439             errorBelch("%s: unhandled ELF relocation(RelA) type %lu\n",
4440                   oc->fileName, (lnat)ELF_R_TYPE(info));
4441             return 0;
4442       }
4443
4444    }
4445    return 1;
4446 }
4447
4448 static int
4449 ocResolve_ELF ( ObjectCode* oc )
4450 {
4451    char *strtab;
4452    int   shnum, ok;
4453    Elf_Sym*  stab  = NULL;
4454    char*     ehdrC = (char*)(oc->image);
4455    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
4456    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
4457
4458    /* first find "the" symbol table */
4459    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
4460
4461    /* also go find the string table */
4462    strtab = findElfSection ( ehdrC, SHT_STRTAB );
4463
4464    if (stab == NULL || strtab == NULL) {
4465       errorBelch("%s: can't find string or symbol table", oc->fileName);
4466       return 0;
4467    }
4468
4469    /* Process the relocation sections. */
4470    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
4471       if (shdr[shnum].sh_type == SHT_REL) {
4472          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
4473                                        shnum, stab, strtab );
4474          if (!ok) return ok;
4475       }
4476       else
4477       if (shdr[shnum].sh_type == SHT_RELA) {
4478          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
4479                                         shnum, stab, strtab );
4480          if (!ok) return ok;
4481       }
4482    }
4483
4484 #if defined(powerpc_HOST_ARCH)
4485    ocFlushInstructionCache( oc );
4486 #endif
4487
4488    return 1;
4489 }
4490
4491 /*
4492  * PowerPC & X86_64 ELF specifics
4493  */
4494
4495 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
4496
4497 static int ocAllocateSymbolExtras_ELF( ObjectCode *oc )
4498 {
4499   Elf_Ehdr *ehdr;
4500   Elf_Shdr* shdr;
4501   int i;
4502
4503   ehdr = (Elf_Ehdr *) oc->image;
4504   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
4505
4506   for( i = 0; i < ehdr->e_shnum; i++ )
4507     if( shdr[i].sh_type == SHT_SYMTAB )
4508       break;
4509
4510   if( i == ehdr->e_shnum )
4511   {
4512     errorBelch( "This ELF file contains no symtab" );
4513     return 0;
4514   }
4515
4516   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
4517   {
4518     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
4519       (int) shdr[i].sh_entsize, (int) sizeof( Elf_Sym ) );
4520
4521     return 0;
4522   }
4523
4524   return ocAllocateSymbolExtras( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
4525 }
4526
4527 #endif /* powerpc */
4528
4529 #endif /* ELF */
4530
4531 /* --------------------------------------------------------------------------
4532  * Mach-O specifics
4533  * ------------------------------------------------------------------------*/
4534
4535 #if defined(OBJFORMAT_MACHO)
4536
4537 /*
4538   Support for MachO linking on Darwin/MacOS X
4539   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
4540
4541   I hereby formally apologize for the hackish nature of this code.
4542   Things that need to be done:
4543   *) implement ocVerifyImage_MachO
4544   *) add still more sanity checks.
4545 */
4546
4547 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4548 #define mach_header mach_header_64
4549 #define segment_command segment_command_64
4550 #define section section_64
4551 #define nlist nlist_64
4552 #endif
4553
4554 #ifdef powerpc_HOST_ARCH
4555 static int
4556 ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4557 {
4558     struct mach_header *header = (struct mach_header *) oc->image;
4559     struct load_command *lc = (struct load_command *) (header + 1);
4560     unsigned i;
4561
4562     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: start\n"));
4563
4564     for (i = 0; i < header->ncmds; i++) {
4565         if (lc->cmd == LC_SYMTAB) {
4566
4567                 // Find out the first and last undefined external
4568                 // symbol, so we don't have to allocate too many
4569             // jump islands/GOT entries.
4570
4571             struct symtab_command *symLC = (struct symtab_command *) lc;
4572             unsigned min = symLC->nsyms, max = 0;
4573             struct nlist *nlist =
4574                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
4575                       : NULL;
4576
4577             for (i = 0; i < symLC->nsyms; i++) {
4578
4579                 if (nlist[i].n_type & N_STAB) {
4580                     ;
4581                 } else if (nlist[i].n_type & N_EXT) {
4582
4583                     if((nlist[i].n_type & N_TYPE) == N_UNDF
4584                         && (nlist[i].n_value == 0)) {
4585
4586                         if (i < min) {
4587                             min = i;
4588                         }
4589
4590                         if (i > max) {
4591                             max = i;
4592                     }
4593                 }
4594             }
4595             }
4596
4597             if (max >= min) {
4598                 return ocAllocateSymbolExtras(oc, max - min + 1, min);
4599             }
4600
4601             break;
4602         }
4603
4604         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4605     }
4606
4607     return ocAllocateSymbolExtras(oc,0,0);
4608 }
4609
4610 #endif
4611 #ifdef x86_64_HOST_ARCH
4612 static int
4613 ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4614 {
4615     struct mach_header *header = (struct mach_header *) oc->image;
4616     struct load_command *lc = (struct load_command *) (header + 1);
4617     unsigned i;
4618
4619     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: start\n"));
4620
4621     for (i = 0; i < header->ncmds; i++) {
4622         if (lc->cmd == LC_SYMTAB) {
4623
4624                 // Just allocate one entry for every symbol
4625             struct symtab_command *symLC = (struct symtab_command *) lc;
4626
4627             IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: allocate %d symbols\n", symLC->nsyms));
4628             IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: done\n"));
4629             return ocAllocateSymbolExtras(oc, symLC->nsyms, 0);
4630         }
4631
4632         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4633     }
4634
4635     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: allocated no symbols\n"));
4636     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: done\n"));
4637     return ocAllocateSymbolExtras(oc,0,0);
4638 }
4639 #endif
4640
4641 static int
4642 ocVerifyImage_MachO(ObjectCode * oc)
4643 {
4644     char *image = (char*) oc->image;
4645     struct mach_header *header = (struct mach_header*) image;
4646
4647     IF_DEBUG(linker, debugBelch("ocVerifyImage_MachO: start\n"));
4648
4649 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4650     if(header->magic != MH_MAGIC_64) {
4651         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4652                    oc->fileName, MH_MAGIC_64, header->magic);
4653         return 0;
4654     }
4655 #else
4656     if(header->magic != MH_MAGIC) {
4657         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4658                    oc->fileName, MH_MAGIC, header->magic);
4659         return 0;
4660     }
4661 #endif
4662
4663     // FIXME: do some more verifying here
4664     IF_DEBUG(linker, debugBelch("ocVerifyImage_MachO: done\n"));
4665     return 1;
4666 }
4667
4668 static int
4669 resolveImports(
4670     ObjectCode* oc,
4671     char *image,
4672     struct symtab_command *symLC,
4673     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
4674     unsigned long *indirectSyms,
4675     struct nlist *nlist)
4676 {
4677     unsigned i;
4678     size_t itemSize = 4;
4679
4680     IF_DEBUG(linker, debugBelch("resolveImports: start\n"));
4681
4682 #if i386_HOST_ARCH
4683     int isJumpTable = 0;
4684
4685     if (strcmp(sect->sectname,"__jump_table") == 0) {
4686         isJumpTable = 1;
4687         itemSize = 5;
4688         ASSERT(sect->reserved2 == itemSize);
4689     }
4690
4691 #endif
4692
4693     for(i=0; i*itemSize < sect->size;i++)
4694     {
4695         // according to otool, reserved1 contains the first index into the indirect symbol table
4696         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
4697         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4698         void *addr = NULL;
4699
4700         IF_DEBUG(linker, debugBelch("resolveImports: resolving %s\n", nm));
4701
4702         if ((symbol->n_type & N_TYPE) == N_UNDF
4703             && (symbol->n_type & N_EXT) && (symbol->n_value != 0)) {
4704             addr = (void*) (symbol->n_value);
4705             IF_DEBUG(linker, debugBelch("resolveImports: undefined external %s has value %p\n", nm, addr));
4706         } else {
4707             addr = lookupSymbol(nm);
4708             IF_DEBUG(linker, debugBelch("resolveImports: looking up %s, %p\n", nm, addr));
4709         }
4710         if (!addr)
4711         {
4712             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
4713             return 0;
4714         }
4715         ASSERT(addr);
4716
4717 #if i386_HOST_ARCH
4718         if (isJumpTable) {
4719             checkProddableBlock(oc,image + sect->offset + i*itemSize);
4720
4721             *(image + sect->offset + i * itemSize) = 0xe9; // jmp opcode
4722             *(unsigned*)(image + sect->offset + i*itemSize + 1)
4723                 = (char*)addr - (image + sect->offset + i*itemSize + 5);
4724         }
4725         else
4726 #endif
4727         {
4728             checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
4729             ((void**)(image + sect->offset))[i] = addr;
4730         }
4731     }
4732
4733     IF_DEBUG(linker, debugBelch("resolveImports: done\n"));
4734     return 1;
4735 }
4736
4737 static unsigned long relocateAddress(
4738     ObjectCode* oc,
4739     int nSections,
4740     struct section* sections,
4741     unsigned long address)
4742 {
4743     int i;
4744     IF_DEBUG(linker, debugBelch("relocateAddress: start\n"));
4745     for (i = 0; i < nSections; i++)
4746     {
4747             IF_DEBUG(linker, debugBelch("    relocating address in section %d\n", i));
4748         if (sections[i].addr <= address
4749             && address < sections[i].addr + sections[i].size)
4750         {
4751             return (unsigned long)oc->image
4752                     + sections[i].offset + address - sections[i].addr;
4753         }
4754     }
4755     barf("Invalid Mach-O file:"
4756          "Address out of bounds while relocating object file");
4757     return 0;
4758 }
4759
4760 static int relocateSection(
4761     ObjectCode* oc,
4762     char *image,
4763     struct symtab_command *symLC, struct nlist *nlist,
4764     int nSections, struct section* sections, struct section *sect)
4765 {
4766     struct relocation_info *relocs;
4767     int i, n;
4768
4769     IF_DEBUG(linker, debugBelch("relocateSection: start\n"));
4770
4771     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
4772         return 1;
4773     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
4774         return 1;
4775     else if(!strcmp(sect->sectname,"__la_sym_ptr2"))
4776         return 1;
4777     else if(!strcmp(sect->sectname,"__la_sym_ptr3"))
4778         return 1;
4779
4780     n = sect->nreloc;
4781     IF_DEBUG(linker, debugBelch("relocateSection: number of relocations: %d\n", n));
4782
4783     relocs = (struct relocation_info*) (image + sect->reloff);
4784
4785     for(i=0;i<n;i++)
4786     {
4787 #ifdef x86_64_HOST_ARCH
4788         struct relocation_info *reloc = &relocs[i];
4789
4790         char    *thingPtr = image + sect->offset + reloc->r_address;
4791         uint64_t thing;
4792         /* We shouldn't need to initialise this, but gcc on OS X 64 bit
4793            complains that it may be used uninitialized if we don't */
4794         uint64_t value = 0;
4795         uint64_t baseValue;
4796         int type = reloc->r_type;
4797
4798         checkProddableBlock(oc,thingPtr);
4799         switch(reloc->r_length)
4800         {
4801             case 0:
4802                 thing = *(uint8_t*)thingPtr;
4803                 baseValue = (uint64_t)thingPtr + 1;
4804                 break;
4805             case 1:
4806                 thing = *(uint16_t*)thingPtr;
4807                 baseValue = (uint64_t)thingPtr + 2;
4808                 break;
4809             case 2:
4810                 thing = *(uint32_t*)thingPtr;
4811                 baseValue = (uint64_t)thingPtr + 4;
4812                 break;
4813             case 3:
4814                 thing = *(uint64_t*)thingPtr;
4815                 baseValue = (uint64_t)thingPtr + 8;
4816                 break;
4817             default:
4818                 barf("Unknown size.");
4819         }
4820
4821         IF_DEBUG(linker,
4822                  debugBelch("relocateSection: length = %d, thing = %" PRId64 ", baseValue = %p\n",
4823                             reloc->r_length, thing, (char *)baseValue));
4824
4825         if (type == X86_64_RELOC_GOT
4826            || type == X86_64_RELOC_GOT_LOAD)
4827         {
4828             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4829             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4830
4831             IF_DEBUG(linker, debugBelch("relocateSection: making jump island for %s, extern = %d, X86_64_RELOC_GOT\n", nm, reloc->r_extern));
4832             ASSERT(reloc->r_extern);
4833             value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, (unsigned long)lookupSymbol(nm))->addr;
4834
4835             type = X86_64_RELOC_SIGNED;
4836         }
4837         else if(reloc->r_extern)
4838         {
4839             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4840             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4841
4842             IF_DEBUG(linker, debugBelch("relocateSection: looking up external symbol %s\n", nm));
4843             IF_DEBUG(linker, debugBelch("               : type  = %d\n", symbol->n_type));
4844             IF_DEBUG(linker, debugBelch("               : sect  = %d\n", symbol->n_sect));
4845             IF_DEBUG(linker, debugBelch("               : desc  = %d\n", symbol->n_desc));
4846             IF_DEBUG(linker, debugBelch("               : value = %p\n", (void *)symbol->n_value));
4847             if ((symbol->n_type & N_TYPE) == N_SECT) {
4848                 value = relocateAddress(oc, nSections, sections,
4849                                         symbol->n_value);
4850                 IF_DEBUG(linker, debugBelch("relocateSection, defined external symbol %s, relocated address %p\n", nm, (void *)value));
4851             }
4852             else {
4853                 value = (uint64_t) lookupSymbol(nm);
4854                 IF_DEBUG(linker, debugBelch("relocateSection: external symbol %s, address %p\n", nm, (void *)value));
4855             }
4856         }
4857         else
4858         {
4859             // If the relocation is not through the global offset table
4860             // or external, then set the value to the baseValue.  This
4861             // will leave displacements into the __const section
4862             // unchanged (as they ought to be).
4863
4864             value = baseValue;
4865         }
4866
4867         IF_DEBUG(linker, debugBelch("relocateSection: value = %p\n", (void *)value));
4868
4869         if (type == X86_64_RELOC_BRANCH)
4870         {
4871             if((int32_t)(value - baseValue) != (int64_t)(value - baseValue))
4872             {
4873                 ASSERT(reloc->r_extern);
4874                 value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)
4875                                         -> jumpIsland;
4876             }
4877             ASSERT((int32_t)(value - baseValue) == (int64_t)(value - baseValue));
4878             type = X86_64_RELOC_SIGNED;
4879         }
4880
4881         switch(type)
4882         {
4883             case X86_64_RELOC_UNSIGNED:
4884                 ASSERT(!reloc->r_pcrel);
4885                 thing += value;
4886                 break;
4887             case X86_64_RELOC_SIGNED:
4888             case X86_64_RELOC_SIGNED_1:
4889             case X86_64_RELOC_SIGNED_2:
4890             case X86_64_RELOC_SIGNED_4:
4891                 ASSERT(reloc->r_pcrel);
4892                 thing += value - baseValue;
4893                 break;
4894             case X86_64_RELOC_SUBTRACTOR:
4895                 ASSERT(!reloc->r_pcrel);
4896                 thing -= value;
4897                 break;
4898             default:
4899                 barf("unkown relocation");
4900         }
4901
4902         switch(reloc->r_length)
4903         {
4904             case 0:
4905                 *(uint8_t*)thingPtr = thing;
4906                 break;
4907             case 1:
4908                 *(uint16_t*)thingPtr = thing;
4909                 break;
4910             case 2:
4911                 *(uint32_t*)thingPtr = thing;
4912                 break;
4913             case 3:
4914                 *(uint64_t*)thingPtr = thing;
4915                 break;
4916         }
4917 #else
4918         if(relocs[i].r_address & R_SCATTERED)
4919         {
4920             struct scattered_relocation_info *scat =
4921                 (struct scattered_relocation_info*) &relocs[i];
4922
4923             if(!scat->r_pcrel)
4924             {
4925                 if(scat->r_length == 2)
4926                 {
4927                     unsigned long word = 0;
4928                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
4929                     checkProddableBlock(oc,wordPtr);
4930
4931                     // Note on relocation types:
4932                     // i386 uses the GENERIC_RELOC_* types,
4933                     // while ppc uses special PPC_RELOC_* types.
4934                     // *_RELOC_VANILLA and *_RELOC_PAIR have the same value
4935                     // in both cases, all others are different.
4936                     // Therefore, we use GENERIC_RELOC_VANILLA
4937                     // and GENERIC_RELOC_PAIR instead of the PPC variants,
4938                     // and use #ifdefs for the other types.
4939
4940                     // Step 1: Figure out what the relocated value should be
4941                     if (scat->r_type == GENERIC_RELOC_VANILLA) {
4942                         word = *wordPtr
4943                              + (unsigned long) relocateAddress(oc,
4944                                                                 nSections,
4945                                                                 sections,
4946                                                                 scat->r_value)
4947                                         - scat->r_value;
4948                     }
4949 #ifdef powerpc_HOST_ARCH
4950                     else if(scat->r_type == PPC_RELOC_SECTDIFF
4951                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
4952                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
4953                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF
4954                         || scat->r_type == PPC_RELOC_LOCAL_SECTDIFF)
4955 #else
4956                     else if(scat->r_type == GENERIC_RELOC_SECTDIFF
4957                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
4958 #endif
4959                     {
4960                         struct scattered_relocation_info *pair =
4961                                 (struct scattered_relocation_info*) &relocs[i+1];
4962
4963                         if (!pair->r_scattered || pair->r_type != GENERIC_RELOC_PAIR) {
4964                             barf("Invalid Mach-O file: "
4965                                  "RELOC_*_SECTDIFF not followed by RELOC_PAIR");
4966                         }
4967
4968                         word = (unsigned long)
4969                                (relocateAddress(oc, nSections, sections, scat->r_value)
4970                               - relocateAddress(oc, nSections, sections, pair->r_value));
4971                         i++;
4972                     }
4973 #ifdef powerpc_HOST_ARCH
4974                     else if(scat->r_type == PPC_RELOC_HI16
4975                          || scat->r_type == PPC_RELOC_LO16
4976                          || scat->r_type == PPC_RELOC_HA16
4977                          || scat->r_type == PPC_RELOC_LO14)
4978                     {   // these are generated by label+offset things
4979                         struct relocation_info *pair = &relocs[i+1];
4980
4981                         if ((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR) {
4982                             barf("Invalid Mach-O file: "
4983                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
4984                         }
4985
4986                         if(scat->r_type == PPC_RELOC_LO16)
4987                         {
4988                             word = ((unsigned short*) wordPtr)[1];
4989                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4990                         }
4991                         else if(scat->r_type == PPC_RELOC_LO14)
4992                         {
4993                             barf("Unsupported Relocation: PPC_RELOC_LO14");
4994                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
4995                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4996                         }
4997                         else if(scat->r_type == PPC_RELOC_HI16)
4998                         {
4999                             word = ((unsigned short*) wordPtr)[1] << 16;
5000                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
5001                         }
5002                         else if(scat->r_type == PPC_RELOC_HA16)
5003                         {
5004                             word = ((unsigned short*) wordPtr)[1] << 16;
5005                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
5006                         }
5007
5008
5009                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
5010                                                 - scat->r_value;
5011
5012                         i++;
5013                     }
5014  #endif
5015                     else {
5016                         barf ("Don't know how to handle this Mach-O "
5017                               "scattered relocation entry: "
5018                               "object file %s; entry type %ld; "
5019                               "address %#lx\n",
5020                               OC_INFORMATIVE_FILENAME(oc),
5021                               scat->r_type,
5022                               scat->r_address);
5023                         return 0;
5024                      }
5025
5026 #ifdef powerpc_HOST_ARCH
5027                     if(scat->r_type == GENERIC_RELOC_VANILLA
5028                         || scat->r_type == PPC_RELOC_SECTDIFF)
5029 #else
5030                     if(scat->r_type == GENERIC_RELOC_VANILLA
5031                         || scat->r_type == GENERIC_RELOC_SECTDIFF
5032                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
5033 #endif
5034                     {
5035                         *wordPtr = word;
5036                     }
5037 #ifdef powerpc_HOST_ARCH
5038                     else if (scat->r_type == PPC_RELOC_LO16_SECTDIFF
5039                           || scat->r_type == PPC_RELOC_LO16)
5040                     {
5041                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
5042                     }
5043                     else if (scat->r_type == PPC_RELOC_HI16_SECTDIFF
5044                           || scat->r_type == PPC_RELOC_HI16)
5045                     {
5046                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
5047                     }
5048                     else if (scat->r_type == PPC_RELOC_HA16_SECTDIFF
5049                           || scat->r_type == PPC_RELOC_HA16)
5050                     {
5051                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
5052                             + ((word & (1<<15)) ? 1 : 0);
5053                     }
5054 #endif
5055                 }
5056                 else
5057                 {
5058                     barf("Can't handle Mach-O scattered relocation entry "
5059                          "with this r_length tag: "
5060                          "object file %s; entry type %ld; "
5061                          "r_length tag %ld; address %#lx\n",
5062                          OC_INFORMATIVE_FILENAME(oc),
5063                          scat->r_type,
5064                          scat->r_length,
5065                          scat->r_address);
5066                     return 0;
5067                 }
5068             }
5069             else /* scat->r_pcrel */
5070             {
5071                 barf("Don't know how to handle *PC-relative* Mach-O "
5072                      "scattered relocation entry: "
5073                      "object file %s; entry type %ld; address %#lx\n",
5074                      OC_INFORMATIVE_FILENAME(oc),
5075                      scat->r_type,
5076                      scat->r_address);
5077                return 0;
5078             }
5079
5080         }
5081         else /* !(relocs[i].r_address & R_SCATTERED) */
5082         {
5083             struct relocation_info *reloc = &relocs[i];
5084             if (reloc->r_pcrel && !reloc->r_extern) {
5085                 IF_DEBUG(linker, debugBelch("relocateSection: pc relative but not external, skipping\n"));
5086                 continue;
5087             }
5088
5089             if (reloc->r_length == 2) {
5090                 unsigned long word = 0;
5091 #ifdef powerpc_HOST_ARCH
5092                 unsigned long jumpIsland = 0;
5093                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
5094                                                       // to avoid warning and to catch
5095                                                       // bugs.
5096 #endif
5097
5098                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
5099                 checkProddableBlock(oc,wordPtr);
5100
5101                 if (reloc->r_type == GENERIC_RELOC_VANILLA) {
5102                     word = *wordPtr;
5103                 }
5104 #ifdef powerpc_HOST_ARCH
5105                 else if (reloc->r_type == PPC_RELOC_LO16) {
5106                     word = ((unsigned short*) wordPtr)[1];
5107                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
5108                 }
5109                 else if (reloc->r_type == PPC_RELOC_HI16) {
5110                     word = ((unsigned short*) wordPtr)[1] << 16;
5111                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
5112                 }
5113                 else if (reloc->r_type == PPC_RELOC_HA16) {
5114                     word = ((unsigned short*) wordPtr)[1] << 16;
5115                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
5116                 }
5117                 else if (reloc->r_type == PPC_RELOC_BR24) {
5118                     word = *wordPtr;
5119                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
5120                 }
5121 #endif
5122                 else {
5123                     barf("Can't handle this Mach-O relocation entry "
5124                          "(not scattered): "
5125                          "object file %s; entry type %ld; address %#lx\n",
5126                          OC_INFORMATIVE_FILENAME(oc),
5127                          reloc->r_type,
5128                          reloc->r_address);
5129                     return 0;
5130                 }
5131
5132                 if (!reloc->r_extern) {
5133                     long delta = sections[reloc->r_symbolnum-1].offset
5134                         - sections[reloc->r_symbolnum-1].addr
5135                         + ((long) image);
5136
5137                     word += delta;
5138                 }
5139                 else {
5140                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
5141                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
5142                     void *symbolAddress = lookupSymbol(nm);
5143
5144                     if (!symbolAddress) {
5145                         errorBelch("\nunknown symbol `%s'", nm);
5146                         return 0;
5147                     }
5148
5149                     if (reloc->r_pcrel) {
5150 #ifdef powerpc_HOST_ARCH
5151                             // In the .o file, this should be a relative jump to NULL
5152                             // and we'll change it to a relative jump to the symbol
5153                         ASSERT(word + reloc->r_address == 0);
5154                         jumpIsland = (unsigned long)
5155                                         &makeSymbolExtra(oc,
5156                                                          reloc->r_symbolnum,
5157                                                          (unsigned long) symbolAddress)
5158                                          -> jumpIsland;
5159                         if (jumpIsland != 0) {
5160                             offsetToJumpIsland = word + jumpIsland
5161                                 - (((long)image) + sect->offset - sect->addr);
5162                         }
5163 #endif
5164                         word += (unsigned long) symbolAddress
5165                                 - (((long)image) + sect->offset - sect->addr);
5166                     }
5167                     else {
5168                         word += (unsigned long) symbolAddress;
5169                     }
5170                 }
5171
5172                 if (reloc->r_type == GENERIC_RELOC_VANILLA) {
5173                     *wordPtr = word;
5174                     continue;
5175                 }
5176 #ifdef powerpc_HOST_ARCH
5177                 else if(reloc->r_type == PPC_RELOC_LO16)
5178                 {
5179                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
5180                     i++;
5181                     continue;
5182                 }
5183                 else if(reloc->r_type == PPC_RELOC_HI16)
5184                 {
5185                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
5186                     i++;
5187                     continue;
5188                 }
5189                 else if(reloc->r_type == PPC_RELOC_HA16)
5190                 {
5191                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
5192                         + ((word & (1<<15)) ? 1 : 0);
5193                     i++;
5194                     continue;
5195                 }
5196                 else if(reloc->r_type == PPC_RELOC_BR24)
5197                 {
5198                     if ((word & 0x03) != 0) {
5199                         barf("%s: unconditional relative branch with a displacement "
5200                              "which isn't a multiple of 4 bytes: %#lx",
5201                              OC_INFORMATIVE_FILENAME(oc),
5202                              word);
5203                     }
5204
5205                     if((word & 0xFE000000) != 0xFE000000 &&
5206                         (word & 0xFE000000) != 0x00000000) {
5207                         // The branch offset is too large.
5208                         // Therefore, we try to use a jump island.
5209                         if (jumpIsland == 0) {
5210                             barf("%s: unconditional relative branch out of range: "
5211                                  "no jump island available: %#lx",
5212                                  OC_INFORMATIVE_FILENAME(oc),
5213                                  word);
5214                         }
5215
5216                         word = offsetToJumpIsland;
5217
5218                         if((word & 0xFE000000) != 0xFE000000 &&
5219                             (word & 0xFE000000) != 0x00000000) {
5220                             barf("%s: unconditional relative branch out of range: "
5221                                  "jump island out of range: %#lx",
5222                                  OC_INFORMATIVE_FILENAME(oc),
5223                                  word);
5224                     }
5225                     }
5226                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
5227                     continue;
5228                 }
5229 #endif
5230             }
5231             else
5232             {
5233                  barf("Can't handle Mach-O relocation entry (not scattered) "
5234                       "with this r_length tag: "
5235                       "object file %s; entry type %ld; "
5236                       "r_length tag %ld; address %#lx\n",
5237                       OC_INFORMATIVE_FILENAME(oc),
5238                       reloc->r_type,
5239                       reloc->r_length,
5240                       reloc->r_address);
5241                  return 0;
5242             }
5243         }
5244 #endif
5245     }
5246
5247     IF_DEBUG(linker, debugBelch("relocateSection: done\n"));
5248     return 1;
5249 }
5250
5251 static int
5252 ocGetNames_MachO(ObjectCode* oc)
5253 {
5254     char *image = (char*) oc->image;
5255     struct mach_header *header = (struct mach_header*) image;
5256     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
5257     unsigned i,curSymbol = 0;
5258     struct segment_command *segLC = NULL;
5259     struct section *sections;
5260     struct symtab_command *symLC = NULL;
5261     struct nlist *nlist;
5262     unsigned long commonSize = 0;
5263     char    *commonStorage = NULL;
5264     unsigned long commonCounter;
5265
5266     IF_DEBUG(linker,debugBelch("ocGetNames_MachO: start\n"));
5267
5268     for(i=0;i<header->ncmds;i++)
5269     {
5270         if (lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64) {
5271             segLC = (struct segment_command*) lc;
5272         }
5273         else if (lc->cmd == LC_SYMTAB) {
5274             symLC = (struct symtab_command*) lc;
5275         }
5276
5277         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
5278     }
5279
5280     sections = (struct section*) (segLC+1);
5281     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
5282                   : NULL;
5283
5284     if (!segLC) {
5285         barf("ocGetNames_MachO: no segment load command");
5286     }
5287
5288     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: will load %d sections\n", segLC->nsects));
5289     for(i=0;i<segLC->nsects;i++)
5290     {
5291         IF_DEBUG(linker, debugBelch("ocGetNames_MachO: section %d\n", i));
5292
5293         if (sections[i].size == 0) {
5294             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: found a zero length section, skipping\n"));
5295             continue;
5296         }
5297
5298         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
5299         {
5300             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
5301                                       "ocGetNames_MachO(common symbols)");
5302             sections[i].offset = zeroFillArea - image;
5303         }
5304
5305         if (!strcmp(sections[i].sectname,"__text")) {
5306
5307             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __text section\n"));
5308             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
5309                 (void*) (image + sections[i].offset),
5310                 (void*) (image + sections[i].offset + sections[i].size));
5311         }
5312         else if (!strcmp(sections[i].sectname,"__const")) {
5313
5314             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __const section\n"));
5315             addSection(oc, SECTIONKIND_RWDATA,
5316                 (void*) (image + sections[i].offset),
5317                 (void*) (image + sections[i].offset + sections[i].size));
5318         }
5319         else if (!strcmp(sections[i].sectname,"__data")) {
5320
5321             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __data section\n"));
5322             addSection(oc, SECTIONKIND_RWDATA,
5323                 (void*) (image + sections[i].offset),
5324                 (void*) (image + sections[i].offset + sections[i].size));
5325         }
5326         else if(!strcmp(sections[i].sectname,"__bss")
5327                 || !strcmp(sections[i].sectname,"__common")) {
5328
5329             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __bss section\n"));
5330             addSection(oc, SECTIONKIND_RWDATA,
5331                 (void*) (image + sections[i].offset),
5332                 (void*) (image + sections[i].offset + sections[i].size));
5333         }
5334         addProddableBlock(oc,
5335                           (void *) (image + sections[i].offset),
5336                                         sections[i].size);
5337     }
5338
5339         // count external symbols defined here
5340     oc->n_symbols = 0;
5341     if (symLC) {
5342         for (i = 0; i < symLC->nsyms; i++) {
5343             if (nlist[i].n_type & N_STAB) {
5344                 ;
5345             }
5346             else if(nlist[i].n_type & N_EXT)
5347             {
5348                 if((nlist[i].n_type & N_TYPE) == N_UNDF
5349                     && (nlist[i].n_value != 0))
5350                 {
5351                     commonSize += nlist[i].n_value;
5352                     oc->n_symbols++;
5353                 }
5354                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
5355                     oc->n_symbols++;
5356             }
5357         }
5358     }
5359     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: %d external symbols\n", oc->n_symbols));
5360     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
5361                                    "ocGetNames_MachO(oc->symbols)");
5362
5363     if(symLC)
5364     {
5365         for(i=0;i<symLC->nsyms;i++)
5366         {
5367             if(nlist[i].n_type & N_STAB)
5368                 ;
5369             else if((nlist[i].n_type & N_TYPE) == N_SECT)
5370             {
5371                 if(nlist[i].n_type & N_EXT)
5372                 {
5373                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
5374                     if ((nlist[i].n_desc & N_WEAK_DEF) && lookupSymbol(nm)) {
5375                         // weak definition, and we already have a definition
5376                         IF_DEBUG(linker, debugBelch("    weak: %s\n", nm));
5377                     }
5378                     else
5379                     {
5380                             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: inserting %s\n", nm));
5381                             ghciInsertStrHashTable(oc->fileName, symhash, nm,
5382                                                     image
5383                                                     + sections[nlist[i].n_sect-1].offset
5384                                                     - sections[nlist[i].n_sect-1].addr
5385                                                     + nlist[i].n_value);
5386                             oc->symbols[curSymbol++] = nm;
5387                     }
5388                 }
5389                 else
5390                 {
5391                     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: \t...not external, skipping\n"));
5392                 }
5393             }
5394             else
5395             {
5396                 IF_DEBUG(linker, debugBelch("ocGetNames_MachO: \t...not defined in this section, skipping\n"));
5397             }
5398         }
5399     }
5400
5401     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
5402     commonCounter = (unsigned long)commonStorage;
5403
5404     if (symLC) {
5405         for (i = 0; i < symLC->nsyms; i++) {
5406             if((nlist[i].n_type & N_TYPE) == N_UNDF
5407              && (nlist[i].n_type & N_EXT)
5408              && (nlist[i].n_value != 0)) {
5409
5410                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
5411                 unsigned long sz = nlist[i].n_value;
5412
5413                 nlist[i].n_value = commonCounter;
5414
5415                 IF_DEBUG(linker, debugBelch("ocGetNames_MachO: inserting common symbol: %s\n", nm));
5416                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
5417                                        (void*)commonCounter);
5418                 oc->symbols[curSymbol++] = nm;
5419
5420                 commonCounter += sz;
5421             }
5422         }
5423     }
5424
5425     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: done\n"));
5426     return 1;
5427 }
5428
5429 static int
5430 ocResolve_MachO(ObjectCode* oc)
5431 {
5432     char *image = (char*) oc->image;
5433     struct mach_header *header = (struct mach_header*) image;
5434     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
5435     unsigned i;
5436     struct segment_command *segLC = NULL;
5437     struct section *sections;
5438     struct symtab_command *symLC = NULL;
5439     struct dysymtab_command *dsymLC = NULL;
5440     struct nlist *nlist;
5441
5442     IF_DEBUG(linker, debugBelch("ocResolve_MachO: start\n"));
5443     for (i = 0; i < header->ncmds; i++)
5444     {
5445         if (lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64) {
5446             segLC = (struct segment_command*) lc;
5447             IF_DEBUG(linker, debugBelch("ocResolve_MachO: found a 32 or 64 bit segment load command\n"));
5448         }
5449         else if (lc->cmd == LC_SYMTAB) {
5450             symLC = (struct symtab_command*) lc;
5451             IF_DEBUG(linker, debugBelch("ocResolve_MachO: found a symbol table load command\n"));
5452         }
5453         else if (lc->cmd == LC_DYSYMTAB) {
5454             dsymLC = (struct dysymtab_command*) lc;
5455             IF_DEBUG(linker, debugBelch("ocResolve_MachO: found a dynamic symbol table load command\n"));
5456         }
5457
5458         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
5459     }
5460
5461     sections = (struct section*) (segLC+1);
5462     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
5463                   : NULL;
5464
5465     if(dsymLC)
5466     {
5467         unsigned long *indirectSyms
5468             = (unsigned long*) (image + dsymLC->indirectsymoff);
5469
5470         IF_DEBUG(linker, debugBelch("ocResolve_MachO: resolving dsymLC\n"));
5471         for (i = 0; i < segLC->nsects; i++)
5472         {
5473             if(    !strcmp(sections[i].sectname,"__la_symbol_ptr")
5474                 || !strcmp(sections[i].sectname,"__la_sym_ptr2")
5475                 || !strcmp(sections[i].sectname,"__la_sym_ptr3"))
5476             {
5477                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5478                     return 0;
5479             }
5480             else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr")
5481                 ||  !strcmp(sections[i].sectname,"__pointers"))
5482             {
5483                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5484                     return 0;
5485             }
5486             else if(!strcmp(sections[i].sectname,"__jump_table"))
5487             {
5488                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5489                     return 0;
5490             }
5491             else
5492             {
5493                 IF_DEBUG(linker, debugBelch("ocResolve_MachO: unknown section\n"));
5494             }
5495         }
5496     }
5497
5498     for(i=0;i<segLC->nsects;i++)
5499     {
5500             IF_DEBUG(linker, debugBelch("ocResolve_MachO: relocating section %d\n", i));
5501
5502         if (!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
5503             return 0;
5504     }
5505
5506 #if defined (powerpc_HOST_ARCH)
5507     ocFlushInstructionCache( oc );
5508 #endif
5509
5510     return 1;
5511 }
5512
5513 #ifdef powerpc_HOST_ARCH
5514 /*
5515  * The Mach-O object format uses leading underscores. But not everywhere.
5516  * There is a small number of runtime support functions defined in
5517  * libcc_dynamic.a whose name does not have a leading underscore.
5518  * As a consequence, we can't get their address from C code.
5519  * We have to use inline assembler just to take the address of a function.
5520  * Yuck.
5521  */
5522
5523 extern void* symbolsWithoutUnderscore[];
5524
5525 static void
5526 machoInitSymbolsWithoutUnderscore(void)
5527 {
5528     void **p = symbolsWithoutUnderscore;
5529     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
5530
5531 #undef SymI_NeedsProto
5532 #define SymI_NeedsProto(x)  \
5533     __asm__ volatile(".long " # x);
5534
5535     RTS_MACHO_NOUNDERLINE_SYMBOLS
5536
5537     __asm__ volatile(".text");
5538
5539 #undef SymI_NeedsProto
5540 #define SymI_NeedsProto(x)  \
5541     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
5542
5543     RTS_MACHO_NOUNDERLINE_SYMBOLS
5544
5545 #undef SymI_NeedsProto
5546 }
5547 #endif
5548
5549 #ifndef USE_MMAP
5550 /*
5551  * Figure out by how much to shift the entire Mach-O file in memory
5552  * when loading so that its single segment ends up 16-byte-aligned
5553  */
5554 static int
5555 machoGetMisalignment( FILE * f )
5556 {
5557     struct mach_header header;
5558     int misalignment;
5559
5560     {
5561         int n = fread(&header, sizeof(header), 1, f);
5562         if (n != 1) {
5563             barf("machoGetMisalignment: can't read the Mach-O header");
5564         }
5565     }
5566     fseek(f, -sizeof(header), SEEK_CUR);
5567
5568 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
5569     if(header.magic != MH_MAGIC_64) {
5570         barf("Bad magic. Expected: %08x, got: %08x.",
5571              MH_MAGIC_64, header.magic);
5572     }
5573 #else
5574     if(header.magic != MH_MAGIC) {
5575         barf("Bad magic. Expected: %08x, got: %08x.",
5576              MH_MAGIC, header.magic);
5577     }
5578 #endif
5579
5580     misalignment = (header.sizeofcmds + sizeof(header))
5581                     & 0xF;
5582
5583     return misalignment ? (16 - misalignment) : 0;
5584 }
5585 #endif
5586
5587 #endif