rts/Linker.c

   1 /* -----------------------------------------------------------------------------
   2  *
   3  * (c) The GHC Team, 2000-2004
   4  *
   5  * RTS Object Linker
   6  *
   7  * ---------------------------------------------------------------------------*/
   8
   9 #if 0
  10 #include "PosixSource.h"
  11 #endif
  12
  13 /* Linux needs _GNU_SOURCE to get RTLD_DEFAULT from <dlfcn.h> and
  14    MREMAP_MAYMOVE from <sys/mman.h>.
  15  */
  16 #ifdef __linux__
  17 #define _GNU_SOURCE
  18 #endif
  19
  20 #include "Rts.h"
  21 #include "HsFFI.h"
  22
  23 #include "sm/Storage.h"
  24 #include "Stats.h"
  25 #include "Hash.h"
  26 #include "LinkerInternals.h"
  27 #include "RtsUtils.h"
  28 #include "Trace.h"
  29 #include "StgPrimFloat.h" // for __int_encodeFloat etc.
  30 #include "Stable.h"
  31
  32 #if !defined(mingw32_HOST_OS)
  33 #include "posix/Signals.h"
  34 #endif
  35
  36 #if defined(mingw32_HOST_OS)
  37 // get protos for is*()
  38 #include <ctype.h>
  39 #endif
  40
  41 #ifdef HAVE_SYS_TYPES_H
  42 #include <sys/types.h>
  43 #endif
  44
  45 #include <stdlib.h>
  46 #include <string.h>
  47 #include <stdio.h>
  48 #include <assert.h>
  49
  50 #ifdef HAVE_SYS_STAT_H
  51 #include <sys/stat.h>
  52 #endif
  53
  54 #if defined(HAVE_DLFCN_H)
  55 #include <dlfcn.h>
  56 #endif
  57
  58 #if defined(cygwin32_HOST_OS)
  59 #ifdef HAVE_DIRENT_H
  60 #include <dirent.h>
  61 #endif
  62
  63 #ifdef HAVE_SYS_TIME_H
  64 #include <sys/time.h>
  65 #endif
  66 #include <regex.h>
  67 #include <sys/fcntl.h>
  68 #include <sys/termios.h>
  69 #include <sys/utime.h>
  70 #include <sys/utsname.h>
  71 #include <sys/wait.h>
  72 #endif
  73
  74 #if defined(linux_HOST_OS) || defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS) || defined(darwin_HOST_OS)
  75 #define USE_MMAP
  76 #include <fcntl.h>
  77 #include <sys/mman.h>
  78
  79 #ifdef HAVE_UNISTD_H
  80 #include <unistd.h>
  81 #endif
  82
  83 #endif
  84
  85 #if defined(linux_HOST_OS) || defined(solaris2_HOST_OS) || defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  86 #  define OBJFORMAT_ELF
  87 #  include <regex.h>    // regex is already used by dlopen() so this is OK
  88                         // to use here without requiring an additional lib
  89 #elif defined(cygwin32_HOST_OS) || defined (mingw32_HOST_OS)
  90 #  define OBJFORMAT_PEi386
  91 #  include <windows.h>
  92 #  include <math.h>
  93 #elif defined(darwin_HOST_OS)
  94 #  define OBJFORMAT_MACHO
  95 #  include <regex.h>
  96 #  include <mach-o/loader.h>
  97 #  include <mach-o/nlist.h>
  98 #  include <mach-o/reloc.h>
  99 #if !defined(HAVE_DLFCN_H)
 100 #  include <mach-o/dyld.h>
 101 #endif
 102 #if defined(powerpc_HOST_ARCH)
 103 #  include <mach-o/ppc/reloc.h>
 104 #endif
 105 #if defined(x86_64_HOST_ARCH)
 106 #  include <mach-o/x86_64/reloc.h>
 107 #endif
 108 #endif
 109
 110 #if defined(x86_64_HOST_ARCH) && defined(darwin_HOST_OS)
 111 #define ALWAYS_PIC
 112 #endif
 113
 114 /* Hash table mapping symbol names to Symbol */
 115 static /*Str*/HashTable *symhash;
 116
 117 /* Hash table mapping symbol names to StgStablePtr */
 118 static /*Str*/HashTable *stablehash;
 119
 120 /* List of currently loaded objects */
 121 ObjectCode *objects = NULL;     /* initially empty */
 122
 123 #if defined(OBJFORMAT_ELF)
 124 static int ocVerifyImage_ELF    ( ObjectCode* oc );
 125 static int ocGetNames_ELF       ( ObjectCode* oc );
 126 static int ocResolve_ELF        ( ObjectCode* oc );
 127 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 128 static int ocAllocateSymbolExtras_ELF ( ObjectCode* oc );
 129 #endif
 130 #elif defined(OBJFORMAT_PEi386)
 131 static int ocVerifyImage_PEi386 ( ObjectCode* oc );
 132 static int ocGetNames_PEi386    ( ObjectCode* oc );
 133 static int ocResolve_PEi386     ( ObjectCode* oc );
 134 static void *lookupSymbolInDLLs ( unsigned char *lbl );
 135 static void zapTrailingAtSign   ( unsigned char *sym );
 136 #elif defined(OBJFORMAT_MACHO)
 137 static int ocVerifyImage_MachO    ( ObjectCode* oc );
 138 static int ocGetNames_MachO       ( ObjectCode* oc );
 139 static int ocResolve_MachO        ( ObjectCode* oc );
 140
 141 #ifndef USE_MMAP
 142 static int machoGetMisalignment( FILE * );
 143 #endif
 144 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 145 static int ocAllocateSymbolExtras_MachO ( ObjectCode* oc );
 146 #endif
 147 #ifdef powerpc_HOST_ARCH
 148 static void machoInitSymbolsWithoutUnderscore( void );
 149 #endif
 150 #endif
 151
 152 /* on x86_64 we have a problem with relocating symbol references in
 153  * code that was compiled without -fPIC.  By default, the small memory
 154  * model is used, which assumes that symbol references can fit in a
 155  * 32-bit slot.  The system dynamic linker makes this work for
 156  * references to shared libraries by either (a) allocating a jump
 157  * table slot for code references, or (b) moving the symbol at load
 158  * time (and copying its contents, if necessary) for data references.
 159  *
 160  * We unfortunately can't tell whether symbol references are to code
 161  * or data.  So for now we assume they are code (the vast majority
 162  * are), and allocate jump-table slots.  Unfortunately this will
 163  * SILENTLY generate crashing code for data references.  This hack is
 164  * enabled by X86_64_ELF_NONPIC_HACK.
 165  *
 166  * One workaround is to use shared Haskell libraries.  This is
 167  * coming.  Another workaround is to keep the static libraries but
 168  * compile them with -fPIC, because that will generate PIC references
 169  * to data which can be relocated.  The PIC code is still too green to
 170  * do this systematically, though.
 171  *
 172  * See bug #781
 173  * See thread http://www.haskell.org/pipermail/cvs-ghc/2007-September/038458.html
 174  *
 175  * Naming Scheme for Symbol Macros
 176  *
 177  * SymI_*: symbol is internal to the RTS. It resides in an object
 178  *         file/library that is statically.
 179  * SymE_*: symbol is external to the RTS library. It might be linked
 180  *         dynamically.
 181  *
 182  * Sym*_HasProto  : the symbol prototype is imported in an include file
 183  *                  or defined explicitly
 184  * Sym*_NeedsProto: the symbol is undefined and we add a dummy
 185  *                  default proto extern void sym(void);
 186  */
 187 #define X86_64_ELF_NONPIC_HACK 1
 188
 189 /* Link objects into the lower 2Gb on x86_64.  GHC assumes the
 190  * small memory model on this architecture (see gcc docs,
 191  * -mcmodel=small).
 192  *
 193  * MAP_32BIT not available on OpenBSD/amd64
 194  */
 195 #if defined(x86_64_HOST_ARCH) && defined(MAP_32BIT)
 196 #define TRY_MAP_32BIT MAP_32BIT
 197 #else
 198 #define TRY_MAP_32BIT 0
 199 #endif
 200
 201 /*
 202  * Due to the small memory model (see above), on x86_64 we have to map
 203  * all our non-PIC object files into the low 2Gb of the address space
 204  * (why 2Gb and not 4Gb?  Because all addresses must be reachable
 205  * using a 32-bit signed PC-relative offset). On Linux we can do this
 206  * using the MAP_32BIT flag to mmap(), however on other OSs
 207  * (e.g. *BSD, see #2063, and also on Linux inside Xen, see #2512), we
 208  * can't do this.  So on these systems, we have to pick a base address
 209  * in the low 2Gb of the address space and try to allocate memory from
 210  * there.
 211  *
 212  * We pick a default address based on the OS, but also make this
 213  * configurable via an RTS flag (+RTS -xm)
 214  */
 215 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
 216
 217 #if defined(MAP_32BIT)
 218 // Try to use MAP_32BIT
 219 #define MMAP_32BIT_BASE_DEFAULT 0
 220 #else
 221 // A guess: 1Gb.
 222 #define MMAP_32BIT_BASE_DEFAULT 0x40000000
 223 #endif
 224
 225 static void *mmap_32bit_base = (void *)MMAP_32BIT_BASE_DEFAULT;
 226 #endif
 227
 228 /* MAP_ANONYMOUS is MAP_ANON on some systems, e.g. OpenBSD */
 229 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
 230 #define MAP_ANONYMOUS MAP_ANON
 231 #endif
 232
 233 /* -----------------------------------------------------------------------------
 234  * Built-in symbols from the RTS
 235  */
 236
 237 typedef struct _RtsSymbolVal {
 238     char   *lbl;
 239     void   *addr;
 240 } RtsSymbolVal;
 241
 242 #define Maybe_Stable_Names      SymI_HasProto(stg_mkWeakzh)                     \
 243                                 SymI_HasProto(stg_mkWeakForeignEnvzh)           \
 244                                 SymI_HasProto(stg_makeStableNamezh)             \
 245                                 SymI_HasProto(stg_finalizzeWeakzh)
 246
 247 #if !defined (mingw32_HOST_OS)
 248 #define RTS_POSIX_ONLY_SYMBOLS                  \
 249       SymI_HasProto(__hscore_get_saved_termios) \
 250       SymI_HasProto(__hscore_set_saved_termios) \
 251       SymI_HasProto(shutdownHaskellAndSignal)   \
 252       SymI_HasProto(lockFile)                   \
 253       SymI_HasProto(unlockFile)                 \
 254       SymI_HasProto(signal_handlers)            \
 255       SymI_HasProto(stg_sig_install)            \
 256       SymI_NeedsProto(nocldstop)
 257 #endif
 258
 259 #if defined (cygwin32_HOST_OS)
 260 #define RTS_MINGW_ONLY_SYMBOLS /**/
 261 /* Don't have the ability to read import libs / archives, so
 262  * we have to stupidly list a lot of what libcygwin.a
 263  * exports; sigh.
 264  */
 265 #define RTS_CYGWIN_ONLY_SYMBOLS                          \
 266       SymI_HasProto(regfree)                             \
 267       SymI_HasProto(regexec)                             \
 268       SymI_HasProto(regerror)                            \
 269       SymI_HasProto(regcomp)                             \
 270       SymI_HasProto(__errno)                             \
 271       SymI_HasProto(access)                              \
 272       SymI_HasProto(chmod)                               \
 273       SymI_HasProto(chdir)                               \
 274       SymI_HasProto(close)                               \
 275       SymI_HasProto(creat)                               \
 276       SymI_HasProto(dup)                                 \
 277       SymI_HasProto(dup2)                                \
 278       SymI_HasProto(fstat)                               \
 279       SymI_HasProto(fcntl)                               \
 280       SymI_HasProto(getcwd)                              \
 281       SymI_HasProto(getenv)                              \
 282       SymI_HasProto(lseek)                               \
 283       SymI_HasProto(open)                                \
 284       SymI_HasProto(fpathconf)                           \
 285       SymI_HasProto(pathconf)                            \
 286       SymI_HasProto(stat)                                \
 287       SymI_HasProto(pow)                                 \
 288       SymI_HasProto(tanh)                                \
 289       SymI_HasProto(cosh)                                \
 290       SymI_HasProto(sinh)                                \
 291       SymI_HasProto(atan)                                \
 292       SymI_HasProto(acos)                                \
 293       SymI_HasProto(asin)                                \
 294       SymI_HasProto(tan)                                 \
 295       SymI_HasProto(cos)                                 \
 296       SymI_HasProto(sin)                                 \
 297       SymI_HasProto(exp)                                 \
 298       SymI_HasProto(log)                                 \
 299       SymI_HasProto(sqrt)                                \
 300       SymI_HasProto(localtime_r)                         \
 301       SymI_HasProto(gmtime_r)                            \
 302       SymI_HasProto(mktime)                              \
 303       SymI_NeedsProto(_imp___tzname)                     \
 304       SymI_HasProto(gettimeofday)                        \
 305       SymI_HasProto(timezone)                            \
 306       SymI_HasProto(tcgetattr)                           \
 307       SymI_HasProto(tcsetattr)                           \
 308       SymI_HasProto(memcpy)                              \
 309       SymI_HasProto(memmove)                             \
 310       SymI_HasProto(realloc)                             \
 311       SymI_HasProto(malloc)                              \
 312       SymI_HasProto(free)                                \
 313       SymI_HasProto(fork)                                \
 314       SymI_HasProto(lstat)                               \
 315       SymI_HasProto(isatty)                              \
 316       SymI_HasProto(mkdir)                               \
 317       SymI_HasProto(opendir)                             \
 318       SymI_HasProto(readdir)                             \
 319       SymI_HasProto(rewinddir)                           \
 320       SymI_HasProto(closedir)                            \
 321       SymI_HasProto(link)                                \
 322       SymI_HasProto(mkfifo)                              \
 323       SymI_HasProto(pipe)                                \
 324       SymI_HasProto(read)                                \
 325       SymI_HasProto(rename)                              \
 326       SymI_HasProto(rmdir)                               \
 327       SymI_HasProto(select)                              \
 328       SymI_HasProto(system)                              \
 329       SymI_HasProto(write)                               \
 330       SymI_HasProto(strcmp)                              \
 331       SymI_HasProto(strcpy)                              \
 332       SymI_HasProto(strncpy)                             \
 333       SymI_HasProto(strerror)                            \
 334       SymI_HasProto(sigaddset)                           \
 335       SymI_HasProto(sigemptyset)                         \
 336       SymI_HasProto(sigprocmask)                         \
 337       SymI_HasProto(umask)                               \
 338       SymI_HasProto(uname)                               \
 339       SymI_HasProto(unlink)                              \
 340       SymI_HasProto(utime)                               \
 341       SymI_HasProto(waitpid)
 342
 343 #elif !defined(mingw32_HOST_OS)
 344 #define RTS_MINGW_ONLY_SYMBOLS /**/
 345 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 346 #else /* defined(mingw32_HOST_OS) */
 347 #define RTS_POSIX_ONLY_SYMBOLS  /**/
 348 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 349
 350 /* Extra syms gen'ed by mingw-2's gcc-3.2: */
 351 #if __GNUC__>=3
 352 #define RTS_MINGW_EXTRA_SYMS                    \
 353       SymI_NeedsProto(_imp____mb_cur_max)       \
 354       SymI_NeedsProto(_imp___pctype)
 355 #else
 356 #define RTS_MINGW_EXTRA_SYMS
 357 #endif
 358
 359 #if HAVE_GETTIMEOFDAY
 360 #define RTS_MINGW_GETTIMEOFDAY_SYM SymI_NeedsProto(gettimeofday)
 361 #else
 362 #define RTS_MINGW_GETTIMEOFDAY_SYM /**/
 363 #endif
 364
 365 #if HAVE___MINGW_VFPRINTF
 366 #define RTS___MINGW_VFPRINTF_SYM SymI_HasProto(__mingw_vfprintf)
 367 #else
 368 #define RTS___MINGW_VFPRINTF_SYM /**/
 369 #endif
 370
 371 /* These are statically linked from the mingw libraries into the ghc
 372    executable, so we have to employ this hack. */
 373 #define RTS_MINGW_ONLY_SYMBOLS                           \
 374       SymI_HasProto(stg_asyncReadzh)                     \
 375       SymI_HasProto(stg_asyncWritezh)                    \
 376       SymI_HasProto(stg_asyncDoProczh)                   \
 377       SymI_HasProto(memset)                              \
 378       SymI_HasProto(inet_ntoa)                           \
 379       SymI_HasProto(inet_addr)                           \
 380       SymI_HasProto(htonl)                               \
 381       SymI_HasProto(recvfrom)                            \
 382       SymI_HasProto(listen)                              \
 383       SymI_HasProto(bind)                                \
 384       SymI_HasProto(shutdown)                            \
 385       SymI_HasProto(connect)                             \
 386       SymI_HasProto(htons)                               \
 387       SymI_HasProto(ntohs)                               \
 388       SymI_HasProto(getservbyname)                       \
 389       SymI_HasProto(getservbyport)                       \
 390       SymI_HasProto(getprotobynumber)                    \
 391       SymI_HasProto(getprotobyname)                      \
 392       SymI_HasProto(gethostbyname)                       \
 393       SymI_HasProto(gethostbyaddr)                       \
 394       SymI_HasProto(gethostname)                         \
 395       SymI_HasProto(strcpy)                              \
 396       SymI_HasProto(strncpy)                             \
 397       SymI_HasProto(abort)                               \
 398       SymI_NeedsProto(_alloca)                           \
 399       SymI_HasProto(isxdigit)                          \
 400       SymI_HasProto(isupper)                           \
 401       SymI_HasProto(ispunct)                           \
 402       SymI_HasProto(islower)                           \
 403       SymI_HasProto(isspace)                           \
 404       SymI_HasProto(isprint)                           \
 405       SymI_HasProto(isdigit)                           \
 406       SymI_HasProto(iscntrl)                           \
 407       SymI_HasProto(isalpha)                           \
 408       SymI_HasProto(isalnum)                           \
 409       SymI_HasProto(isascii)                           \
 410       RTS___MINGW_VFPRINTF_SYM                           \
 411       SymI_HasProto(strcmp)                              \
 412       SymI_HasProto(memmove)                             \
 413       SymI_HasProto(realloc)                             \
 414       SymI_HasProto(malloc)                              \
 415       SymI_HasProto(pow)                                 \
 416       SymI_HasProto(tanh)                                \
 417       SymI_HasProto(cosh)                                \
 418       SymI_HasProto(sinh)                                \
 419       SymI_HasProto(atan)                                \
 420       SymI_HasProto(acos)                                \
 421       SymI_HasProto(asin)                                \
 422       SymI_HasProto(tan)                                 \
 423       SymI_HasProto(cos)                                 \
 424       SymI_HasProto(sin)                                 \
 425       SymI_HasProto(exp)                                 \
 426       SymI_HasProto(log)                                 \
 427       SymI_HasProto(sqrt)                                \
 428       SymI_HasProto(powf)                                \
 429       SymI_HasProto(tanhf)                               \
 430       SymI_HasProto(coshf)                               \
 431       SymI_HasProto(sinhf)                               \
 432       SymI_HasProto(atanf)                               \
 433       SymI_HasProto(acosf)                               \
 434       SymI_HasProto(asinf)                               \
 435       SymI_HasProto(tanf)                                \
 436       SymI_HasProto(cosf)                                \
 437       SymI_HasProto(sinf)                                \
 438       SymI_HasProto(expf)                                \
 439       SymI_HasProto(logf)                                \
 440       SymI_HasProto(sqrtf)                               \
 441       SymI_HasProto(erf)                                \
 442       SymI_HasProto(erfc)                                \
 443       SymI_HasProto(erff)                                \
 444       SymI_HasProto(erfcf)                               \
 445       SymI_HasProto(memcpy)                              \
 446       SymI_HasProto(rts_InstallConsoleEvent)             \
 447       SymI_HasProto(rts_ConsoleHandlerDone)              \
 448       SymI_NeedsProto(mktime)                            \
 449       SymI_NeedsProto(_imp___timezone)                   \
 450       SymI_NeedsProto(_imp___tzname)                     \
 451       SymI_NeedsProto(_imp__tzname)                      \
 452       SymI_NeedsProto(_imp___iob)                        \
 453       SymI_NeedsProto(_imp___osver)                      \
 454       SymI_NeedsProto(localtime)                         \
 455       SymI_NeedsProto(gmtime)                            \
 456       SymI_NeedsProto(opendir)                           \
 457       SymI_NeedsProto(readdir)                           \
 458       SymI_NeedsProto(rewinddir)                         \
 459       RTS_MINGW_EXTRA_SYMS                               \
 460       RTS_MINGW_GETTIMEOFDAY_SYM                         \
 461       SymI_NeedsProto(closedir)
 462 #endif
 463
 464 #if defined(darwin_HOST_OS) && HAVE_PRINTF_LDBLSTUB
 465 #define RTS_DARWIN_ONLY_SYMBOLS                             \
 466      SymI_NeedsProto(asprintf$LDBLStub)                     \
 467      SymI_NeedsProto(err$LDBLStub)                          \
 468      SymI_NeedsProto(errc$LDBLStub)                         \
 469      SymI_NeedsProto(errx$LDBLStub)                         \
 470      SymI_NeedsProto(fprintf$LDBLStub)                      \
 471      SymI_NeedsProto(fscanf$LDBLStub)                       \
 472      SymI_NeedsProto(fwprintf$LDBLStub)                     \
 473      SymI_NeedsProto(fwscanf$LDBLStub)                      \
 474      SymI_NeedsProto(printf$LDBLStub)                       \
 475      SymI_NeedsProto(scanf$LDBLStub)                        \
 476      SymI_NeedsProto(snprintf$LDBLStub)                     \
 477      SymI_NeedsProto(sprintf$LDBLStub)                      \
 478      SymI_NeedsProto(sscanf$LDBLStub)                       \
 479      SymI_NeedsProto(strtold$LDBLStub)                      \
 480      SymI_NeedsProto(swprintf$LDBLStub)                     \
 481      SymI_NeedsProto(swscanf$LDBLStub)                      \
 482      SymI_NeedsProto(syslog$LDBLStub)                       \
 483      SymI_NeedsProto(vasprintf$LDBLStub)                    \
 484      SymI_NeedsProto(verr$LDBLStub)                         \
 485      SymI_NeedsProto(verrc$LDBLStub)                        \
 486      SymI_NeedsProto(verrx$LDBLStub)                        \
 487      SymI_NeedsProto(vfprintf$LDBLStub)                     \
 488      SymI_NeedsProto(vfscanf$LDBLStub)                      \
 489      SymI_NeedsProto(vfwprintf$LDBLStub)                    \
 490      SymI_NeedsProto(vfwscanf$LDBLStub)                     \
 491      SymI_NeedsProto(vprintf$LDBLStub)                      \
 492      SymI_NeedsProto(vscanf$LDBLStub)                       \
 493      SymI_NeedsProto(vsnprintf$LDBLStub)                    \
 494      SymI_NeedsProto(vsprintf$LDBLStub)                     \
 495      SymI_NeedsProto(vsscanf$LDBLStub)                      \
 496      SymI_NeedsProto(vswprintf$LDBLStub)                    \
 497      SymI_NeedsProto(vswscanf$LDBLStub)                     \
 498      SymI_NeedsProto(vsyslog$LDBLStub)                      \
 499      SymI_NeedsProto(vwarn$LDBLStub)                        \
 500      SymI_NeedsProto(vwarnc$LDBLStub)                       \
 501      SymI_NeedsProto(vwarnx$LDBLStub)                       \
 502      SymI_NeedsProto(vwprintf$LDBLStub)                     \
 503      SymI_NeedsProto(vwscanf$LDBLStub)                      \
 504      SymI_NeedsProto(warn$LDBLStub)                         \
 505      SymI_NeedsProto(warnc$LDBLStub)                        \
 506      SymI_NeedsProto(warnx$LDBLStub)                        \
 507      SymI_NeedsProto(wcstold$LDBLStub)                      \
 508      SymI_NeedsProto(wprintf$LDBLStub)                      \
 509      SymI_NeedsProto(wscanf$LDBLStub)
 510 #else
 511 #define RTS_DARWIN_ONLY_SYMBOLS
 512 #endif
 513
 514 #ifndef SMP
 515 # define MAIN_CAP_SYM SymI_HasProto(MainCapability)
 516 #else
 517 # define MAIN_CAP_SYM
 518 #endif
 519
 520 #if !defined(mingw32_HOST_OS)
 521 #define RTS_USER_SIGNALS_SYMBOLS \
 522    SymI_HasProto(setIOManagerControlFd) \
 523    SymI_HasProto(setIOManagerWakeupFd) \
 524    SymI_HasProto(ioManagerWakeup) \
 525    SymI_HasProto(blockUserSignals) \
 526    SymI_HasProto(unblockUserSignals)
 527 #else
 528 #define RTS_USER_SIGNALS_SYMBOLS     \
 529    SymI_HasProto(ioManagerWakeup) \
 530    SymI_HasProto(sendIOManagerEvent) \
 531    SymI_HasProto(readIOManagerEvent) \
 532    SymI_HasProto(getIOManagerEvent)  \
 533    SymI_HasProto(console_handler)
 534 #endif
 535
 536 #define RTS_LIBFFI_SYMBOLS                                  \
 537      SymE_NeedsProto(ffi_prep_cif)                          \
 538      SymE_NeedsProto(ffi_call)                              \
 539      SymE_NeedsProto(ffi_type_void)                         \
 540      SymE_NeedsProto(ffi_type_float)                        \
 541      SymE_NeedsProto(ffi_type_double)                       \
 542      SymE_NeedsProto(ffi_type_sint64)                       \
 543      SymE_NeedsProto(ffi_type_uint64)                       \
 544      SymE_NeedsProto(ffi_type_sint32)                       \
 545      SymE_NeedsProto(ffi_type_uint32)                       \
 546      SymE_NeedsProto(ffi_type_sint16)                       \
 547      SymE_NeedsProto(ffi_type_uint16)                       \
 548      SymE_NeedsProto(ffi_type_sint8)                        \
 549      SymE_NeedsProto(ffi_type_uint8)                        \
 550      SymE_NeedsProto(ffi_type_pointer)
 551
 552 #ifdef TABLES_NEXT_TO_CODE
 553 #define RTS_RET_SYMBOLS /* nothing */
 554 #else
 555 #define RTS_RET_SYMBOLS                                 \
 556       SymI_HasProto(stg_enter_ret)                      \
 557       SymI_HasProto(stg_gc_fun_ret)                     \
 558       SymI_HasProto(stg_ap_v_ret)                       \
 559       SymI_HasProto(stg_ap_f_ret)                       \
 560       SymI_HasProto(stg_ap_d_ret)                       \
 561       SymI_HasProto(stg_ap_l_ret)                       \
 562       SymI_HasProto(stg_ap_n_ret)                       \
 563       SymI_HasProto(stg_ap_p_ret)                       \
 564       SymI_HasProto(stg_ap_pv_ret)                      \
 565       SymI_HasProto(stg_ap_pp_ret)                      \
 566       SymI_HasProto(stg_ap_ppv_ret)                     \
 567       SymI_HasProto(stg_ap_ppp_ret)                     \
 568       SymI_HasProto(stg_ap_pppv_ret)                    \
 569       SymI_HasProto(stg_ap_pppp_ret)                    \
 570       SymI_HasProto(stg_ap_ppppp_ret)                   \
 571       SymI_HasProto(stg_ap_pppppp_ret)
 572 #endif
 573
 574 /* Modules compiled with -ticky may mention ticky counters */
 575 /* This list should marry up with the one in $(TOP)/includes/stg/Ticky.h */
 576 #define RTS_TICKY_SYMBOLS                       \
 577       SymI_NeedsProto(ticky_entry_ctrs)         \
 578       SymI_NeedsProto(top_ct)                   \
 579                                                 \
 580       SymI_HasProto(ENT_VIA_NODE_ctr)           \
 581       SymI_HasProto(ENT_STATIC_THK_ctr)         \
 582       SymI_HasProto(ENT_DYN_THK_ctr)            \
 583       SymI_HasProto(ENT_STATIC_FUN_DIRECT_ctr)  \
 584       SymI_HasProto(ENT_DYN_FUN_DIRECT_ctr)     \
 585       SymI_HasProto(ENT_STATIC_CON_ctr)         \
 586       SymI_HasProto(ENT_DYN_CON_ctr)            \
 587       SymI_HasProto(ENT_STATIC_IND_ctr)         \
 588       SymI_HasProto(ENT_DYN_IND_ctr)            \
 589       SymI_HasProto(ENT_PERM_IND_ctr)           \
 590       SymI_HasProto(ENT_PAP_ctr)                \
 591       SymI_HasProto(ENT_AP_ctr)                 \
 592       SymI_HasProto(ENT_AP_STACK_ctr)           \
 593       SymI_HasProto(ENT_BH_ctr)                 \
 594       SymI_HasProto(UNKNOWN_CALL_ctr)           \
 595       SymI_HasProto(SLOW_CALL_v_ctr)            \
 596       SymI_HasProto(SLOW_CALL_f_ctr)            \
 597       SymI_HasProto(SLOW_CALL_d_ctr)            \
 598       SymI_HasProto(SLOW_CALL_l_ctr)            \
 599       SymI_HasProto(SLOW_CALL_n_ctr)            \
 600       SymI_HasProto(SLOW_CALL_p_ctr)            \
 601       SymI_HasProto(SLOW_CALL_pv_ctr)           \
 602       SymI_HasProto(SLOW_CALL_pp_ctr)           \
 603       SymI_HasProto(SLOW_CALL_ppv_ctr)          \
 604       SymI_HasProto(SLOW_CALL_ppp_ctr)          \
 605       SymI_HasProto(SLOW_CALL_pppv_ctr)         \
 606       SymI_HasProto(SLOW_CALL_pppp_ctr)         \
 607       SymI_HasProto(SLOW_CALL_ppppp_ctr)                \
 608       SymI_HasProto(SLOW_CALL_pppppp_ctr)               \
 609       SymI_HasProto(SLOW_CALL_OTHER_ctr)                \
 610       SymI_HasProto(ticky_slow_call_unevald)            \
 611       SymI_HasProto(SLOW_CALL_ctr)                      \
 612       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_ctr)          \
 613       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_CHUNKS_ctr)   \
 614       SymI_HasProto(KNOWN_CALL_ctr)                     \
 615       SymI_HasProto(KNOWN_CALL_TOO_FEW_ARGS_ctr)        \
 616       SymI_HasProto(KNOWN_CALL_EXTRA_ARGS_ctr)          \
 617       SymI_HasProto(SLOW_CALL_FUN_TOO_FEW_ctr)          \
 618       SymI_HasProto(SLOW_CALL_FUN_CORRECT_ctr)          \
 619       SymI_HasProto(SLOW_CALL_FUN_TOO_MANY_ctr)         \
 620       SymI_HasProto(SLOW_CALL_PAP_TOO_FEW_ctr)          \
 621       SymI_HasProto(SLOW_CALL_PAP_CORRECT_ctr)          \
 622       SymI_HasProto(SLOW_CALL_PAP_TOO_MANY_ctr)         \
 623       SymI_HasProto(SLOW_CALL_UNEVALD_ctr)              \
 624       SymI_HasProto(UPDF_OMITTED_ctr)           \
 625       SymI_HasProto(UPDF_PUSHED_ctr)            \
 626       SymI_HasProto(CATCHF_PUSHED_ctr)          \
 627       SymI_HasProto(UPDF_RCC_PUSHED_ctr)        \
 628       SymI_HasProto(UPDF_RCC_OMITTED_ctr)       \
 629       SymI_HasProto(UPD_SQUEEZED_ctr)           \
 630       SymI_HasProto(UPD_CON_IN_NEW_ctr)         \
 631       SymI_HasProto(UPD_CON_IN_PLACE_ctr)       \
 632       SymI_HasProto(UPD_PAP_IN_NEW_ctr)         \
 633       SymI_HasProto(UPD_PAP_IN_PLACE_ctr)       \
 634       SymI_HasProto(ALLOC_HEAP_ctr)             \
 635       SymI_HasProto(ALLOC_HEAP_tot)             \
 636       SymI_HasProto(ALLOC_FUN_ctr)              \
 637       SymI_HasProto(ALLOC_FUN_adm)              \
 638       SymI_HasProto(ALLOC_FUN_gds)              \
 639       SymI_HasProto(ALLOC_FUN_slp)              \
 640       SymI_HasProto(UPD_NEW_IND_ctr)            \
 641       SymI_HasProto(UPD_NEW_PERM_IND_ctr)       \
 642       SymI_HasProto(UPD_OLD_IND_ctr)            \
 643       SymI_HasProto(UPD_OLD_PERM_IND_ctr)               \
 644       SymI_HasProto(UPD_BH_UPDATABLE_ctr)               \
 645       SymI_HasProto(UPD_BH_SINGLE_ENTRY_ctr)            \
 646       SymI_HasProto(UPD_CAF_BH_UPDATABLE_ctr)           \
 647       SymI_HasProto(UPD_CAF_BH_SINGLE_ENTRY_ctr)        \
 648       SymI_HasProto(GC_SEL_ABANDONED_ctr)               \
 649       SymI_HasProto(GC_SEL_MINOR_ctr)           \
 650       SymI_HasProto(GC_SEL_MAJOR_ctr)           \
 651       SymI_HasProto(GC_FAILED_PROMOTION_ctr)    \
 652       SymI_HasProto(ALLOC_UP_THK_ctr)           \
 653       SymI_HasProto(ALLOC_SE_THK_ctr)           \
 654       SymI_HasProto(ALLOC_THK_adm)              \
 655       SymI_HasProto(ALLOC_THK_gds)              \
 656       SymI_HasProto(ALLOC_THK_slp)              \
 657       SymI_HasProto(ALLOC_CON_ctr)              \
 658       SymI_HasProto(ALLOC_CON_adm)              \
 659       SymI_HasProto(ALLOC_CON_gds)              \
 660       SymI_HasProto(ALLOC_CON_slp)              \
 661       SymI_HasProto(ALLOC_TUP_ctr)              \
 662       SymI_HasProto(ALLOC_TUP_adm)              \
 663       SymI_HasProto(ALLOC_TUP_gds)              \
 664       SymI_HasProto(ALLOC_TUP_slp)              \
 665       SymI_HasProto(ALLOC_BH_ctr)               \
 666       SymI_HasProto(ALLOC_BH_adm)               \
 667       SymI_HasProto(ALLOC_BH_gds)               \
 668       SymI_HasProto(ALLOC_BH_slp)               \
 669       SymI_HasProto(ALLOC_PRIM_ctr)             \
 670       SymI_HasProto(ALLOC_PRIM_adm)             \
 671       SymI_HasProto(ALLOC_PRIM_gds)             \
 672       SymI_HasProto(ALLOC_PRIM_slp)             \
 673       SymI_HasProto(ALLOC_PAP_ctr)              \
 674       SymI_HasProto(ALLOC_PAP_adm)              \
 675       SymI_HasProto(ALLOC_PAP_gds)              \
 676       SymI_HasProto(ALLOC_PAP_slp)              \
 677       SymI_HasProto(ALLOC_TSO_ctr)              \
 678       SymI_HasProto(ALLOC_TSO_adm)              \
 679       SymI_HasProto(ALLOC_TSO_gds)              \
 680       SymI_HasProto(ALLOC_TSO_slp)              \
 681       SymI_HasProto(RET_NEW_ctr)                \
 682       SymI_HasProto(RET_OLD_ctr)                \
 683       SymI_HasProto(RET_UNBOXED_TUP_ctr)        \
 684       SymI_HasProto(RET_SEMI_loads_avoided)
 685
 686
 687 // On most platforms, the garbage collector rewrites references
 688 //      to small integer and char objects to a set of common, shared ones.
 689 //
 690 // We don't do this when compiling to Windows DLLs at the moment because
 691 //      it doesn't support cross package data references well.
 692 //
 693 #if defined(__PIC__) && defined(mingw32_HOST_OS)
 694 #define RTS_INTCHAR_SYMBOLS
 695 #else
 696 #define RTS_INTCHAR_SYMBOLS                             \
 697       SymI_HasProto(stg_CHARLIKE_closure)               \
 698       SymI_HasProto(stg_INTLIKE_closure)
 699 #endif
 700
 701
 702 #define RTS_SYMBOLS                                     \
 703       Maybe_Stable_Names                                \
 704       RTS_TICKY_SYMBOLS                                 \
 705       SymI_HasProto(StgReturn)                          \
 706       SymI_HasProto(stg_enter_info)                     \
 707       SymI_HasProto(stg_gc_void_info)                   \
 708       SymI_HasProto(__stg_gc_enter_1)                   \
 709       SymI_HasProto(stg_gc_noregs)                      \
 710       SymI_HasProto(stg_gc_unpt_r1_info)                \
 711       SymI_HasProto(stg_gc_unpt_r1)                     \
 712       SymI_HasProto(stg_gc_unbx_r1_info)                \
 713       SymI_HasProto(stg_gc_unbx_r1)                     \
 714       SymI_HasProto(stg_gc_f1_info)                     \
 715       SymI_HasProto(stg_gc_f1)                          \
 716       SymI_HasProto(stg_gc_d1_info)                     \
 717       SymI_HasProto(stg_gc_d1)                          \
 718       SymI_HasProto(stg_gc_l1_info)                     \
 719       SymI_HasProto(stg_gc_l1)                          \
 720       SymI_HasProto(__stg_gc_fun)                       \
 721       SymI_HasProto(stg_gc_fun_info)                    \
 722       SymI_HasProto(stg_gc_gen)                         \
 723       SymI_HasProto(stg_gc_gen_info)                    \
 724       SymI_HasProto(stg_gc_gen_hp)                      \
 725       SymI_HasProto(stg_gc_ut)                          \
 726       SymI_HasProto(stg_gen_yield)                      \
 727       SymI_HasProto(stg_yield_noregs)                   \
 728       SymI_HasProto(stg_yield_to_interpreter)           \
 729       SymI_HasProto(stg_gen_block)                      \
 730       SymI_HasProto(stg_block_noregs)                   \
 731       SymI_HasProto(stg_block_1)                        \
 732       SymI_HasProto(stg_block_takemvar)                 \
 733       SymI_HasProto(stg_block_putmvar)                  \
 734       MAIN_CAP_SYM                                      \
 735       SymI_HasProto(MallocFailHook)                     \
 736       SymI_HasProto(OnExitHook)                         \
 737       SymI_HasProto(OutOfHeapHook)                      \
 738       SymI_HasProto(StackOverflowHook)                  \
 739       SymI_HasProto(addDLL)                             \
 740       SymI_HasProto(__int_encodeDouble)                 \
 741       SymI_HasProto(__word_encodeDouble)                \
 742       SymI_HasProto(__2Int_encodeDouble)                \
 743       SymI_HasProto(__int_encodeFloat)                  \
 744       SymI_HasProto(__word_encodeFloat)                 \
 745       SymI_HasProto(stg_atomicallyzh)                   \
 746       SymI_HasProto(barf)                               \
 747       SymI_HasProto(debugBelch)                         \
 748       SymI_HasProto(errorBelch)                         \
 749       SymI_HasProto(sysErrorBelch)                      \
 750       SymI_HasProto(stg_getMaskingStatezh)              \
 751       SymI_HasProto(stg_maskAsyncExceptionszh)          \
 752       SymI_HasProto(stg_maskUninterruptiblezh)          \
 753       SymI_HasProto(stg_catchzh)                        \
 754       SymI_HasProto(stg_catchRetryzh)                   \
 755       SymI_HasProto(stg_catchSTMzh)                     \
 756       SymI_HasProto(stg_checkzh)                        \
 757       SymI_HasProto(closure_flags)                      \
 758       SymI_HasProto(cmp_thread)                         \
 759       SymI_HasProto(createAdjustor)                     \
 760       SymI_HasProto(stg_decodeDoublezu2Intzh)           \
 761       SymI_HasProto(stg_decodeFloatzuIntzh)             \
 762       SymI_HasProto(defaultsHook)                       \
 763       SymI_HasProto(stg_delayzh)                        \
 764       SymI_HasProto(stg_deRefWeakzh)                    \
 765       SymI_HasProto(stg_deRefStablePtrzh)               \
 766       SymI_HasProto(dirty_MUT_VAR)                      \
 767       SymI_HasProto(stg_forkzh)                         \
 768       SymI_HasProto(stg_forkOnzh)                       \
 769       SymI_HasProto(forkProcess)                        \
 770       SymI_HasProto(forkOS_createThread)                \
 771       SymI_HasProto(freeHaskellFunctionPtr)             \
 772       SymI_HasProto(getOrSetTypeableStore)              \
 773       SymI_HasProto(getOrSetGHCConcSignalSignalHandlerStore)            \
 774       SymI_HasProto(getOrSetGHCConcWindowsPendingDelaysStore)           \
 775       SymI_HasProto(getOrSetGHCConcWindowsIOManagerThreadStore) \
 776       SymI_HasProto(getOrSetGHCConcWindowsProddingStore)                \
 777       SymI_HasProto(getOrSetSystemEventThreadEventManagerStore)         \
 778       SymI_HasProto(getOrSetSystemEventThreadIOManagerThreadStore)              \
 779       SymI_HasProto(genSymZh)                           \
 780       SymI_HasProto(genericRaise)                       \
 781       SymI_HasProto(getProgArgv)                        \
 782       SymI_HasProto(getFullProgArgv)                    \
 783       SymI_HasProto(getStablePtr)                       \
 784       SymI_HasProto(hs_init)                            \
 785       SymI_HasProto(hs_exit)                            \
 786       SymI_HasProto(hs_set_argv)                        \
 787       SymI_HasProto(hs_add_root)                        \
 788       SymI_HasProto(hs_perform_gc)                      \
 789       SymI_HasProto(hs_free_stable_ptr)                 \
 790       SymI_HasProto(hs_free_fun_ptr)                    \
 791       SymI_HasProto(hs_hpc_rootModule)                  \
 792       SymI_HasProto(hs_hpc_module)                      \
 793       SymI_HasProto(initLinker)                         \
 794       SymI_HasProto(stg_unpackClosurezh)                \
 795       SymI_HasProto(stg_getApStackValzh)                \
 796       SymI_HasProto(stg_getSparkzh)                     \
 797       SymI_HasProto(stg_numSparkszh)                    \
 798       SymI_HasProto(stg_isCurrentThreadBoundzh)         \
 799       SymI_HasProto(stg_isEmptyMVarzh)                  \
 800       SymI_HasProto(stg_killThreadzh)                   \
 801       SymI_HasProto(loadObj)                            \
 802       SymI_HasProto(insertStableSymbol)                 \
 803       SymI_HasProto(insertSymbol)                       \
 804       SymI_HasProto(lookupSymbol)                       \
 805       SymI_HasProto(stg_makeStablePtrzh)                \
 806       SymI_HasProto(stg_mkApUpd0zh)                     \
 807       SymI_HasProto(stg_myThreadIdzh)                   \
 808       SymI_HasProto(stg_labelThreadzh)                  \
 809       SymI_HasProto(stg_newArrayzh)                     \
 810       SymI_HasProto(stg_newBCOzh)                       \
 811       SymI_HasProto(stg_newByteArrayzh)                 \
 812       SymI_HasProto_redirect(newCAF, newDynCAF)         \
 813       SymI_HasProto(stg_newMVarzh)                      \
 814       SymI_HasProto(stg_newMutVarzh)                    \
 815       SymI_HasProto(stg_newTVarzh)                      \
 816       SymI_HasProto(stg_noDuplicatezh)                  \
 817       SymI_HasProto(stg_atomicModifyMutVarzh)           \
 818       SymI_HasProto(stg_newPinnedByteArrayzh)           \
 819       SymI_HasProto(stg_newAlignedPinnedByteArrayzh)    \
 820       SymI_HasProto(newSpark)                           \
 821       SymI_HasProto(performGC)                          \
 822       SymI_HasProto(performMajorGC)                     \
 823       SymI_HasProto(prog_argc)                          \
 824       SymI_HasProto(prog_argv)                          \
 825       SymI_HasProto(stg_putMVarzh)                      \
 826       SymI_HasProto(stg_raisezh)                        \
 827       SymI_HasProto(stg_raiseIOzh)                      \
 828       SymI_HasProto(stg_readTVarzh)                     \
 829       SymI_HasProto(stg_readTVarIOzh)                   \
 830       SymI_HasProto(resumeThread)                       \
 831       SymI_HasProto(resolveObjs)                        \
 832       SymI_HasProto(stg_retryzh)                        \
 833       SymI_HasProto(rts_apply)                          \
 834       SymI_HasProto(rts_checkSchedStatus)               \
 835       SymI_HasProto(rts_eval)                           \
 836       SymI_HasProto(rts_evalIO)                         \
 837       SymI_HasProto(rts_evalLazyIO)                     \
 838       SymI_HasProto(rts_evalStableIO)                   \
 839       SymI_HasProto(rts_eval_)                          \
 840       SymI_HasProto(rts_getBool)                        \
 841       SymI_HasProto(rts_getChar)                        \
 842       SymI_HasProto(rts_getDouble)                      \
 843       SymI_HasProto(rts_getFloat)                       \
 844       SymI_HasProto(rts_getInt)                         \
 845       SymI_HasProto(rts_getInt8)                        \
 846       SymI_HasProto(rts_getInt16)                       \
 847       SymI_HasProto(rts_getInt32)                       \
 848       SymI_HasProto(rts_getInt64)                       \
 849       SymI_HasProto(rts_getPtr)                         \
 850       SymI_HasProto(rts_getFunPtr)                      \
 851       SymI_HasProto(rts_getStablePtr)                   \
 852       SymI_HasProto(rts_getThreadId)                    \
 853       SymI_HasProto(rts_getWord)                        \
 854       SymI_HasProto(rts_getWord8)                       \
 855       SymI_HasProto(rts_getWord16)                      \
 856       SymI_HasProto(rts_getWord32)                      \
 857       SymI_HasProto(rts_getWord64)                      \
 858       SymI_HasProto(rts_lock)                           \
 859       SymI_HasProto(rts_mkBool)                         \
 860       SymI_HasProto(rts_mkChar)                         \
 861       SymI_HasProto(rts_mkDouble)                       \
 862       SymI_HasProto(rts_mkFloat)                        \
 863       SymI_HasProto(rts_mkInt)                          \
 864       SymI_HasProto(rts_mkInt8)                         \
 865       SymI_HasProto(rts_mkInt16)                        \
 866       SymI_HasProto(rts_mkInt32)                        \
 867       SymI_HasProto(rts_mkInt64)                        \
 868       SymI_HasProto(rts_mkPtr)                          \
 869       SymI_HasProto(rts_mkFunPtr)                       \
 870       SymI_HasProto(rts_mkStablePtr)                    \
 871       SymI_HasProto(rts_mkString)                       \
 872       SymI_HasProto(rts_mkWord)                         \
 873       SymI_HasProto(rts_mkWord8)                        \
 874       SymI_HasProto(rts_mkWord16)                       \
 875       SymI_HasProto(rts_mkWord32)                       \
 876       SymI_HasProto(rts_mkWord64)                       \
 877       SymI_HasProto(rts_unlock)                         \
 878       SymI_HasProto(rts_unsafeGetMyCapability)          \
 879       SymI_HasProto(rtsSupportsBoundThreads)            \
 880       SymI_HasProto(rts_isProfiled)                     \
 881       SymI_HasProto(setProgArgv)                        \
 882       SymI_HasProto(startupHaskell)                     \
 883       SymI_HasProto(shutdownHaskell)                    \
 884       SymI_HasProto(shutdownHaskellAndExit)             \
 885       SymI_HasProto(stable_ptr_table)                   \
 886       SymI_HasProto(stackOverflow)                      \
 887       SymI_HasProto(stg_CAF_BLACKHOLE_info)             \
 888       SymI_HasProto(stg_BLACKHOLE_info)                 \
 889       SymI_HasProto(__stg_EAGER_BLACKHOLE_info)         \
 890       SymI_HasProto(stg_BLOCKING_QUEUE_CLEAN_info)      \
 891       SymI_HasProto(stg_BLOCKING_QUEUE_DIRTY_info)      \
 892       SymI_HasProto(startTimer)                         \
 893       SymI_HasProto(stg_MVAR_CLEAN_info)                \
 894       SymI_HasProto(stg_MVAR_DIRTY_info)                \
 895       SymI_HasProto(stg_IND_STATIC_info)                \
 896       SymI_HasProto(stg_ARR_WORDS_info)                 \
 897       SymI_HasProto(stg_MUT_ARR_PTRS_DIRTY_info)        \
 898       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN_info)       \
 899       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN0_info)      \
 900       SymI_HasProto(stg_WEAK_info)                      \
 901       SymI_HasProto(stg_ap_v_info)                      \
 902       SymI_HasProto(stg_ap_f_info)                      \
 903       SymI_HasProto(stg_ap_d_info)                      \
 904       SymI_HasProto(stg_ap_l_info)                      \
 905       SymI_HasProto(stg_ap_n_info)                      \
 906       SymI_HasProto(stg_ap_p_info)                      \
 907       SymI_HasProto(stg_ap_pv_info)                     \
 908       SymI_HasProto(stg_ap_pp_info)                     \
 909       SymI_HasProto(stg_ap_ppv_info)                    \
 910       SymI_HasProto(stg_ap_ppp_info)                    \
 911       SymI_HasProto(stg_ap_pppv_info)                   \
 912       SymI_HasProto(stg_ap_pppp_info)                   \
 913       SymI_HasProto(stg_ap_ppppp_info)                  \
 914       SymI_HasProto(stg_ap_pppppp_info)                 \
 915       SymI_HasProto(stg_ap_0_fast)                      \
 916       SymI_HasProto(stg_ap_v_fast)                      \
 917       SymI_HasProto(stg_ap_f_fast)                      \
 918       SymI_HasProto(stg_ap_d_fast)                      \
 919       SymI_HasProto(stg_ap_l_fast)                      \
 920       SymI_HasProto(stg_ap_n_fast)                      \
 921       SymI_HasProto(stg_ap_p_fast)                      \
 922       SymI_HasProto(stg_ap_pv_fast)                     \
 923       SymI_HasProto(stg_ap_pp_fast)                     \
 924       SymI_HasProto(stg_ap_ppv_fast)                    \
 925       SymI_HasProto(stg_ap_ppp_fast)                    \
 926       SymI_HasProto(stg_ap_pppv_fast)                   \
 927       SymI_HasProto(stg_ap_pppp_fast)                   \
 928       SymI_HasProto(stg_ap_ppppp_fast)                  \
 929       SymI_HasProto(stg_ap_pppppp_fast)                 \
 930       SymI_HasProto(stg_ap_1_upd_info)                  \
 931       SymI_HasProto(stg_ap_2_upd_info)                  \
 932       SymI_HasProto(stg_ap_3_upd_info)                  \
 933       SymI_HasProto(stg_ap_4_upd_info)                  \
 934       SymI_HasProto(stg_ap_5_upd_info)                  \
 935       SymI_HasProto(stg_ap_6_upd_info)                  \
 936       SymI_HasProto(stg_ap_7_upd_info)                  \
 937       SymI_HasProto(stg_exit)                           \
 938       SymI_HasProto(stg_sel_0_upd_info)                 \
 939       SymI_HasProto(stg_sel_10_upd_info)                \
 940       SymI_HasProto(stg_sel_11_upd_info)                \
 941       SymI_HasProto(stg_sel_12_upd_info)                \
 942       SymI_HasProto(stg_sel_13_upd_info)                \
 943       SymI_HasProto(stg_sel_14_upd_info)                \
 944       SymI_HasProto(stg_sel_15_upd_info)                \
 945       SymI_HasProto(stg_sel_1_upd_info)                 \
 946       SymI_HasProto(stg_sel_2_upd_info)                 \
 947       SymI_HasProto(stg_sel_3_upd_info)                 \
 948       SymI_HasProto(stg_sel_4_upd_info)                 \
 949       SymI_HasProto(stg_sel_5_upd_info)                 \
 950       SymI_HasProto(stg_sel_6_upd_info)                 \
 951       SymI_HasProto(stg_sel_7_upd_info)                 \
 952       SymI_HasProto(stg_sel_8_upd_info)                 \
 953       SymI_HasProto(stg_sel_9_upd_info)                 \
 954       SymI_HasProto(stg_upd_frame_info)                 \
 955       SymI_HasProto(stg_bh_upd_frame_info)              \
 956       SymI_HasProto(suspendThread)                      \
 957       SymI_HasProto(stg_takeMVarzh)                     \
 958       SymI_HasProto(stg_threadStatuszh)                 \
 959       SymI_HasProto(stg_tryPutMVarzh)                   \
 960       SymI_HasProto(stg_tryTakeMVarzh)                  \
 961       SymI_HasProto(stg_unmaskAsyncExceptionszh)        \
 962       SymI_HasProto(unloadObj)                          \
 963       SymI_HasProto(stg_unsafeThawArrayzh)              \
 964       SymI_HasProto(stg_waitReadzh)                     \
 965       SymI_HasProto(stg_waitWritezh)                    \
 966       SymI_HasProto(stg_writeTVarzh)                    \
 967       SymI_HasProto(stg_yieldzh)                        \
 968       SymI_NeedsProto(stg_interp_constr_entry)          \
 969       SymI_HasProto(stg_arg_bitmaps)                    \
 970       SymI_HasProto(alloc_blocks_lim)                   \
 971       SymI_HasProto(g0)                                 \
 972       SymI_HasProto(allocate)                           \
 973       SymI_HasProto(allocateExec)                       \
 974       SymI_HasProto(freeExec)                           \
 975       SymI_HasProto(getAllocations)                     \
 976       SymI_HasProto(revertCAFs)                         \
 977       SymI_HasProto(RtsFlags)                           \
 978       SymI_NeedsProto(rts_breakpoint_io_action)         \
 979       SymI_NeedsProto(rts_stop_next_breakpoint)         \
 980       SymI_NeedsProto(rts_stop_on_exception)            \
 981       SymI_HasProto(stopTimer)                          \
 982       SymI_HasProto(n_capabilities)                     \
 983       SymI_HasProto(stg_traceCcszh)                     \
 984       SymI_HasProto(stg_traceEventzh)                   \
 985       RTS_USER_SIGNALS_SYMBOLS                          \
 986       RTS_INTCHAR_SYMBOLS
 987
 988
 989 // 64-bit support functions in libgcc.a
 990 #if defined(__GNUC__) && SIZEOF_VOID_P <= 4
 991 #define RTS_LIBGCC_SYMBOLS                             \
 992       SymI_NeedsProto(__divdi3)                        \
 993       SymI_NeedsProto(__udivdi3)                       \
 994       SymI_NeedsProto(__moddi3)                        \
 995       SymI_NeedsProto(__umoddi3)                       \
 996       SymI_NeedsProto(__muldi3)                        \
 997       SymI_NeedsProto(__ashldi3)                       \
 998       SymI_NeedsProto(__ashrdi3)                       \
 999       SymI_NeedsProto(__lshrdi3)
1000 #else
1001 #define RTS_LIBGCC_SYMBOLS
1002 #endif
1003
1004 #if defined(darwin_HOST_OS) && defined(powerpc_HOST_ARCH)
1005       // Symbols that don't have a leading underscore
1006       // on Mac OS X. They have to receive special treatment,
1007       // see machoInitSymbolsWithoutUnderscore()
1008 #define RTS_MACHO_NOUNDERLINE_SYMBOLS           \
1009       SymI_NeedsProto(saveFP)                           \
1010       SymI_NeedsProto(restFP)
1011 #endif
1012
1013 /* entirely bogus claims about types of these symbols */
1014 #define SymI_NeedsProto(vvv)  extern void vvv(void);
1015 #if defined(__PIC__) && defined(mingw32_HOST_OS)
1016 #define SymE_HasProto(vvv)    SymE_HasProto(vvv);
1017 #define SymE_NeedsProto(vvv)    extern void _imp__ ## vvv (void);
1018 #else
1019 #define SymE_NeedsProto(vvv)  SymI_NeedsProto(vvv);
1020 #define SymE_HasProto(vvv)    SymI_HasProto(vvv)
1021 #endif
1022 #define SymI_HasProto(vvv) /**/
1023 #define SymI_HasProto_redirect(vvv,xxx) /**/
1024 RTS_SYMBOLS
1025 RTS_RET_SYMBOLS
1026 RTS_POSIX_ONLY_SYMBOLS
1027 RTS_MINGW_ONLY_SYMBOLS
1028 RTS_CYGWIN_ONLY_SYMBOLS
1029 RTS_DARWIN_ONLY_SYMBOLS
1030 RTS_LIBGCC_SYMBOLS
1031 RTS_LIBFFI_SYMBOLS
1032 #undef SymI_NeedsProto
1033 #undef SymI_HasProto
1034 #undef SymI_HasProto_redirect
1035 #undef SymE_HasProto
1036 #undef SymE_NeedsProto
1037
1038 #ifdef LEADING_UNDERSCORE
1039 #define MAYBE_LEADING_UNDERSCORE_STR(s) ("_" s)
1040 #else
1041 #define MAYBE_LEADING_UNDERSCORE_STR(s) (s)
1042 #endif
1043
1044 #define SymI_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1045                     (void*)(&(vvv)) },
1046 #define SymE_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1047             (void*)DLL_IMPORT_DATA_REF(vvv) },
1048
1049 #define SymI_NeedsProto(vvv) SymI_HasProto(vvv)
1050 #define SymE_NeedsProto(vvv) SymE_HasProto(vvv)
1051
1052 // SymI_HasProto_redirect allows us to redirect references to one symbol to
1053 // another symbol.  See newCAF/newDynCAF for an example.
1054 #define SymI_HasProto_redirect(vvv,xxx) \
1055     { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1056       (void*)(&(xxx)) },
1057
1058 static RtsSymbolVal rtsSyms[] = {
1059       RTS_SYMBOLS
1060       RTS_RET_SYMBOLS
1061       RTS_POSIX_ONLY_SYMBOLS
1062       RTS_MINGW_ONLY_SYMBOLS
1063       RTS_CYGWIN_ONLY_SYMBOLS
1064       RTS_DARWIN_ONLY_SYMBOLS
1065       RTS_LIBGCC_SYMBOLS
1066       RTS_LIBFFI_SYMBOLS
1067 #if defined(darwin_HOST_OS) && defined(i386_HOST_ARCH)
1068       // dyld stub code contains references to this,
1069       // but it should never be called because we treat
1070       // lazy pointers as nonlazy.
1071       { "dyld_stub_binding_helper", (void*)0xDEADBEEF },
1072 #endif
1073       { 0, 0 } /* sentinel */
1074 };
1075
1076
1077
1078 /* -----------------------------------------------------------------------------
1079  * Insert symbols into hash tables, checking for duplicates.
1080  */
1081
1082 static void ghciInsertStrHashTable ( char* obj_name,
1083                                      HashTable *table,
1084                                      char* key,
1085                                      void *data
1086                                    )
1087 {
1088    if (lookupHashTable(table, (StgWord)key) == NULL)
1089    {
1090       insertStrHashTable(table, (StgWord)key, data);
1091       return;
1092    }
1093    debugBelch(
1094       "\n\n"
1095       "GHCi runtime linker: fatal error: I found a duplicate definition for symbol\n"
1096       "   %s\n"
1097       "whilst processing object file\n"
1098       "   %s\n"
1099       "This could be caused by:\n"
1100       "   * Loading two different object files which export the same symbol\n"
1101       "   * Specifying the same object file twice on the GHCi command line\n"
1102       "   * An incorrect `package.conf' entry, causing some object to be\n"
1103       "     loaded twice.\n"
1104       "GHCi cannot safely continue in this situation.  Exiting now.  Sorry.\n"
1105       "\n",
1106       (char*)key,
1107       obj_name
1108    );
1109    exit(1);
1110 }
1111 /* -----------------------------------------------------------------------------
1112  * initialize the object linker
1113  */
1114
1115
1116 static int linker_init_done = 0 ;
1117
1118 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1119 static void *dl_prog_handle;
1120 static regex_t re_invalid;
1121 static regex_t re_realso;
1122 #ifdef THREADED_RTS
1123 static Mutex dl_mutex; // mutex to protect dlopen/dlerror critical section
1124 #endif
1125 #endif
1126
1127 void
1128 initLinker( void )
1129 {
1130     RtsSymbolVal *sym;
1131 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1132     int compileResult;
1133 #endif
1134
1135     /* Make initLinker idempotent, so we can call it
1136        before evey relevant operation; that means we
1137        don't need to initialise the linker separately */
1138     if (linker_init_done == 1) { return; } else {
1139       linker_init_done = 1;
1140     }
1141
1142 #if defined(THREADED_RTS) && (defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO))
1143     initMutex(&dl_mutex);
1144 #endif
1145     stablehash = allocStrHashTable();
1146     symhash = allocStrHashTable();
1147
1148     /* populate the symbol table with stuff from the RTS */
1149     for (sym = rtsSyms; sym->lbl != NULL; sym++) {
1150         ghciInsertStrHashTable("(GHCi built-in symbols)",
1151                                symhash, sym->lbl, sym->addr);
1152     }
1153 #   if defined(OBJFORMAT_MACHO) && defined(powerpc_HOST_ARCH)
1154     machoInitSymbolsWithoutUnderscore();
1155 #   endif
1156
1157 #   if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1158 #   if defined(RTLD_DEFAULT)
1159     dl_prog_handle = RTLD_DEFAULT;
1160 #   else
1161     dl_prog_handle = dlopen(NULL, RTLD_LAZY);
1162 #   endif /* RTLD_DEFAULT */
1163
1164     compileResult = regcomp(&re_invalid,
1165            "(([^ \t()])+\\.so([^ \t:()])*):([ \t])*invalid ELF header",
1166            REG_EXTENDED);
1167     ASSERT( compileResult == 0 );
1168     compileResult = regcomp(&re_realso,
1169            "GROUP *\\( *(([^ )])+)",
1170            REG_EXTENDED);
1171     ASSERT( compileResult == 0 );
1172 #   endif
1173
1174 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1175     if (RtsFlags.MiscFlags.linkerMemBase != 0) {
1176         // User-override for mmap_32bit_base
1177         mmap_32bit_base = (void*)RtsFlags.MiscFlags.linkerMemBase;
1178     }
1179 #endif
1180
1181 #if defined(mingw32_HOST_OS)
1182     /*
1183      * These two libraries cause problems when added to the static link,
1184      * but are necessary for resolving symbols in GHCi, hence we load
1185      * them manually here.
1186      */
1187     addDLL("msvcrt");
1188     addDLL("kernel32");
1189 #endif
1190 }
1191
1192 void
1193 exitLinker( void ) {
1194 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1195    if (linker_init_done == 1) {
1196       regfree(&re_invalid);
1197       regfree(&re_realso);
1198 #ifdef THREADED_RTS
1199       closeMutex(&dl_mutex);
1200 #endif
1201    }
1202 #endif
1203 }
1204
1205 /* -----------------------------------------------------------------------------
1206  *                  Loading DLL or .so dynamic libraries
1207  * -----------------------------------------------------------------------------
1208  *
1209  * Add a DLL from which symbols may be found.  In the ELF case, just
1210  * do RTLD_GLOBAL-style add, so no further messing around needs to
1211  * happen in order that symbols in the loaded .so are findable --
1212  * lookupSymbol() will subsequently see them by dlsym on the program's
1213  * dl-handle.  Returns NULL if success, otherwise ptr to an err msg.
1214  *
1215  * In the PEi386 case, open the DLLs and put handles to them in a
1216  * linked list.  When looking for a symbol, try all handles in the
1217  * list.  This means that we need to load even DLLs that are guaranteed
1218  * to be in the ghc.exe image already, just so we can get a handle
1219  * to give to loadSymbol, so that we can find the symbols.  For such
1220  * libraries, the LoadLibrary call should be a no-op except for returning
1221  * the handle.
1222  *
1223  */
1224
1225 #if defined(OBJFORMAT_PEi386)
1226 /* A record for storing handles into DLLs. */
1227
1228 typedef
1229    struct _OpenedDLL {
1230       char*              name;
1231       struct _OpenedDLL* next;
1232       HINSTANCE instance;
1233    }
1234    OpenedDLL;
1235
1236 /* A list thereof. */
1237 static OpenedDLL* opened_dlls = NULL;
1238 #endif
1239
1240 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1241
1242 static const char *
1243 internal_dlopen(const char *dll_name)
1244 {
1245    void *hdl;
1246    const char *errmsg;
1247    char *errmsg_copy;
1248
1249    // omitted: RTLD_NOW
1250    // see http://www.haskell.org/pipermail/cvs-ghc/2007-September/038570.html
1251    IF_DEBUG(linker,
1252       debugBelch("internal_dlopen: dll_name = '%s'\n", dll_name));
1253
1254    //-------------- Begin critical section ------------------
1255    // This critical section is necessary because dlerror() is not
1256    // required to be reentrant (see POSIX -- IEEE Std 1003.1-2008)
1257    // Also, the error message returned must be copied to preserve it
1258    // (see POSIX also)
1259
1260    ACQUIRE_LOCK(&dl_mutex);
1261    hdl = dlopen(dll_name, RTLD_LAZY | RTLD_GLOBAL);
1262
1263    errmsg = NULL;
1264    if (hdl == NULL) {
1265       /* dlopen failed; return a ptr to the error msg. */
1266       errmsg = dlerror();
1267       if (errmsg == NULL) errmsg = "addDLL: unknown error";
1268       errmsg_copy = stgMallocBytes(strlen(errmsg)+1, "addDLL");
1269       strcpy(errmsg_copy, errmsg);
1270       errmsg = errmsg_copy;
1271    }
1272    RELEASE_LOCK(&dl_mutex);
1273    //--------------- End critical section -------------------
1274
1275    return errmsg;
1276 }
1277 #  endif
1278
1279 const char *
1280 addDLL( char *dll_name )
1281 {
1282 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1283    /* ------------------- ELF DLL loader ------------------- */
1284
1285 #define NMATCH 5
1286    regmatch_t match[NMATCH];
1287    const char *errmsg;
1288    FILE* fp;
1289    size_t match_length;
1290 #define MAXLINE 1000
1291    char line[MAXLINE];
1292    int result;
1293
1294    initLinker();
1295
1296    IF_DEBUG(linker, debugBelch("addDLL: dll_name = '%s'\n", dll_name));
1297    errmsg = internal_dlopen(dll_name);
1298
1299    if (errmsg == NULL) {
1300       return NULL;
1301    }
1302
1303    // GHC Trac ticket #2615
1304    // On some systems (e.g., Gentoo Linux) dynamic files (e.g. libc.so)
1305    // contain linker scripts rather than ELF-format object code. This
1306    // code handles the situation by recognizing the real object code
1307    // file name given in the linker script.
1308    //
1309    // If an "invalid ELF header" error occurs, it is assumed that the
1310    // .so file contains a linker script instead of ELF object code.
1311    // In this case, the code looks for the GROUP ( ... ) linker
1312    // directive. If one is found, the first file name inside the
1313    // parentheses is treated as the name of a dynamic library and the
1314    // code attempts to dlopen that file. If this is also unsuccessful,
1315    // an error message is returned.
1316
1317    // see if the error message is due to an invalid ELF header
1318    IF_DEBUG(linker, debugBelch("errmsg = '%s'\n", errmsg));
1319    result = regexec(&re_invalid, errmsg, (size_t) NMATCH, match, 0);
1320    IF_DEBUG(linker, debugBelch("result = %i\n", result));
1321    if (result == 0) {
1322       // success -- try to read the named file as a linker script
1323       match_length = (size_t) stg_min((match[1].rm_eo - match[1].rm_so),
1324                                  MAXLINE-1);
1325       strncpy(line, (errmsg+(match[1].rm_so)),match_length);
1326       line[match_length] = '\0'; // make sure string is null-terminated
1327       IF_DEBUG(linker, debugBelch ("file name = '%s'\n", line));
1328       if ((fp = fopen(line, "r")) == NULL) {
1329          return errmsg; // return original error if open fails
1330       }
1331       // try to find a GROUP ( ... ) command
1332       while (fgets(line, MAXLINE, fp) != NULL) {
1333          IF_DEBUG(linker, debugBelch("input line = %s", line));
1334          if (regexec(&re_realso, line, (size_t) NMATCH, match, 0) == 0) {
1335             // success -- try to dlopen the first named file
1336             IF_DEBUG(linker, debugBelch("match%s\n",""));
1337             line[match[1].rm_eo] = '\0';
1338             errmsg = internal_dlopen(line+match[1].rm_so);
1339             break;
1340          }
1341          // if control reaches here, no GROUP ( ... ) directive was found
1342          // and the original error message is returned to the caller
1343       }
1344       fclose(fp);
1345    }
1346    return errmsg;
1347
1348 #  elif defined(OBJFORMAT_PEi386)
1349    /* ------------------- Win32 DLL loader ------------------- */
1350
1351    char*      buf;
1352    OpenedDLL* o_dll;
1353    HINSTANCE  instance;
1354
1355    initLinker();
1356
1357    /* debugBelch("\naddDLL; dll_name = `%s'\n", dll_name); */
1358
1359    /* See if we've already got it, and ignore if so. */
1360    for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
1361       if (0 == strcmp(o_dll->name, dll_name))
1362          return NULL;
1363    }
1364
1365    /* The file name has no suffix (yet) so that we can try
1366       both foo.dll and foo.drv
1367
1368       The documentation for LoadLibrary says:
1369         If no file name extension is specified in the lpFileName
1370         parameter, the default library extension .dll is
1371         appended. However, the file name string can include a trailing
1372         point character (.) to indicate that the module name has no
1373         extension. */
1374
1375    buf = stgMallocBytes(strlen(dll_name) + 10, "addDLL");
1376    sprintf(buf, "%s.DLL", dll_name);
1377    instance = LoadLibrary(buf);
1378    if (instance == NULL) {
1379        if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1380        // KAA: allow loading of drivers (like winspool.drv)
1381        sprintf(buf, "%s.DRV", dll_name);
1382        instance = LoadLibrary(buf);
1383        if (instance == NULL) {
1384            if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1385            // #1883: allow loading of unix-style libfoo.dll DLLs
1386            sprintf(buf, "lib%s.DLL", dll_name);
1387            instance = LoadLibrary(buf);
1388            if (instance == NULL) {
1389                goto error;
1390            }
1391        }
1392    }
1393    stgFree(buf);
1394
1395    /* Add this DLL to the list of DLLs in which to search for symbols. */
1396    o_dll = stgMallocBytes( sizeof(OpenedDLL), "addDLL" );
1397    o_dll->name     = stgMallocBytes(1+strlen(dll_name), "addDLL");
1398    strcpy(o_dll->name, dll_name);
1399    o_dll->instance = instance;
1400    o_dll->next     = opened_dlls;
1401    opened_dlls     = o_dll;
1402
1403    return NULL;
1404
1405 error:
1406    stgFree(buf);
1407    sysErrorBelch(dll_name);
1408
1409    /* LoadLibrary failed; return a ptr to the error msg. */
1410    return "addDLL: could not load DLL";
1411
1412 #  else
1413    barf("addDLL: not implemented on this platform");
1414 #  endif
1415 }
1416
1417 /* -----------------------------------------------------------------------------
1418  * insert a stable symbol in the hash table
1419  */
1420
1421 void
1422 insertStableSymbol(char* obj_name, char* key, StgPtr p)
1423 {
1424   ghciInsertStrHashTable(obj_name, stablehash, key, getStablePtr(p));
1425 }
1426
1427
1428 /* -----------------------------------------------------------------------------
1429  * insert a symbol in the hash table
1430  */
1431 void
1432 insertSymbol(char* obj_name, char* key, void* data)
1433 {
1434   ghciInsertStrHashTable(obj_name, symhash, key, data);
1435 }
1436
1437 /* -----------------------------------------------------------------------------
1438  * lookup a symbol in the hash table
1439  */
1440 void *
1441 lookupSymbol( char *lbl )
1442 {
1443     void *val;
1444     initLinker() ;
1445     ASSERT(symhash != NULL);
1446     val = lookupStrHashTable(symhash, lbl);
1447
1448     if (val == NULL) {
1449 #       if defined(OBJFORMAT_ELF)
1450         return dlsym(dl_prog_handle, lbl);
1451 #       elif defined(OBJFORMAT_MACHO)
1452 #       if HAVE_DLFCN_H
1453         /* On OS X 10.3 and later, we use dlsym instead of the old legacy
1454            interface.
1455
1456            HACK: On OS X, global symbols are prefixed with an underscore.
1457                  However, dlsym wants us to omit the leading underscore from the
1458                  symbol name. For now, we simply strip it off here (and ONLY
1459                  here).
1460         */
1461         ASSERT(lbl[0] == '_');
1462         return dlsym(dl_prog_handle, lbl+1);
1463 #       else
1464         if(NSIsSymbolNameDefined(lbl)) {
1465             NSSymbol symbol = NSLookupAndBindSymbol(lbl);
1466             return NSAddressOfSymbol(symbol);
1467         } else {
1468             return NULL;
1469         }
1470 #       endif /* HAVE_DLFCN_H */
1471 #       elif defined(OBJFORMAT_PEi386)
1472         void* sym;
1473
1474         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1475         if (sym != NULL) { return sym; };
1476
1477         // Also try looking up the symbol without the @N suffix.  Some
1478         // DLLs have the suffixes on their symbols, some don't.
1479         zapTrailingAtSign ( (unsigned char*)lbl );
1480         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1481         if (sym != NULL) { return sym; };
1482         return NULL;
1483
1484 #       else
1485         ASSERT(2+2 == 5);
1486         return NULL;
1487 #       endif
1488     } else {
1489         return val;
1490     }
1491 }
1492
1493 /* -----------------------------------------------------------------------------
1494  * Debugging aid: look in GHCi's object symbol tables for symbols
1495  * within DELTA bytes of the specified address, and show their names.
1496  */
1497 #ifdef DEBUG
1498 void ghci_enquire ( char* addr );
1499
1500 void ghci_enquire ( char* addr )
1501 {
1502    int   i;
1503    char* sym;
1504    char* a;
1505    const int DELTA = 64;
1506    ObjectCode* oc;
1507
1508    initLinker();
1509
1510    for (oc = objects; oc; oc = oc->next) {
1511       for (i = 0; i < oc->n_symbols; i++) {
1512          sym = oc->symbols[i];
1513          if (sym == NULL) continue;
1514          a = NULL;
1515          if (a == NULL) {
1516             a = lookupStrHashTable(symhash, sym);
1517          }
1518          if (a == NULL) {
1519              // debugBelch("ghci_enquire: can't find %s\n", sym);
1520          }
1521          else if (addr-DELTA <= a && a <= addr+DELTA) {
1522             debugBelch("%p + %3d  ==  `%s'\n", addr, (int)(a - addr), sym);
1523          }
1524       }
1525    }
1526 }
1527 #endif
1528
1529 #ifdef USE_MMAP
1530 #define ROUND_UP(x,size) ((x + size - 1) & ~(size - 1))
1531
1532 static void *
1533 mmapForLinker (size_t bytes, nat flags, int fd)
1534 {
1535    void *map_addr = NULL;
1536    void *result;
1537    int pagesize, size;
1538    static nat fixed = 0;
1539
1540    pagesize = getpagesize();
1541    size = ROUND_UP(bytes, pagesize);
1542
1543 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1544 mmap_again:
1545
1546    if (mmap_32bit_base != 0) {
1547        map_addr = mmap_32bit_base;
1548    }
1549 #endif
1550
1551    result = mmap(map_addr, size, PROT_EXEC|PROT_READ|PROT_WRITE,
1552                     MAP_PRIVATE|TRY_MAP_32BIT|fixed|flags, fd, 0);
1553
1554    if (result == MAP_FAILED) {
1555        sysErrorBelch("mmap %lu bytes at %p",(lnat)size,map_addr);
1556        errorBelch("Try specifying an address with +RTS -xm<addr> -RTS");
1557        stg_exit(EXIT_FAILURE);
1558    }
1559
1560 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1561    if (mmap_32bit_base != 0) {
1562        if (result == map_addr) {
1563            mmap_32bit_base = (StgWord8*)map_addr + size;
1564        } else {
1565            if ((W_)result > 0x80000000) {
1566                // oops, we were given memory over 2Gb
1567 #if defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS)
1568                // Some platforms require MAP_FIXED.  This is normally
1569                // a bad idea, because MAP_FIXED will overwrite
1570                // existing mappings.
1571                munmap(result,size);
1572                fixed = MAP_FIXED;
1573                goto mmap_again;
1574 #else
1575                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);
1576 #endif
1577            } else {
1578                // hmm, we were given memory somewhere else, but it's
1579                // still under 2Gb so we can use it.  Next time, ask
1580                // for memory right after the place we just got some
1581                mmap_32bit_base = (StgWord8*)result + size;
1582            }
1583        }
1584    } else {
1585        if ((W_)result > 0x80000000) {
1586            // oops, we were given memory over 2Gb
1587            // ... try allocating memory somewhere else?;
1588            debugTrace(DEBUG_linker,"MAP_32BIT didn't work; gave us %lu bytes at 0x%p", bytes, result);
1589            munmap(result, size);
1590
1591            // Set a base address and try again... (guess: 1Gb)
1592            mmap_32bit_base = (void*)0x40000000;
1593            goto mmap_again;
1594        }
1595    }
1596 #endif
1597
1598    return result;
1599 }
1600 #endif // USE_MMAP
1601
1602 /* -----------------------------------------------------------------------------
1603  * Load an obj (populate the global symbol table, but don't resolve yet)
1604  *
1605  * Returns: 1 if ok, 0 on error.
1606  */
1607 HsInt
1608 loadObj( char *path )
1609 {
1610    ObjectCode* oc;
1611    struct stat st;
1612    int r;
1613 #ifdef USE_MMAP
1614    int fd;
1615 #else
1616    FILE *f;
1617 #endif
1618    IF_DEBUG(linker, debugBelch("loadObj %s\n", path));
1619    initLinker();
1620
1621    /* debugBelch("loadObj %s\n", path ); */
1622
1623    /* Check that we haven't already loaded this object.
1624       Ignore requests to load multiple times */
1625    {
1626        ObjectCode *o;
1627        int is_dup = 0;
1628        for (o = objects; o; o = o->next) {
1629           if (0 == strcmp(o->fileName, path)) {
1630              is_dup = 1;
1631              break; /* don't need to search further */
1632           }
1633        }
1634        if (is_dup) {
1635           IF_DEBUG(linker, debugBelch(
1636             "GHCi runtime linker: warning: looks like you're trying to load the\n"
1637             "same object file twice:\n"
1638             "   %s\n"
1639             "GHCi will ignore this, but be warned.\n"
1640             , path));
1641           return 1; /* success */
1642        }
1643    }
1644
1645    oc = stgMallocBytes(sizeof(ObjectCode), "loadObj(oc)");
1646
1647 #  if defined(OBJFORMAT_ELF)
1648    oc->formatName = "ELF";
1649 #  elif defined(OBJFORMAT_PEi386)
1650    oc->formatName = "PEi386";
1651 #  elif defined(OBJFORMAT_MACHO)
1652    oc->formatName = "Mach-O";
1653 #  else
1654    stgFree(oc);
1655    barf("loadObj: not implemented on this platform");
1656 #  endif
1657
1658    r = stat(path, &st);
1659    if (r == -1) {
1660        IF_DEBUG(linker, debugBelch("File doesn't exist\n"));
1661        return 0;
1662    }
1663
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    oc->fileSize          = st.st_size;
1669    oc->symbols           = NULL;
1670    oc->sections          = NULL;
1671    oc->proddables        = NULL;
1672
1673    /* chain it onto the list of objects */
1674    oc->next              = objects;
1675    objects               = oc;
1676
1677 #ifdef USE_MMAP
1678    /* On many architectures malloc'd memory isn't executable, so we need to use mmap. */
1679
1680 #if defined(openbsd_HOST_OS)
1681    fd = open(path, O_RDONLY, S_IRUSR);
1682 #else
1683    fd = open(path, O_RDONLY);
1684 #endif
1685    if (fd == -1)
1686       barf("loadObj: can't open `%s'", path);
1687
1688    oc->image = mmapForLinker(oc->fileSize, 0, fd);
1689
1690    close(fd);
1691
1692 #else /* !USE_MMAP */
1693    /* load the image into memory */
1694    f = fopen(path, "rb");
1695    if (!f)
1696        barf("loadObj: can't read `%s'", path);
1697
1698 #   if defined(mingw32_HOST_OS)
1699         // TODO: We would like to use allocateExec here, but allocateExec
1700         //       cannot currently allocate blocks large enough.
1701     oc->image = VirtualAlloc(NULL, oc->fileSize, MEM_RESERVE | MEM_COMMIT,
1702                              PAGE_EXECUTE_READWRITE);
1703 #   elif defined(darwin_HOST_OS)
1704     // In a Mach-O .o file, all sections can and will be misaligned
1705     // if the total size of the headers is not a multiple of the
1706     // desired alignment. This is fine for .o files that only serve
1707     // as input for the static linker, but it's not fine for us,
1708     // as SSE (used by gcc for floating point) and Altivec require
1709     // 16-byte alignment.
1710     // We calculate the correct alignment from the header before
1711     // reading the file, and then we misalign oc->image on purpose so
1712     // that the actual sections end up aligned again.
1713    oc->misalignment = machoGetMisalignment(f);
1714    oc->image = stgMallocBytes(oc->fileSize + oc->misalignment, "loadObj(image)");
1715    oc->image += oc->misalignment;
1716 #  else
1717    oc->image = stgMallocBytes(oc->fileSize, "loadObj(image)");
1718 #  endif
1719
1720    {
1721        int n;
1722        n = fread ( oc->image, 1, oc->fileSize, f );
1723        if (n != oc->fileSize)
1724            barf("loadObj: error whilst reading `%s'", path);
1725    }
1726    fclose(f);
1727 #endif /* USE_MMAP */
1728
1729 #  if defined(OBJFORMAT_MACHO) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
1730    r = ocAllocateSymbolExtras_MachO ( oc );
1731    if (!r) {
1732        IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO failed\n"));
1733        return r;
1734    }
1735 #  elif defined(OBJFORMAT_ELF) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
1736    r = ocAllocateSymbolExtras_ELF ( oc );
1737    if (!r) {
1738        IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_ELF failed\n"));
1739        return r;
1740    }
1741 #endif
1742
1743    /* verify the in-memory image */
1744 #  if defined(OBJFORMAT_ELF)
1745    r = ocVerifyImage_ELF ( oc );
1746 #  elif defined(OBJFORMAT_PEi386)
1747    r = ocVerifyImage_PEi386 ( oc );
1748 #  elif defined(OBJFORMAT_MACHO)
1749    r = ocVerifyImage_MachO ( oc );
1750 #  else
1751    barf("loadObj: no verify method");
1752 #  endif
1753    if (!r) {
1754        IF_DEBUG(linker, debugBelch("ocVerifyImage_* failed\n"));
1755        return r;
1756    }
1757
1758    /* build the symbol list for this image */
1759 #  if defined(OBJFORMAT_ELF)
1760    r = ocGetNames_ELF ( oc );
1761 #  elif defined(OBJFORMAT_PEi386)
1762    r = ocGetNames_PEi386 ( oc );
1763 #  elif defined(OBJFORMAT_MACHO)
1764    r = ocGetNames_MachO ( oc );
1765 #  else
1766    barf("loadObj: no getNames method");
1767 #  endif
1768    if (!r) {
1769        IF_DEBUG(linker, debugBelch("ocGetNames_* failed\n"));
1770        return r;
1771    }
1772
1773    /* loaded, but not resolved yet */
1774    oc->status = OBJECT_LOADED;
1775
1776    return 1;
1777 }
1778
1779 /* -----------------------------------------------------------------------------
1780  * resolve all the currently unlinked objects in memory
1781  *
1782  * Returns: 1 if ok, 0 on error.
1783  */
1784 HsInt
1785 resolveObjs( void )
1786 {
1787     ObjectCode *oc;
1788     int r;
1789
1790     initLinker();
1791
1792     for (oc = objects; oc; oc = oc->next) {
1793         if (oc->status != OBJECT_RESOLVED) {
1794 #           if defined(OBJFORMAT_ELF)
1795             r = ocResolve_ELF ( oc );
1796 #           elif defined(OBJFORMAT_PEi386)
1797             r = ocResolve_PEi386 ( oc );
1798 #           elif defined(OBJFORMAT_MACHO)
1799             r = ocResolve_MachO ( oc );
1800 #           else
1801             barf("resolveObjs: not implemented on this platform");
1802 #           endif
1803             if (!r) { return r; }
1804             oc->status = OBJECT_RESOLVED;
1805         }
1806     }
1807     return 1;
1808 }
1809
1810 /* -----------------------------------------------------------------------------
1811  * delete an object from the pool
1812  */
1813 HsInt
1814 unloadObj( char *path )
1815 {
1816     ObjectCode *oc, *prev;
1817
1818     ASSERT(symhash != NULL);
1819     ASSERT(objects != NULL);
1820
1821     initLinker();
1822
1823     prev = NULL;
1824     for (oc = objects; oc; prev = oc, oc = oc->next) {
1825         if (!strcmp(oc->fileName,path)) {
1826
1827             /* Remove all the mappings for the symbols within this
1828              * object..
1829              */
1830             {
1831                 int i;
1832                 for (i = 0; i < oc->n_symbols; i++) {
1833                    if (oc->symbols[i] != NULL) {
1834                        removeStrHashTable(symhash, oc->symbols[i], NULL);
1835                    }
1836                 }
1837             }
1838
1839             if (prev == NULL) {
1840                 objects = oc->next;
1841             } else {
1842                 prev->next = oc->next;
1843             }
1844
1845             // We're going to leave this in place, in case there are
1846             // any pointers from the heap into it:
1847                 // #ifdef mingw32_HOST_OS
1848                 //  VirtualFree(oc->image);
1849                 // #else
1850             //  stgFree(oc->image);
1851             // #endif
1852             stgFree(oc->fileName);
1853             stgFree(oc->symbols);
1854             stgFree(oc->sections);
1855             stgFree(oc);
1856             return 1;
1857         }
1858     }
1859
1860     errorBelch("unloadObj: can't find `%s' to unload", path);
1861     return 0;
1862 }
1863
1864 /* -----------------------------------------------------------------------------
1865  * Sanity checking.  For each ObjectCode, maintain a list of address ranges
1866  * which may be prodded during relocation, and abort if we try and write
1867  * outside any of these.
1868  */
1869 static void addProddableBlock ( ObjectCode* oc, void* start, int size )
1870 {
1871    ProddableBlock* pb
1872       = stgMallocBytes(sizeof(ProddableBlock), "addProddableBlock");
1873    /* debugBelch("aPB %p %p %d\n", oc, start, size); */
1874    ASSERT(size > 0);
1875    pb->start      = start;
1876    pb->size       = size;
1877    pb->next       = oc->proddables;
1878    oc->proddables = pb;
1879 }
1880
1881 static void checkProddableBlock ( ObjectCode* oc, void* addr )
1882 {
1883    ProddableBlock* pb;
1884    for (pb = oc->proddables; pb != NULL; pb = pb->next) {
1885       char* s = (char*)(pb->start);
1886       char* e = s + pb->size - 1;
1887       char* a = (char*)addr;
1888       /* Assumes that the biggest fixup involves a 4-byte write.  This
1889          probably needs to be changed to 8 (ie, +7) on 64-bit
1890          plats. */
1891       if (a >= s && (a+3) <= e) return;
1892    }
1893    barf("checkProddableBlock: invalid fixup in runtime linker");
1894 }
1895
1896 /* -----------------------------------------------------------------------------
1897  * Section management.
1898  */
1899 static void addSection ( ObjectCode* oc, SectionKind kind,
1900                          void* start, void* end )
1901 {
1902    Section* s   = stgMallocBytes(sizeof(Section), "addSection");
1903    s->start     = start;
1904    s->end       = end;
1905    s->kind      = kind;
1906    s->next      = oc->sections;
1907    oc->sections = s;
1908    /*
1909    debugBelch("addSection: %p-%p (size %d), kind %d\n",
1910                    start, ((char*)end)-1, end - start + 1, kind );
1911    */
1912 }
1913
1914
1915 /* --------------------------------------------------------------------------
1916  * Symbol Extras.
1917  * This is about allocating a small chunk of memory for every symbol in the
1918  * object file. We make sure that the SymboLExtras are always "in range" of
1919  * limited-range PC-relative instructions on various platforms by allocating
1920  * them right next to the object code itself.
1921  */
1922
1923 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
1924
1925 /*
1926   ocAllocateSymbolExtras
1927
1928   Allocate additional space at the end of the object file image to make room
1929   for jump islands (powerpc, x86_64) and GOT entries (x86_64).
1930
1931   PowerPC relative branch instructions have a 24 bit displacement field.
1932   As PPC code is always 4-byte-aligned, this yields a +-32MB range.
1933   If a particular imported symbol is outside this range, we have to redirect
1934   the jump to a short piece of new code that just loads the 32bit absolute
1935   address and jumps there.
1936   On x86_64, PC-relative jumps and PC-relative accesses to the GOT are limited
1937   to 32 bits (+-2GB).
1938
1939   This function just allocates space for one SymbolExtra for every
1940   undefined symbol in the object file. The code for the jump islands is
1941   filled in by makeSymbolExtra below.
1942 */
1943
1944 static int ocAllocateSymbolExtras( ObjectCode* oc, int count, int first )
1945 {
1946 #ifdef USE_MMAP
1947   int pagesize, n, m;
1948 #endif
1949   int aligned;
1950 #ifndef USE_MMAP
1951   int misalignment = 0;
1952 #ifdef darwin_HOST_OS
1953   misalignment = oc->misalignment;
1954 #endif
1955 #endif
1956
1957   if( count > 0 )
1958   {
1959     // round up to the nearest 4
1960     aligned = (oc->fileSize + 3) & ~3;
1961
1962 #ifdef USE_MMAP
1963     pagesize = getpagesize();
1964     n = ROUND_UP( oc->fileSize, pagesize );
1965     m = ROUND_UP( aligned + sizeof (SymbolExtra) * count, pagesize );
1966
1967     /* we try to use spare space at the end of the last page of the
1968      * image for the jump islands, but if there isn't enough space
1969      * then we have to map some (anonymously, remembering MAP_32BIT).
1970      */
1971     if( m > n ) // we need to allocate more pages
1972     {
1973         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
1974                                           MAP_ANONYMOUS, -1);
1975     }
1976     else
1977     {
1978         oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
1979     }
1980 #else
1981     oc->image -= misalignment;
1982     oc->image = stgReallocBytes( oc->image,
1983                                  misalignment +
1984                                  aligned + sizeof (SymbolExtra) * count,
1985                                  "ocAllocateSymbolExtras" );
1986     oc->image += misalignment;
1987
1988     oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
1989 #endif /* USE_MMAP */
1990
1991     memset( oc->symbol_extras, 0, sizeof (SymbolExtra) * count );
1992   }
1993   else
1994     oc->symbol_extras = NULL;
1995
1996   oc->first_symbol_extra = first;
1997   oc->n_symbol_extras = count;
1998
1999   return 1;
2000 }
2001
2002 static SymbolExtra* makeSymbolExtra( ObjectCode* oc,
2003                                      unsigned long symbolNumber,
2004                                      unsigned long target )
2005 {
2006   SymbolExtra *extra;
2007
2008   ASSERT( symbolNumber >= oc->first_symbol_extra
2009         && symbolNumber - oc->first_symbol_extra < oc->n_symbol_extras);
2010
2011   extra = &oc->symbol_extras[symbolNumber - oc->first_symbol_extra];
2012
2013 #ifdef powerpc_HOST_ARCH
2014   // lis r12, hi16(target)
2015   extra->jumpIsland.lis_r12     = 0x3d80;
2016   extra->jumpIsland.hi_addr     = target >> 16;
2017
2018   // ori r12, r12, lo16(target)
2019   extra->jumpIsland.ori_r12_r12 = 0x618c;
2020   extra->jumpIsland.lo_addr     = target & 0xffff;
2021
2022   // mtctr r12
2023   extra->jumpIsland.mtctr_r12   = 0x7d8903a6;
2024
2025   // bctr
2026   extra->jumpIsland.bctr        = 0x4e800420;
2027 #endif
2028 #ifdef x86_64_HOST_ARCH
2029         // jmp *-14(%rip)
2030   static uint8_t jmp[] = { 0xFF, 0x25, 0xF2, 0xFF, 0xFF, 0xFF };
2031   extra->addr = target;
2032   memcpy(extra->jumpIsland, jmp, 6);
2033 #endif
2034
2035   return extra;
2036 }
2037
2038 #endif
2039
2040 /* --------------------------------------------------------------------------
2041  * PowerPC specifics (instruction cache flushing)
2042  * ------------------------------------------------------------------------*/
2043
2044 #ifdef powerpc_HOST_ARCH
2045 /*
2046    ocFlushInstructionCache
2047
2048    Flush the data & instruction caches.
2049    Because the PPC has split data/instruction caches, we have to
2050    do that whenever we modify code at runtime.
2051  */
2052
2053 static void ocFlushInstructionCache( ObjectCode *oc )
2054 {
2055     int n = (oc->fileSize + sizeof( SymbolExtra ) * oc->n_symbol_extras + 3) / 4;
2056     unsigned long *p = (unsigned long *) oc->image;
2057
2058     while( n-- )
2059     {
2060         __asm__ volatile ( "dcbf 0,%0\n\t"
2061                            "sync\n\t"
2062                            "icbi 0,%0"
2063                            :
2064                            : "r" (p)
2065                          );
2066         p++;
2067     }
2068     __asm__ volatile ( "sync\n\t"
2069                        "isync"
2070                      );
2071 }
2072 #endif
2073
2074 /* --------------------------------------------------------------------------
2075  * PEi386 specifics (Win32 targets)
2076  * ------------------------------------------------------------------------*/
2077
2078 /* The information for this linker comes from
2079       Microsoft Portable Executable
2080       and Common Object File Format Specification
2081       revision 5.1 January 1998
2082    which SimonM says comes from the MS Developer Network CDs.
2083
2084    It can be found there (on older CDs), but can also be found
2085    online at:
2086
2087       http://www.microsoft.com/hwdev/hardware/PECOFF.asp
2088
2089    (this is Rev 6.0 from February 1999).
2090
2091    Things move, so if that fails, try searching for it via
2092
2093       http://www.google.com/search?q=PE+COFF+specification
2094
2095    The ultimate reference for the PE format is the Winnt.h
2096    header file that comes with the Platform SDKs; as always,
2097    implementations will drift wrt their documentation.
2098
2099    A good background article on the PE format is Matt Pietrek's
2100    March 1994 article in Microsoft System Journal (MSJ)
2101    (Vol.9, No. 3): "Peering Inside the PE: A Tour of the
2102    Win32 Portable Executable File Format." The info in there
2103    has recently been updated in a two part article in
2104    MSDN magazine, issues Feb and March 2002,
2105    "Inside Windows: An In-Depth Look into the Win32 Portable
2106    Executable File Format"
2107
2108    John Levine's book "Linkers and Loaders" contains useful
2109    info on PE too.
2110 */
2111
2112
2113 #if defined(OBJFORMAT_PEi386)
2114
2115
2116
2117 typedef unsigned char  UChar;
2118 typedef unsigned short UInt16;
2119 typedef unsigned int   UInt32;
2120 typedef          int   Int32;
2121
2122
2123 typedef
2124    struct {
2125       UInt16 Machine;
2126       UInt16 NumberOfSections;
2127       UInt32 TimeDateStamp;
2128       UInt32 PointerToSymbolTable;
2129       UInt32 NumberOfSymbols;
2130       UInt16 SizeOfOptionalHeader;
2131       UInt16 Characteristics;
2132    }
2133    COFF_header;
2134
2135 #define sizeof_COFF_header 20
2136
2137
2138 typedef
2139    struct {
2140       UChar  Name[8];
2141       UInt32 VirtualSize;
2142       UInt32 VirtualAddress;
2143       UInt32 SizeOfRawData;
2144       UInt32 PointerToRawData;
2145       UInt32 PointerToRelocations;
2146       UInt32 PointerToLinenumbers;
2147       UInt16 NumberOfRelocations;
2148       UInt16 NumberOfLineNumbers;
2149       UInt32 Characteristics;
2150    }
2151    COFF_section;
2152
2153 #define sizeof_COFF_section 40
2154
2155
2156 typedef
2157    struct {
2158       UChar  Name[8];
2159       UInt32 Value;
2160       UInt16 SectionNumber;
2161       UInt16 Type;
2162       UChar  StorageClass;
2163       UChar  NumberOfAuxSymbols;
2164    }
2165    COFF_symbol;
2166
2167 #define sizeof_COFF_symbol 18
2168
2169
2170 typedef
2171    struct {
2172       UInt32 VirtualAddress;
2173       UInt32 SymbolTableIndex;
2174       UInt16 Type;
2175    }
2176    COFF_reloc;
2177
2178 #define sizeof_COFF_reloc 10
2179
2180
2181 /* From PE spec doc, section 3.3.2 */
2182 /* Note use of MYIMAGE_* since IMAGE_* are already defined in
2183    windows.h -- for the same purpose, but I want to know what I'm
2184    getting, here. */
2185 #define MYIMAGE_FILE_RELOCS_STRIPPED     0x0001
2186 #define MYIMAGE_FILE_EXECUTABLE_IMAGE    0x0002
2187 #define MYIMAGE_FILE_DLL                 0x2000
2188 #define MYIMAGE_FILE_SYSTEM              0x1000
2189 #define MYIMAGE_FILE_BYTES_REVERSED_HI   0x8000
2190 #define MYIMAGE_FILE_BYTES_REVERSED_LO   0x0080
2191 #define MYIMAGE_FILE_32BIT_MACHINE       0x0100
2192
2193 /* From PE spec doc, section 5.4.2 and 5.4.4 */
2194 #define MYIMAGE_SYM_CLASS_EXTERNAL       2
2195 #define MYIMAGE_SYM_CLASS_STATIC         3
2196 #define MYIMAGE_SYM_UNDEFINED            0
2197
2198 /* From PE spec doc, section 4.1 */
2199 #define MYIMAGE_SCN_CNT_CODE             0x00000020
2200 #define MYIMAGE_SCN_CNT_INITIALIZED_DATA 0x00000040
2201 #define MYIMAGE_SCN_LNK_NRELOC_OVFL      0x01000000
2202
2203 /* From PE spec doc, section 5.2.1 */
2204 #define MYIMAGE_REL_I386_DIR32           0x0006
2205 #define MYIMAGE_REL_I386_REL32           0x0014
2206
2207
2208 /* We use myindex to calculate array addresses, rather than
2209    simply doing the normal subscript thing.  That's because
2210    some of the above structs have sizes which are not
2211    a whole number of words.  GCC rounds their sizes up to a
2212    whole number of words, which means that the address calcs
2213    arising from using normal C indexing or pointer arithmetic
2214    are just plain wrong.  Sigh.
2215 */
2216 static UChar *
2217 myindex ( int scale, void* base, int index )
2218 {
2219    return
2220       ((UChar*)base) + scale * index;
2221 }
2222
2223
2224 static void
2225 printName ( UChar* name, UChar* strtab )
2226 {
2227    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2228       UInt32 strtab_offset = * (UInt32*)(name+4);
2229       debugBelch("%s", strtab + strtab_offset );
2230    } else {
2231       int i;
2232       for (i = 0; i < 8; i++) {
2233          if (name[i] == 0) break;
2234          debugBelch("%c", name[i] );
2235       }
2236    }
2237 }
2238
2239
2240 static void
2241 copyName ( UChar* name, UChar* strtab, UChar* dst, int dstSize )
2242 {
2243    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2244       UInt32 strtab_offset = * (UInt32*)(name+4);
2245       strncpy ( (char*)dst, (char*)strtab+strtab_offset, dstSize );
2246       dst[dstSize-1] = 0;
2247    } else {
2248       int i = 0;
2249       while (1) {
2250          if (i >= 8) break;
2251          if (name[i] == 0) break;
2252          dst[i] = name[i];
2253          i++;
2254       }
2255       dst[i] = 0;
2256    }
2257 }
2258
2259
2260 static UChar *
2261 cstring_from_COFF_symbol_name ( UChar* name, UChar* strtab )
2262 {
2263    UChar* newstr;
2264    /* If the string is longer than 8 bytes, look in the
2265       string table for it -- this will be correctly zero terminated.
2266    */
2267    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2268       UInt32 strtab_offset = * (UInt32*)(name+4);
2269       return ((UChar*)strtab) + strtab_offset;
2270    }
2271    /* Otherwise, if shorter than 8 bytes, return the original,
2272       which by defn is correctly terminated.
2273    */
2274    if (name[7]==0) return name;
2275    /* The annoying case: 8 bytes.  Copy into a temporary
2276       (which is never freed ...)
2277    */
2278    newstr = stgMallocBytes(9, "cstring_from_COFF_symbol_name");
2279    ASSERT(newstr);
2280    strncpy((char*)newstr,(char*)name,8);
2281    newstr[8] = 0;
2282    return newstr;
2283 }
2284
2285
2286 /* Just compares the short names (first 8 chars) */
2287 static COFF_section *
2288 findPEi386SectionCalled ( ObjectCode* oc,  UChar* name )
2289 {
2290    int i;
2291    COFF_header* hdr
2292       = (COFF_header*)(oc->image);
2293    COFF_section* sectab
2294       = (COFF_section*) (
2295            ((UChar*)(oc->image))
2296            + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2297         );
2298    for (i = 0; i < hdr->NumberOfSections; i++) {
2299       UChar* n1;
2300       UChar* n2;
2301       COFF_section* section_i
2302          = (COFF_section*)
2303            myindex ( sizeof_COFF_section, sectab, i );
2304       n1 = (UChar*) &(section_i->Name);
2305       n2 = name;
2306       if (n1[0]==n2[0] && n1[1]==n2[1] && n1[2]==n2[2] &&
2307           n1[3]==n2[3] && n1[4]==n2[4] && n1[5]==n2[5] &&
2308           n1[6]==n2[6] && n1[7]==n2[7])
2309          return section_i;
2310    }
2311
2312    return NULL;
2313 }
2314
2315
2316 static void
2317 zapTrailingAtSign ( UChar* sym )
2318 {
2319 #  define my_isdigit(c) ((c) >= '0' && (c) <= '9')
2320    int i, j;
2321    if (sym[0] == 0) return;
2322    i = 0;
2323    while (sym[i] != 0) i++;
2324    i--;
2325    j = i;
2326    while (j > 0 && my_isdigit(sym[j])) j--;
2327    if (j > 0 && sym[j] == '@' && j != i) sym[j] = 0;
2328 #  undef my_isdigit
2329 }
2330
2331 static void *
2332 lookupSymbolInDLLs ( UChar *lbl )
2333 {
2334     OpenedDLL* o_dll;
2335     void *sym;
2336
2337     for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
2338         /* debugBelch("look in %s for %s\n", o_dll->name, lbl); */
2339
2340         if (lbl[0] == '_') {
2341             /* HACK: if the name has an initial underscore, try stripping
2342                it off & look that up first. I've yet to verify whether there's
2343                a Rule that governs whether an initial '_' *should always* be
2344                stripped off when mapping from import lib name to the DLL name.
2345             */
2346             sym = GetProcAddress(o_dll->instance, (char*)(lbl+1));
2347             if (sym != NULL) {
2348                 /*debugBelch("found %s in %s\n", lbl+1,o_dll->name);*/
2349                 return sym;
2350             }
2351         }
2352         sym = GetProcAddress(o_dll->instance, (char*)lbl);
2353         if (sym != NULL) {
2354             /*debugBelch("found %s in %s\n", lbl,o_dll->name);*/
2355             return sym;
2356            }
2357     }
2358     return NULL;
2359 }
2360
2361
2362 static int
2363 ocVerifyImage_PEi386 ( ObjectCode* oc )
2364 {
2365    int i;
2366    UInt32 j, noRelocs;
2367    COFF_header*  hdr;
2368    COFF_section* sectab;
2369    COFF_symbol*  symtab;
2370    UChar*        strtab;
2371    /* debugBelch("\nLOADING %s\n", oc->fileName); */
2372    hdr = (COFF_header*)(oc->image);
2373    sectab = (COFF_section*) (
2374                ((UChar*)(oc->image))
2375                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2376             );
2377    symtab = (COFF_symbol*) (
2378                ((UChar*)(oc->image))
2379                + hdr->PointerToSymbolTable
2380             );
2381    strtab = ((UChar*)symtab)
2382             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2383
2384    if (hdr->Machine != 0x14c) {
2385       errorBelch("%s: Not x86 PEi386", oc->fileName);
2386       return 0;
2387    }
2388    if (hdr->SizeOfOptionalHeader != 0) {
2389       errorBelch("%s: PEi386 with nonempty optional header", oc->fileName);
2390       return 0;
2391    }
2392    if ( /* (hdr->Characteristics & MYIMAGE_FILE_RELOCS_STRIPPED) || */
2393         (hdr->Characteristics & MYIMAGE_FILE_EXECUTABLE_IMAGE) ||
2394         (hdr->Characteristics & MYIMAGE_FILE_DLL) ||
2395         (hdr->Characteristics & MYIMAGE_FILE_SYSTEM) ) {
2396       errorBelch("%s: Not a PEi386 object file", oc->fileName);
2397       return 0;
2398    }
2399    if ( (hdr->Characteristics & MYIMAGE_FILE_BYTES_REVERSED_HI)
2400         /* || !(hdr->Characteristics & MYIMAGE_FILE_32BIT_MACHINE) */ ) {
2401       errorBelch("%s: Invalid PEi386 word size or endiannness: %d",
2402                  oc->fileName,
2403                  (int)(hdr->Characteristics));
2404       return 0;
2405    }
2406    /* If the string table size is way crazy, this might indicate that
2407       there are more than 64k relocations, despite claims to the
2408       contrary.  Hence this test. */
2409    /* debugBelch("strtab size %d\n", * (UInt32*)strtab); */
2410 #if 0
2411    if ( (*(UInt32*)strtab) > 600000 ) {
2412       /* Note that 600k has no special significance other than being
2413          big enough to handle the almost-2MB-sized lumps that
2414          constitute HSwin32*.o. */
2415       debugBelch("PEi386 object has suspiciously large string table; > 64k relocs?");
2416       return 0;
2417    }
2418 #endif
2419
2420    /* No further verification after this point; only debug printing. */
2421    i = 0;
2422    IF_DEBUG(linker, i=1);
2423    if (i == 0) return 1;
2424
2425    debugBelch( "sectab offset = %d\n", ((UChar*)sectab) - ((UChar*)hdr) );
2426    debugBelch( "symtab offset = %d\n", ((UChar*)symtab) - ((UChar*)hdr) );
2427    debugBelch( "strtab offset = %d\n", ((UChar*)strtab) - ((UChar*)hdr) );
2428
2429    debugBelch("\n" );
2430    debugBelch( "Machine:           0x%x\n", (UInt32)(hdr->Machine) );
2431    debugBelch( "# sections:        %d\n",   (UInt32)(hdr->NumberOfSections) );
2432    debugBelch( "time/date:         0x%x\n", (UInt32)(hdr->TimeDateStamp) );
2433    debugBelch( "symtab offset:     %d\n",   (UInt32)(hdr->PointerToSymbolTable) );
2434    debugBelch( "# symbols:         %d\n",   (UInt32)(hdr->NumberOfSymbols) );
2435    debugBelch( "sz of opt hdr:     %d\n",   (UInt32)(hdr->SizeOfOptionalHeader) );
2436    debugBelch( "characteristics:   0x%x\n", (UInt32)(hdr->Characteristics) );
2437
2438    /* Print the section table. */
2439    debugBelch("\n" );
2440    for (i = 0; i < hdr->NumberOfSections; i++) {
2441       COFF_reloc* reltab;
2442       COFF_section* sectab_i
2443          = (COFF_section*)
2444            myindex ( sizeof_COFF_section, sectab, i );
2445       debugBelch(
2446                 "\n"
2447                 "section %d\n"
2448                 "     name `",
2449                 i
2450               );
2451       printName ( sectab_i->Name, strtab );
2452       debugBelch(
2453                 "'\n"
2454                 "    vsize %d\n"
2455                 "    vaddr %d\n"
2456                 "  data sz %d\n"
2457                 " data off %d\n"
2458                 "  num rel %d\n"
2459                 "  off rel %d\n"
2460                 "  ptr raw 0x%x\n",
2461                 sectab_i->VirtualSize,
2462                 sectab_i->VirtualAddress,
2463                 sectab_i->SizeOfRawData,
2464                 sectab_i->PointerToRawData,
2465                 sectab_i->NumberOfRelocations,
2466                 sectab_i->PointerToRelocations,
2467                 sectab_i->PointerToRawData
2468               );
2469       reltab = (COFF_reloc*) (
2470                   ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2471                );
2472
2473       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2474         /* If the relocation field (a short) has overflowed, the
2475          * real count can be found in the first reloc entry.
2476          *
2477          * See Section 4.1 (last para) of the PE spec (rev6.0).
2478          */
2479         COFF_reloc* rel = (COFF_reloc*)
2480                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2481         noRelocs = rel->VirtualAddress;
2482         j = 1;
2483       } else {
2484         noRelocs = sectab_i->NumberOfRelocations;
2485         j = 0;
2486       }
2487
2488       for (; j < noRelocs; j++) {
2489          COFF_symbol* sym;
2490          COFF_reloc* rel = (COFF_reloc*)
2491                            myindex ( sizeof_COFF_reloc, reltab, j );
2492          debugBelch(
2493                    "        type 0x%-4x   vaddr 0x%-8x   name `",
2494                    (UInt32)rel->Type,
2495                    rel->VirtualAddress );
2496          sym = (COFF_symbol*)
2497                myindex ( sizeof_COFF_symbol, symtab, rel->SymbolTableIndex );
2498          /* Hmm..mysterious looking offset - what's it for? SOF */
2499          printName ( sym->Name, strtab -10 );
2500          debugBelch("'\n" );
2501       }
2502
2503       debugBelch("\n" );
2504    }
2505    debugBelch("\n" );
2506    debugBelch("string table has size 0x%x\n", * (UInt32*)strtab );
2507    debugBelch("---START of string table---\n");
2508    for (i = 4; i < *(Int32*)strtab; i++) {
2509       if (strtab[i] == 0)
2510          debugBelch("\n"); else
2511          debugBelch("%c", strtab[i] );
2512    }
2513    debugBelch("--- END  of string table---\n");
2514
2515    debugBelch("\n" );
2516    i = 0;
2517    while (1) {
2518       COFF_symbol* symtab_i;
2519       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2520       symtab_i = (COFF_symbol*)
2521                  myindex ( sizeof_COFF_symbol, symtab, i );
2522       debugBelch(
2523                 "symbol %d\n"
2524                 "     name `",
2525                 i
2526               );
2527       printName ( symtab_i->Name, strtab );
2528       debugBelch(
2529                 "'\n"
2530                 "    value 0x%x\n"
2531                 "   1+sec# %d\n"
2532                 "     type 0x%x\n"
2533                 "   sclass 0x%x\n"
2534                 "     nAux %d\n",
2535                 symtab_i->Value,
2536                 (Int32)(symtab_i->SectionNumber),
2537                 (UInt32)symtab_i->Type,
2538                 (UInt32)symtab_i->StorageClass,
2539                 (UInt32)symtab_i->NumberOfAuxSymbols
2540               );
2541       i += symtab_i->NumberOfAuxSymbols;
2542       i++;
2543    }
2544
2545    debugBelch("\n" );
2546    return 1;
2547 }
2548
2549
2550 static int
2551 ocGetNames_PEi386 ( ObjectCode* oc )
2552 {
2553    COFF_header*  hdr;
2554    COFF_section* sectab;
2555    COFF_symbol*  symtab;
2556    UChar*        strtab;
2557
2558    UChar* sname;
2559    void*  addr;
2560    int    i;
2561
2562    hdr = (COFF_header*)(oc->image);
2563    sectab = (COFF_section*) (
2564                ((UChar*)(oc->image))
2565                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2566             );
2567    symtab = (COFF_symbol*) (
2568                ((UChar*)(oc->image))
2569                + hdr->PointerToSymbolTable
2570             );
2571    strtab = ((UChar*)(oc->image))
2572             + hdr->PointerToSymbolTable
2573             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2574
2575    /* Allocate space for any (local, anonymous) .bss sections. */
2576
2577    for (i = 0; i < hdr->NumberOfSections; i++) {
2578       UInt32 bss_sz;
2579       UChar* zspace;
2580       COFF_section* sectab_i
2581          = (COFF_section*)
2582            myindex ( sizeof_COFF_section, sectab, i );
2583       if (0 != strcmp((char*)sectab_i->Name, ".bss")) continue;
2584       /* sof 10/05: the PE spec text isn't too clear regarding what
2585        * the SizeOfRawData field is supposed to hold for object
2586        * file sections containing just uninitialized data -- for executables,
2587        * it is supposed to be zero; unclear what it's supposed to be
2588        * for object files. However, VirtualSize is guaranteed to be
2589        * zero for object files, which definitely suggests that SizeOfRawData
2590        * will be non-zero (where else would the size of this .bss section be
2591        * stored?) Looking at the COFF_section info for incoming object files,
2592        * this certainly appears to be the case.
2593        *
2594        * => I suspect we've been incorrectly handling .bss sections in (relocatable)
2595        * object files up until now. This turned out to bite us with ghc-6.4.1's use
2596        * of gcc-3.4.x, which has started to emit initially-zeroed-out local 'static'
2597        * variable decls into to the .bss section. (The specific function in Q which
2598        * triggered this is libraries/base/cbits/dirUtils.c:__hscore_getFolderPath())
2599        */
2600       if (sectab_i->VirtualSize == 0 && sectab_i->SizeOfRawData == 0) continue;
2601       /* This is a non-empty .bss section.  Allocate zeroed space for
2602          it, and set its PointerToRawData field such that oc->image +
2603          PointerToRawData == addr_of_zeroed_space.  */
2604       bss_sz = sectab_i->VirtualSize;
2605       if ( bss_sz < sectab_i->SizeOfRawData) { bss_sz = sectab_i->SizeOfRawData; }
2606       zspace = stgCallocBytes(1, bss_sz, "ocGetNames_PEi386(anonymous bss)");
2607       sectab_i->PointerToRawData = ((UChar*)zspace) - ((UChar*)(oc->image));
2608       addProddableBlock(oc, zspace, bss_sz);
2609       /* debugBelch("BSS anon section at 0x%x\n", zspace); */
2610    }
2611
2612    /* Copy section information into the ObjectCode. */
2613
2614    for (i = 0; i < hdr->NumberOfSections; i++) {
2615       UChar* start;
2616       UChar* end;
2617       UInt32 sz;
2618
2619       SectionKind kind
2620          = SECTIONKIND_OTHER;
2621       COFF_section* sectab_i
2622          = (COFF_section*)
2623            myindex ( sizeof_COFF_section, sectab, i );
2624       IF_DEBUG(linker, debugBelch("section name = %s\n", sectab_i->Name ));
2625
2626 #     if 0
2627       /* I'm sure this is the Right Way to do it.  However, the
2628          alternative of testing the sectab_i->Name field seems to
2629          work ok with Cygwin.
2630       */
2631       if (sectab_i->Characteristics & MYIMAGE_SCN_CNT_CODE ||
2632           sectab_i->Characteristics & MYIMAGE_SCN_CNT_INITIALIZED_DATA)
2633          kind = SECTIONKIND_CODE_OR_RODATA;
2634 #     endif
2635
2636       if (0==strcmp(".text",(char*)sectab_i->Name) ||
2637           0==strcmp(".rdata",(char*)sectab_i->Name)||
2638           0==strcmp(".rodata",(char*)sectab_i->Name))
2639          kind = SECTIONKIND_CODE_OR_RODATA;
2640       if (0==strcmp(".data",(char*)sectab_i->Name) ||
2641           0==strcmp(".bss",(char*)sectab_i->Name))
2642          kind = SECTIONKIND_RWDATA;
2643
2644       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
2645       sz = sectab_i->SizeOfRawData;
2646       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
2647
2648       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
2649       end   = start + sz - 1;
2650
2651       if (kind == SECTIONKIND_OTHER
2652           /* Ignore sections called which contain stabs debugging
2653              information. */
2654           && 0 != strcmp(".stab", (char*)sectab_i->Name)
2655           && 0 != strcmp(".stabstr", (char*)sectab_i->Name)
2656           /* ignore constructor section for now */
2657           && 0 != strcmp(".ctors", (char*)sectab_i->Name)
2658           /* ignore section generated from .ident */
2659           && 0!= strcmp("/4", (char*)sectab_i->Name)
2660           /* ignore unknown section that appeared in gcc 3.4.5(?) */
2661           && 0!= strcmp(".reloc", (char*)sectab_i->Name)
2662          ) {
2663          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", sectab_i->Name, oc->fileName);
2664          return 0;
2665       }
2666
2667       if (kind != SECTIONKIND_OTHER && end >= start) {
2668          addSection(oc, kind, start, end);
2669          addProddableBlock(oc, start, end - start + 1);
2670       }
2671    }
2672
2673    /* Copy exported symbols into the ObjectCode. */
2674
2675    oc->n_symbols = hdr->NumberOfSymbols;
2676    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
2677                                   "ocGetNames_PEi386(oc->symbols)");
2678    /* Call me paranoid; I don't care. */
2679    for (i = 0; i < oc->n_symbols; i++)
2680       oc->symbols[i] = NULL;
2681
2682    i = 0;
2683    while (1) {
2684       COFF_symbol* symtab_i;
2685       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2686       symtab_i = (COFF_symbol*)
2687                  myindex ( sizeof_COFF_symbol, symtab, i );
2688
2689       addr  = NULL;
2690
2691       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
2692           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
2693          /* This symbol is global and defined, viz, exported */
2694          /* for MYIMAGE_SYMCLASS_EXTERNAL
2695                 && !MYIMAGE_SYM_UNDEFINED,
2696             the address of the symbol is:
2697                 address of relevant section + offset in section
2698          */
2699          COFF_section* sectabent
2700             = (COFF_section*) myindex ( sizeof_COFF_section,
2701                                         sectab,
2702                                         symtab_i->SectionNumber-1 );
2703          addr = ((UChar*)(oc->image))
2704                 + (sectabent->PointerToRawData
2705                    + symtab_i->Value);
2706       }
2707       else
2708       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
2709           && symtab_i->Value > 0) {
2710          /* This symbol isn't in any section at all, ie, global bss.
2711             Allocate zeroed space for it. */
2712          addr = stgCallocBytes(1, symtab_i->Value,
2713                                "ocGetNames_PEi386(non-anonymous bss)");
2714          addSection(oc, SECTIONKIND_RWDATA, addr,
2715                         ((UChar*)addr) + symtab_i->Value - 1);
2716          addProddableBlock(oc, addr, symtab_i->Value);
2717          /* debugBelch("BSS      section at 0x%x\n", addr); */
2718       }
2719
2720       if (addr != NULL ) {
2721          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
2722          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
2723          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
2724          ASSERT(i >= 0 && i < oc->n_symbols);
2725          /* cstring_from_COFF_symbol_name always succeeds. */
2726          oc->symbols[i] = (char*)sname;
2727          ghciInsertStrHashTable(oc->fileName, symhash, (char*)sname, addr);
2728       } else {
2729 #        if 0
2730          debugBelch(
2731                    "IGNORING symbol %d\n"
2732                    "     name `",
2733                    i
2734                  );
2735          printName ( symtab_i->Name, strtab );
2736          debugBelch(
2737                    "'\n"
2738                    "    value 0x%x\n"
2739                    "   1+sec# %d\n"
2740                    "     type 0x%x\n"
2741                    "   sclass 0x%x\n"
2742                    "     nAux %d\n",
2743                    symtab_i->Value,
2744                    (Int32)(symtab_i->SectionNumber),
2745                    (UInt32)symtab_i->Type,
2746                    (UInt32)symtab_i->StorageClass,
2747                    (UInt32)symtab_i->NumberOfAuxSymbols
2748                  );
2749 #        endif
2750       }
2751
2752       i += symtab_i->NumberOfAuxSymbols;
2753       i++;
2754    }
2755
2756    return 1;
2757 }
2758
2759
2760 static int
2761 ocResolve_PEi386 ( ObjectCode* oc )
2762 {
2763    COFF_header*  hdr;
2764    COFF_section* sectab;
2765    COFF_symbol*  symtab;
2766    UChar*        strtab;
2767
2768    UInt32        A;
2769    UInt32        S;
2770    UInt32*       pP;
2771
2772    int i;
2773    UInt32 j, noRelocs;
2774
2775    /* ToDo: should be variable-sized?  But is at least safe in the
2776       sense of buffer-overrun-proof. */
2777    UChar symbol[1000];
2778    /* debugBelch("resolving for %s\n", oc->fileName); */
2779
2780    hdr = (COFF_header*)(oc->image);
2781    sectab = (COFF_section*) (
2782                ((UChar*)(oc->image))
2783                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2784             );
2785    symtab = (COFF_symbol*) (
2786                ((UChar*)(oc->image))
2787                + hdr->PointerToSymbolTable
2788             );
2789    strtab = ((UChar*)(oc->image))
2790             + hdr->PointerToSymbolTable
2791             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2792
2793    for (i = 0; i < hdr->NumberOfSections; i++) {
2794       COFF_section* sectab_i
2795          = (COFF_section*)
2796            myindex ( sizeof_COFF_section, sectab, i );
2797       COFF_reloc* reltab
2798          = (COFF_reloc*) (
2799               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2800            );
2801
2802       /* Ignore sections called which contain stabs debugging
2803          information. */
2804       if (0 == strcmp(".stab", (char*)sectab_i->Name)
2805           || 0 == strcmp(".stabstr", (char*)sectab_i->Name)
2806           || 0 == strcmp(".ctors", (char*)sectab_i->Name))
2807          continue;
2808
2809       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2810         /* If the relocation field (a short) has overflowed, the
2811          * real count can be found in the first reloc entry.
2812          *
2813          * See Section 4.1 (last para) of the PE spec (rev6.0).
2814          *
2815          * Nov2003 update: the GNU linker still doesn't correctly
2816          * handle the generation of relocatable object files with
2817          * overflown relocations. Hence the output to warn of potential
2818          * troubles.
2819          */
2820         COFF_reloc* rel = (COFF_reloc*)
2821                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2822         noRelocs = rel->VirtualAddress;
2823
2824         /* 10/05: we now assume (and check for) a GNU ld that is capable
2825          * of handling object files with (>2^16) of relocs.
2826          */
2827 #if 0
2828         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
2829                    noRelocs);
2830 #endif
2831         j = 1;
2832       } else {
2833         noRelocs = sectab_i->NumberOfRelocations;
2834         j = 0;
2835       }
2836
2837
2838       for (; j < noRelocs; j++) {
2839          COFF_symbol* sym;
2840          COFF_reloc* reltab_j
2841             = (COFF_reloc*)
2842               myindex ( sizeof_COFF_reloc, reltab, j );
2843
2844          /* the location to patch */
2845          pP = (UInt32*)(
2846                  ((UChar*)(oc->image))
2847                  + (sectab_i->PointerToRawData
2848                     + reltab_j->VirtualAddress
2849                     - sectab_i->VirtualAddress )
2850               );
2851          /* the existing contents of pP */
2852          A = *pP;
2853          /* the symbol to connect to */
2854          sym = (COFF_symbol*)
2855                myindex ( sizeof_COFF_symbol,
2856                          symtab, reltab_j->SymbolTableIndex );
2857          IF_DEBUG(linker,
2858                   debugBelch(
2859                             "reloc sec %2d num %3d:  type 0x%-4x   "
2860                             "vaddr 0x%-8x   name `",
2861                             i, j,
2862                             (UInt32)reltab_j->Type,
2863                             reltab_j->VirtualAddress );
2864                             printName ( sym->Name, strtab );
2865                             debugBelch("'\n" ));
2866
2867          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
2868             COFF_section* section_sym
2869                = findPEi386SectionCalled ( oc, sym->Name );
2870             if (!section_sym) {
2871                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
2872                return 0;
2873             }
2874             S = ((UInt32)(oc->image))
2875                 + (section_sym->PointerToRawData
2876                    + sym->Value);
2877          } else {
2878             copyName ( sym->Name, strtab, symbol, 1000-1 );
2879             S = (UInt32) lookupSymbol( (char*)symbol );
2880             if ((void*)S != NULL) goto foundit;
2881             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
2882             return 0;
2883            foundit:;
2884          }
2885          checkProddableBlock(oc, pP);
2886          switch (reltab_j->Type) {
2887             case MYIMAGE_REL_I386_DIR32:
2888                *pP = A + S;
2889                break;
2890             case MYIMAGE_REL_I386_REL32:
2891                /* Tricky.  We have to insert a displacement at
2892                   pP which, when added to the PC for the _next_
2893                   insn, gives the address of the target (S).
2894                   Problem is to know the address of the next insn
2895                   when we only know pP.  We assume that this
2896                   literal field is always the last in the insn,
2897                   so that the address of the next insn is pP+4
2898                   -- hence the constant 4.
2899                   Also I don't know if A should be added, but so
2900                   far it has always been zero.
2901
2902                   SOF 05/2005: 'A' (old contents of *pP) have been observed
2903                   to contain values other than zero (the 'wx' object file
2904                   that came with wxhaskell-0.9.4; dunno how it was compiled..).
2905                   So, add displacement to old value instead of asserting
2906                   A to be zero. Fixes wxhaskell-related crashes, and no other
2907                   ill effects have been observed.
2908
2909                   Update: the reason why we're seeing these more elaborate
2910                   relocations is due to a switch in how the NCG compiles SRTs
2911                   and offsets to them from info tables. SRTs live in .(ro)data,
2912                   while info tables live in .text, causing GAS to emit REL32/DISP32
2913                   relocations with non-zero values. Adding the displacement is
2914                   the right thing to do.
2915                */
2916                *pP = S - ((UInt32)pP) - 4 + A;
2917                break;
2918             default:
2919                debugBelch("%s: unhandled PEi386 relocation type %d",
2920                      oc->fileName, reltab_j->Type);
2921                return 0;
2922          }
2923
2924       }
2925    }
2926
2927    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
2928    return 1;
2929 }
2930
2931 #endif /* defined(OBJFORMAT_PEi386) */
2932
2933
2934 /* --------------------------------------------------------------------------
2935  * ELF specifics
2936  * ------------------------------------------------------------------------*/
2937
2938 #if defined(OBJFORMAT_ELF)
2939
2940 #define FALSE 0
2941 #define TRUE  1
2942
2943 #if defined(sparc_HOST_ARCH)
2944 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
2945 #elif defined(i386_HOST_ARCH)
2946 #  define ELF_TARGET_386    /* Used inside <elf.h> */
2947 #elif defined(x86_64_HOST_ARCH)
2948 #  define ELF_TARGET_X64_64
2949 #  define ELF_64BIT
2950 #endif
2951
2952 #if !defined(openbsd_HOST_OS)
2953 #  include <elf.h>
2954 #else
2955 /* openbsd elf has things in different places, with diff names */
2956 #  include <elf_abi.h>
2957 #  include <machine/reloc.h>
2958 #  define R_386_32    RELOC_32
2959 #  define R_386_PC32  RELOC_PC32
2960 #endif
2961
2962 /* If elf.h doesn't define it */
2963 #  ifndef R_X86_64_PC64
2964 #    define R_X86_64_PC64 24
2965 #  endif
2966
2967 /*
2968  * Define a set of types which can be used for both ELF32 and ELF64
2969  */
2970
2971 #ifdef ELF_64BIT
2972 #define ELFCLASS    ELFCLASS64
2973 #define Elf_Addr    Elf64_Addr
2974 #define Elf_Word    Elf64_Word
2975 #define Elf_Sword   Elf64_Sword
2976 #define Elf_Ehdr    Elf64_Ehdr
2977 #define Elf_Phdr    Elf64_Phdr
2978 #define Elf_Shdr    Elf64_Shdr
2979 #define Elf_Sym     Elf64_Sym
2980 #define Elf_Rel     Elf64_Rel
2981 #define Elf_Rela    Elf64_Rela
2982 #ifndef ELF_ST_TYPE
2983 #define ELF_ST_TYPE ELF64_ST_TYPE
2984 #endif
2985 #ifndef ELF_ST_BIND
2986 #define ELF_ST_BIND ELF64_ST_BIND
2987 #endif
2988 #ifndef ELF_R_TYPE
2989 #define ELF_R_TYPE  ELF64_R_TYPE
2990 #endif
2991 #ifndef ELF_R_SYM
2992 #define ELF_R_SYM   ELF64_R_SYM
2993 #endif
2994 #else
2995 #define ELFCLASS    ELFCLASS32
2996 #define Elf_Addr    Elf32_Addr
2997 #define Elf_Word    Elf32_Word
2998 #define Elf_Sword   Elf32_Sword
2999 #define Elf_Ehdr    Elf32_Ehdr
3000 #define Elf_Phdr    Elf32_Phdr
3001 #define Elf_Shdr    Elf32_Shdr
3002 #define Elf_Sym     Elf32_Sym
3003 #define Elf_Rel     Elf32_Rel
3004 #define Elf_Rela    Elf32_Rela
3005 #ifndef ELF_ST_TYPE
3006 #define ELF_ST_TYPE ELF32_ST_TYPE
3007 #endif
3008 #ifndef ELF_ST_BIND
3009 #define ELF_ST_BIND ELF32_ST_BIND
3010 #endif
3011 #ifndef ELF_R_TYPE
3012 #define ELF_R_TYPE  ELF32_R_TYPE
3013 #endif
3014 #ifndef ELF_R_SYM
3015 #define ELF_R_SYM   ELF32_R_SYM
3016 #endif
3017 #endif
3018
3019
3020 /*
3021  * Functions to allocate entries in dynamic sections.  Currently we simply
3022  * preallocate a large number, and we don't check if a entry for the given
3023  * target already exists (a linear search is too slow).  Ideally these
3024  * entries would be associated with symbols.
3025  */
3026
3027 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
3028 #define GOT_SIZE            0x20000
3029 #define FUNCTION_TABLE_SIZE 0x10000
3030 #define PLT_SIZE            0x08000
3031
3032 #ifdef ELF_NEED_GOT
3033 static Elf_Addr got[GOT_SIZE];
3034 static unsigned int gotIndex;
3035 static Elf_Addr gp_val = (Elf_Addr)got;
3036
3037 static Elf_Addr
3038 allocateGOTEntry(Elf_Addr target)
3039 {
3040    Elf_Addr *entry;
3041
3042    if (gotIndex >= GOT_SIZE)
3043       barf("Global offset table overflow");
3044
3045    entry = &got[gotIndex++];
3046    *entry = target;
3047    return (Elf_Addr)entry;
3048 }
3049 #endif
3050
3051 #ifdef ELF_FUNCTION_DESC
3052 typedef struct {
3053    Elf_Addr ip;
3054    Elf_Addr gp;
3055 } FunctionDesc;
3056
3057 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
3058 static unsigned int functionTableIndex;
3059
3060 static Elf_Addr
3061 allocateFunctionDesc(Elf_Addr target)
3062 {
3063    FunctionDesc *entry;
3064
3065    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
3066       barf("Function table overflow");
3067
3068    entry = &functionTable[functionTableIndex++];
3069    entry->ip = target;
3070    entry->gp = (Elf_Addr)gp_val;
3071    return (Elf_Addr)entry;
3072 }
3073
3074 static Elf_Addr
3075 copyFunctionDesc(Elf_Addr target)
3076 {
3077    FunctionDesc *olddesc = (FunctionDesc *)target;
3078    FunctionDesc *newdesc;
3079
3080    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
3081    newdesc->gp = olddesc->gp;
3082    return (Elf_Addr)newdesc;
3083 }
3084 #endif
3085
3086 #ifdef ELF_NEED_PLT
3087
3088 typedef struct {
3089    unsigned char code[sizeof(plt_code)];
3090 } PLTEntry;
3091
3092 static Elf_Addr
3093 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
3094 {
3095    PLTEntry *plt = (PLTEntry *)oc->plt;
3096    PLTEntry *entry;
3097
3098    if (oc->pltIndex >= PLT_SIZE)
3099       barf("Procedure table overflow");
3100
3101    entry = &plt[oc->pltIndex++];
3102    memcpy(entry->code, plt_code, sizeof(entry->code));
3103    PLT_RELOC(entry->code, target);
3104    return (Elf_Addr)entry;
3105 }
3106
3107 static unsigned int
3108 PLTSize(void)
3109 {
3110    return (PLT_SIZE * sizeof(PLTEntry));
3111 }
3112 #endif
3113
3114
3115 /*
3116  * Generic ELF functions
3117  */
3118
3119 static char *
3120 findElfSection ( void* objImage, Elf_Word sh_type )
3121 {
3122    char* ehdrC = (char*)objImage;
3123    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
3124    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
3125    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3126    char* ptr = NULL;
3127    int i;
3128
3129    for (i = 0; i < ehdr->e_shnum; i++) {
3130       if (shdr[i].sh_type == sh_type
3131           /* Ignore the section header's string table. */
3132           && i != ehdr->e_shstrndx
3133           /* Ignore string tables named .stabstr, as they contain
3134              debugging info. */
3135           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3136          ) {
3137          ptr = ehdrC + shdr[i].sh_offset;
3138          break;
3139       }
3140    }
3141    return ptr;
3142 }
3143
3144 static int
3145 ocVerifyImage_ELF ( ObjectCode* oc )
3146 {
3147    Elf_Shdr* shdr;
3148    Elf_Sym*  stab;
3149    int i, j, nent, nstrtab, nsymtabs;
3150    char* sh_strtab;
3151    char* strtab;
3152
3153    char*     ehdrC = (char*)(oc->image);
3154    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
3155
3156    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
3157        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
3158        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
3159        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
3160       errorBelch("%s: not an ELF object", oc->fileName);
3161       return 0;
3162    }
3163
3164    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
3165       errorBelch("%s: unsupported ELF format", oc->fileName);
3166       return 0;
3167    }
3168
3169    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
3170        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
3171    } else
3172    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
3173        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
3174    } else {
3175        errorBelch("%s: unknown endiannness", oc->fileName);
3176        return 0;
3177    }
3178
3179    if (ehdr->e_type != ET_REL) {
3180       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
3181       return 0;
3182    }
3183    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
3184
3185    IF_DEBUG(linker,debugBelch( "Architecture is " ));
3186    switch (ehdr->e_machine) {
3187       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
3188 #ifdef EM_SPARC32PLUS
3189       case EM_SPARC32PLUS:
3190 #endif
3191       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
3192 #ifdef EM_IA_64
3193       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
3194 #endif
3195       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
3196 #ifdef EM_X86_64
3197       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
3198 #elif defined(EM_AMD64)
3199       case EM_AMD64: IF_DEBUG(linker,debugBelch( "amd64" )); break;
3200 #endif
3201       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
3202                      errorBelch("%s: unknown architecture (e_machine == %d)"
3203                                 , oc->fileName, ehdr->e_machine);
3204                      return 0;
3205    }
3206
3207    IF_DEBUG(linker,debugBelch(
3208              "\nSection header table: start %ld, n_entries %d, ent_size %d\n",
3209              (long)ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
3210
3211    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
3212
3213    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3214
3215    if (ehdr->e_shstrndx == SHN_UNDEF) {
3216       errorBelch("%s: no section header string table", oc->fileName);
3217       return 0;
3218    } else {
3219       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
3220                           ehdr->e_shstrndx));
3221       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3222    }
3223
3224    for (i = 0; i < ehdr->e_shnum; i++) {
3225       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
3226       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
3227       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
3228       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
3229       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
3230                ehdrC + shdr[i].sh_offset,
3231                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
3232
3233       if (shdr[i].sh_type == SHT_REL) {
3234           IF_DEBUG(linker,debugBelch("Rel  " ));
3235       } else if (shdr[i].sh_type == SHT_RELA) {
3236           IF_DEBUG(linker,debugBelch("RelA " ));
3237       } else {
3238           IF_DEBUG(linker,debugBelch("     "));
3239       }
3240       if (sh_strtab) {
3241           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
3242       }
3243    }
3244
3245    IF_DEBUG(linker,debugBelch( "\nString tables" ));
3246    strtab = NULL;
3247    nstrtab = 0;
3248    for (i = 0; i < ehdr->e_shnum; i++) {
3249       if (shdr[i].sh_type == SHT_STRTAB
3250           /* Ignore the section header's string table. */
3251           && i != ehdr->e_shstrndx
3252           /* Ignore string tables named .stabstr, as they contain
3253              debugging info. */
3254           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3255          ) {
3256          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
3257          strtab = ehdrC + shdr[i].sh_offset;
3258          nstrtab++;
3259       }
3260    }
3261    if (nstrtab != 1) {
3262       errorBelch("%s: no string tables, or too many", oc->fileName);
3263       return 0;
3264    }
3265
3266    nsymtabs = 0;
3267    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
3268    for (i = 0; i < ehdr->e_shnum; i++) {
3269       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3270       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
3271       nsymtabs++;
3272       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3273       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3274       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%ld rem)\n",
3275                nent,
3276                (long)shdr[i].sh_size % sizeof(Elf_Sym)
3277              ));
3278       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
3279          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
3280          return 0;
3281       }
3282       for (j = 0; j < nent; j++) {
3283          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
3284          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
3285                              (int)stab[j].st_shndx,
3286                              (int)stab[j].st_size,
3287                              (char*)stab[j].st_value ));
3288
3289          IF_DEBUG(linker,debugBelch("type=" ));
3290          switch (ELF_ST_TYPE(stab[j].st_info)) {
3291             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
3292             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
3293             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
3294             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
3295             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
3296             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
3297          }
3298          IF_DEBUG(linker,debugBelch("  " ));
3299
3300          IF_DEBUG(linker,debugBelch("bind=" ));
3301          switch (ELF_ST_BIND(stab[j].st_info)) {
3302             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
3303             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
3304             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
3305             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
3306          }
3307          IF_DEBUG(linker,debugBelch("  " ));
3308
3309          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
3310       }
3311    }
3312
3313    if (nsymtabs == 0) {
3314       errorBelch("%s: didn't find any symbol tables", oc->fileName);
3315       return 0;
3316    }
3317
3318    return 1;
3319 }
3320
3321 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
3322 {
3323     *is_bss = FALSE;
3324
3325     if (hdr->sh_type == SHT_PROGBITS
3326         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
3327         /* .text-style section */
3328         return SECTIONKIND_CODE_OR_RODATA;
3329     }
3330
3331     if (hdr->sh_type == SHT_PROGBITS
3332             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3333             /* .data-style section */
3334             return SECTIONKIND_RWDATA;
3335     }
3336
3337     if (hdr->sh_type == SHT_PROGBITS
3338         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
3339         /* .rodata-style section */
3340         return SECTIONKIND_CODE_OR_RODATA;
3341     }
3342
3343     if (hdr->sh_type == SHT_NOBITS
3344         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3345         /* .bss-style section */
3346         *is_bss = TRUE;
3347         return SECTIONKIND_RWDATA;
3348     }
3349
3350     return SECTIONKIND_OTHER;
3351 }
3352
3353
3354 static int
3355 ocGetNames_ELF ( ObjectCode* oc )
3356 {
3357    int i, j, k, nent;
3358    Elf_Sym* stab;
3359
3360    char*     ehdrC    = (char*)(oc->image);
3361    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
3362    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
3363    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3364
3365    ASSERT(symhash != NULL);
3366
3367    if (!strtab) {
3368       errorBelch("%s: no strtab", oc->fileName);
3369       return 0;
3370    }
3371
3372    k = 0;
3373    for (i = 0; i < ehdr->e_shnum; i++) {
3374       /* Figure out what kind of section it is.  Logic derived from
3375          Figure 1.14 ("Special Sections") of the ELF document
3376          ("Portable Formats Specification, Version 1.1"). */
3377       int         is_bss = FALSE;
3378       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
3379
3380       if (is_bss && shdr[i].sh_size > 0) {
3381          /* This is a non-empty .bss section.  Allocate zeroed space for
3382             it, and set its .sh_offset field such that
3383             ehdrC + .sh_offset == addr_of_zeroed_space.  */
3384          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
3385                                        "ocGetNames_ELF(BSS)");
3386          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
3387          /*
3388          debugBelch("BSS section at 0x%x, size %d\n",
3389                          zspace, shdr[i].sh_size);
3390          */
3391       }
3392
3393       /* fill in the section info */
3394       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
3395          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
3396          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
3397                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
3398       }
3399
3400       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3401
3402       /* copy stuff into this module's object symbol table */
3403       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3404       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3405
3406       oc->n_symbols = nent;
3407       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3408                                    "ocGetNames_ELF(oc->symbols)");
3409
3410       for (j = 0; j < nent; j++) {
3411
3412          char  isLocal = FALSE; /* avoids uninit-var warning */
3413          char* ad      = NULL;
3414          char* nm      = strtab + stab[j].st_name;
3415          int   secno   = stab[j].st_shndx;
3416
3417          /* Figure out if we want to add it; if so, set ad to its
3418             address.  Otherwise leave ad == NULL. */
3419
3420          if (secno == SHN_COMMON) {
3421             isLocal = FALSE;
3422             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
3423             /*
3424             debugBelch("COMMON symbol, size %d name %s\n",
3425                             stab[j].st_size, nm);
3426             */
3427             /* Pointless to do addProddableBlock() for this area,
3428                since the linker should never poke around in it. */
3429          }
3430          else
3431          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
3432                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
3433               )
3434               /* and not an undefined symbol */
3435               && stab[j].st_shndx != SHN_UNDEF
3436               /* and not in a "special section" */
3437               && stab[j].st_shndx < SHN_LORESERVE
3438               &&
3439               /* and it's a not a section or string table or anything silly */
3440               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
3441                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
3442                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
3443               )
3444             ) {
3445             /* Section 0 is the undefined section, hence > and not >=. */
3446             ASSERT(secno > 0 && secno < ehdr->e_shnum);
3447             /*
3448             if (shdr[secno].sh_type == SHT_NOBITS) {
3449                debugBelch("   BSS symbol, size %d off %d name %s\n",
3450                                stab[j].st_size, stab[j].st_value, nm);
3451             }
3452             */
3453             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
3454             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
3455                isLocal = TRUE;
3456             } else {
3457 #ifdef ELF_FUNCTION_DESC
3458                /* dlsym() and the initialisation table both give us function
3459                 * descriptors, so to be consistent we store function descriptors
3460                 * in the symbol table */
3461                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
3462                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
3463 #endif
3464                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s\n",
3465                                       ad, oc->fileName, nm ));
3466                isLocal = FALSE;
3467             }
3468          }
3469
3470          /* And the decision is ... */
3471
3472          if (ad != NULL) {
3473             ASSERT(nm != NULL);
3474             oc->symbols[j] = nm;
3475             /* Acquire! */
3476             if (isLocal) {
3477                /* Ignore entirely. */
3478             } else {
3479                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
3480             }
3481          } else {
3482             /* Skip. */
3483             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
3484                                    strtab + stab[j].st_name ));
3485             /*
3486             debugBelch(
3487                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
3488                     (int)ELF_ST_BIND(stab[j].st_info),
3489                     (int)ELF_ST_TYPE(stab[j].st_info),
3490                     (int)stab[j].st_shndx,
3491                     strtab + stab[j].st_name
3492                    );
3493             */
3494             oc->symbols[j] = NULL;
3495          }
3496
3497       }
3498    }
3499
3500    return 1;
3501 }
3502
3503 /* Do ELF relocations which lack an explicit addend.  All x86-linux
3504    relocations appear to be of this form. */
3505 static int
3506 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
3507                          Elf_Shdr* shdr, int shnum,
3508                          Elf_Sym*  stab, char* strtab )
3509 {
3510    int j;
3511    char *symbol;
3512    Elf_Word* targ;
3513    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
3514    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
3515    int target_shndx = shdr[shnum].sh_info;
3516    int symtab_shndx = shdr[shnum].sh_link;
3517
3518    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
3519    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
3520    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
3521                           target_shndx, symtab_shndx ));
3522
3523    /* Skip sections that we're not interested in. */
3524    {
3525        int is_bss;
3526        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
3527        if (kind == SECTIONKIND_OTHER) {
3528            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
3529            return 1;
3530        }
3531    }
3532
3533    for (j = 0; j < nent; j++) {
3534       Elf_Addr offset = rtab[j].r_offset;
3535       Elf_Addr info   = rtab[j].r_info;
3536
3537       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
3538       Elf_Word* pP = (Elf_Word*)P;
3539       Elf_Addr  A  = *pP;
3540       Elf_Addr  S;
3541       void*     S_tmp;
3542       Elf_Addr  value;
3543       StgStablePtr stablePtr;
3544       StgPtr stableVal;
3545
3546       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
3547                              j, (void*)offset, (void*)info ));
3548       if (!info) {
3549          IF_DEBUG(linker,debugBelch( " ZERO" ));
3550          S = 0;
3551       } else {
3552          Elf_Sym sym = stab[ELF_R_SYM(info)];
3553          /* First see if it is a local symbol. */
3554          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
3555             /* Yes, so we can get the address directly from the ELF symbol
3556                table. */
3557             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
3558             S = (Elf_Addr)
3559                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
3560                        + stab[ELF_R_SYM(info)].st_value);
3561
3562          } else {
3563             symbol = strtab + sym.st_name;
3564             stablePtr = (StgStablePtr)lookupHashTable(stablehash, (StgWord)symbol);
3565             if (NULL == stablePtr) {
3566               /* No, so look up the name in our global table. */
3567               S_tmp = lookupSymbol( symbol );
3568               S = (Elf_Addr)S_tmp;
3569             } else {
3570               stableVal = deRefStablePtr( stablePtr );
3571               S_tmp = stableVal;
3572               S = (Elf_Addr)S_tmp;
3573             }
3574          }
3575          if (!S) {
3576             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3577             return 0;
3578          }
3579          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
3580       }
3581
3582       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
3583                              (void*)P, (void*)S, (void*)A ));
3584       checkProddableBlock ( oc, pP );
3585
3586       value = S + A;
3587
3588       switch (ELF_R_TYPE(info)) {
3589 #        ifdef i386_HOST_ARCH
3590          case R_386_32:   *pP = value;     break;
3591          case R_386_PC32: *pP = value - P; break;
3592 #        endif
3593          default:
3594             errorBelch("%s: unhandled ELF relocation(Rel) type %lu\n",
3595                   oc->fileName, (lnat)ELF_R_TYPE(info));
3596             return 0;
3597       }
3598
3599    }
3600    return 1;
3601 }
3602
3603 /* Do ELF relocations for which explicit addends are supplied.
3604    sparc-solaris relocations appear to be of this form. */
3605 static int
3606 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
3607                           Elf_Shdr* shdr, int shnum,
3608                           Elf_Sym*  stab, char* strtab )
3609 {
3610    int j;
3611    char *symbol = NULL;
3612    Elf_Addr targ;
3613    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
3614    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
3615    int target_shndx = shdr[shnum].sh_info;
3616    int symtab_shndx = shdr[shnum].sh_link;
3617
3618    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
3619    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
3620    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
3621                           target_shndx, symtab_shndx ));
3622
3623    for (j = 0; j < nent; j++) {
3624 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
3625       /* This #ifdef only serves to avoid unused-var warnings. */
3626       Elf_Addr  offset = rtab[j].r_offset;
3627       Elf_Addr  P      = targ + offset;
3628 #endif
3629       Elf_Addr  info   = rtab[j].r_info;
3630       Elf_Addr  A      = rtab[j].r_addend;
3631       Elf_Addr  S;
3632       void*     S_tmp;
3633       Elf_Addr  value;
3634 #     if defined(sparc_HOST_ARCH)
3635       Elf_Word* pP = (Elf_Word*)P;
3636       Elf_Word  w1, w2;
3637 #     elif defined(powerpc_HOST_ARCH)
3638       Elf_Sword delta;
3639 #     endif
3640
3641       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
3642                              j, (void*)offset, (void*)info,
3643                                 (void*)A ));
3644       if (!info) {
3645          IF_DEBUG(linker,debugBelch( " ZERO" ));
3646          S = 0;
3647       } else {
3648          Elf_Sym sym = stab[ELF_R_SYM(info)];
3649          /* First see if it is a local symbol. */
3650          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
3651             /* Yes, so we can get the address directly from the ELF symbol
3652                table. */
3653             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
3654             S = (Elf_Addr)
3655                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
3656                        + stab[ELF_R_SYM(info)].st_value);
3657 #ifdef ELF_FUNCTION_DESC
3658             /* Make a function descriptor for this function */
3659             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
3660                S = allocateFunctionDesc(S + A);
3661                A = 0;
3662             }
3663 #endif
3664          } else {
3665             /* No, so look up the name in our global table. */
3666             symbol = strtab + sym.st_name;
3667             S_tmp = lookupSymbol( symbol );
3668             S = (Elf_Addr)S_tmp;
3669
3670 #ifdef ELF_FUNCTION_DESC
3671             /* If a function, already a function descriptor - we would
3672                have to copy it to add an offset. */
3673             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
3674                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
3675 #endif
3676          }
3677          if (!S) {
3678            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3679            return 0;
3680          }
3681          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
3682       }
3683
3684       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
3685                                         (void*)P, (void*)S, (void*)A ));
3686       /* checkProddableBlock ( oc, (void*)P ); */
3687
3688       value = S + A;
3689
3690       switch (ELF_R_TYPE(info)) {
3691 #        if defined(sparc_HOST_ARCH)
3692          case R_SPARC_WDISP30:
3693             w1 = *pP & 0xC0000000;
3694             w2 = (Elf_Word)((value - P) >> 2);
3695             ASSERT((w2 & 0xC0000000) == 0);
3696             w1 |= w2;
3697             *pP = w1;
3698             break;
3699          case R_SPARC_HI22:
3700             w1 = *pP & 0xFFC00000;
3701             w2 = (Elf_Word)(value >> 10);
3702             ASSERT((w2 & 0xFFC00000) == 0);
3703             w1 |= w2;
3704             *pP = w1;
3705             break;
3706          case R_SPARC_LO10:
3707             w1 = *pP & ~0x3FF;
3708             w2 = (Elf_Word)(value & 0x3FF);
3709             ASSERT((w2 & ~0x3FF) == 0);
3710             w1 |= w2;
3711             *pP = w1;
3712             break;
3713
3714          /* According to the Sun documentation:
3715             R_SPARC_UA32
3716             This relocation type resembles R_SPARC_32, except it refers to an
3717             unaligned word. That is, the word to be relocated must be treated
3718             as four separate bytes with arbitrary alignment, not as a word
3719             aligned according to the architecture requirements.
3720          */
3721          case R_SPARC_UA32:
3722             w2  = (Elf_Word)value;
3723
3724             // SPARC doesn't do misaligned writes of 32 bit words,
3725             //       so we have to do this one byte-at-a-time.
3726             char *pPc   = (char*)pP;
3727             pPc[0]      = (char) ((Elf_Word)(w2 & 0xff000000) >> 24);
3728             pPc[1]      = (char) ((Elf_Word)(w2 & 0x00ff0000) >> 16);
3729             pPc[2]      = (char) ((Elf_Word)(w2 & 0x0000ff00) >> 8);
3730             pPc[3]      = (char) ((Elf_Word)(w2 & 0x000000ff));
3731             break;
3732
3733          case R_SPARC_32:
3734             w2 = (Elf_Word)value;
3735             *pP = w2;
3736             break;
3737 #        elif defined(powerpc_HOST_ARCH)
3738          case R_PPC_ADDR16_LO:
3739             *(Elf32_Half*) P = value;
3740             break;
3741
3742          case R_PPC_ADDR16_HI:
3743             *(Elf32_Half*) P = value >> 16;
3744             break;
3745
3746          case R_PPC_ADDR16_HA:
3747             *(Elf32_Half*) P = (value + 0x8000) >> 16;
3748             break;
3749
3750          case R_PPC_ADDR32:
3751             *(Elf32_Word *) P = value;
3752             break;
3753
3754          case R_PPC_REL32:
3755             *(Elf32_Word *) P = value - P;
3756             break;
3757
3758          case R_PPC_REL24:
3759             delta = value - P;
3760
3761             if( delta << 6 >> 6 != delta )
3762             {
3763                value = (Elf_Addr) (&makeSymbolExtra( oc, ELF_R_SYM(info), value )
3764                                         ->jumpIsland);
3765                delta = value - P;
3766
3767                if( value == 0 || delta << 6 >> 6 != delta )
3768                {
3769                   barf( "Unable to make SymbolExtra for #%d",
3770                         ELF_R_SYM(info) );
3771                   return 0;
3772                }
3773             }
3774
3775             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
3776                                           | (delta & 0x3fffffc);
3777             break;
3778 #        endif
3779
3780 #if x86_64_HOST_ARCH
3781       case R_X86_64_64:
3782           *(Elf64_Xword *)P = value;
3783           break;
3784
3785       case R_X86_64_PC32:
3786       {
3787 #if defined(ALWAYS_PIC)
3788           barf("R_X86_64_PC32 relocation, but ALWAYS_PIC.");
3789 #else
3790           StgInt64 off = value - P;
3791           if (off >= 0x7fffffffL || off < -0x80000000L) {
3792 #if X86_64_ELF_NONPIC_HACK
3793               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3794                                                 -> jumpIsland;
3795               off = pltAddress + A - P;
3796 #else
3797               barf("R_X86_64_PC32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
3798                    symbol, off, oc->fileName );
3799 #endif
3800           }
3801           *(Elf64_Word *)P = (Elf64_Word)off;
3802 #endif
3803           break;
3804       }
3805
3806       case R_X86_64_PC64:
3807       {
3808           StgInt64 off = value - P;
3809           *(Elf64_Word *)P = (Elf64_Word)off;
3810           break;
3811       }
3812
3813       case R_X86_64_32:
3814 #if defined(ALWAYS_PIC)
3815           barf("R_X86_64_32 relocation, but ALWAYS_PIC.");
3816 #else
3817           if (value >= 0x7fffffffL) {
3818 #if X86_64_ELF_NONPIC_HACK
3819               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3820                                                 -> jumpIsland;
3821               value = pltAddress + A;
3822 #else
3823               barf("R_X86_64_32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
3824                    symbol, value, oc->fileName );
3825 #endif
3826           }
3827           *(Elf64_Word *)P = (Elf64_Word)value;
3828 #endif
3829           break;
3830
3831       case R_X86_64_32S:
3832 #if defined(ALWAYS_PIC)
3833           barf("R_X86_64_32S relocation, but ALWAYS_PIC.");
3834 #else
3835           if ((StgInt64)value > 0x7fffffffL || (StgInt64)value < -0x80000000L) {
3836 #if X86_64_ELF_NONPIC_HACK
3837               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3838                                                 -> jumpIsland;
3839               value = pltAddress + A;
3840 #else
3841               barf("R_X86_64_32S relocation out of range: %s = %p\nRecompile %s with -fPIC.",
3842                    symbol, value, oc->fileName );
3843 #endif
3844           }
3845           *(Elf64_Sword *)P = (Elf64_Sword)value;
3846 #endif
3847           break;
3848
3849       case R_X86_64_GOTPCREL:
3850       {
3851           StgInt64 gotAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)->addr;
3852           StgInt64 off = gotAddress + A - P;
3853           *(Elf64_Word *)P = (Elf64_Word)off;
3854           break;
3855       }
3856
3857       case R_X86_64_PLT32:
3858       {
3859 #if defined(ALWAYS_PIC)
3860           barf("R_X86_64_PLT32 relocation, but ALWAYS_PIC.");
3861 #else
3862           StgInt64 off = value - P;
3863           if (off >= 0x7fffffffL || off < -0x80000000L) {
3864               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3865                                                     -> jumpIsland;
3866               off = pltAddress + A - P;
3867           }
3868           *(Elf64_Word *)P = (Elf64_Word)off;
3869 #endif
3870           break;
3871       }
3872 #endif
3873
3874          default:
3875             errorBelch("%s: unhandled ELF relocation(RelA) type %lu\n",
3876                   oc->fileName, (lnat)ELF_R_TYPE(info));
3877             return 0;
3878       }
3879
3880    }
3881    return 1;
3882 }
3883
3884 static int
3885 ocResolve_ELF ( ObjectCode* oc )
3886 {
3887    char *strtab;
3888    int   shnum, ok;
3889    Elf_Sym*  stab  = NULL;
3890    char*     ehdrC = (char*)(oc->image);
3891    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
3892    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3893
3894    /* first find "the" symbol table */
3895    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
3896
3897    /* also go find the string table */
3898    strtab = findElfSection ( ehdrC, SHT_STRTAB );
3899
3900    if (stab == NULL || strtab == NULL) {
3901       errorBelch("%s: can't find string or symbol table", oc->fileName);
3902       return 0;
3903    }
3904
3905    /* Process the relocation sections. */
3906    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
3907       if (shdr[shnum].sh_type == SHT_REL) {
3908          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
3909                                        shnum, stab, strtab );
3910          if (!ok) return ok;
3911       }
3912       else
3913       if (shdr[shnum].sh_type == SHT_RELA) {
3914          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
3915                                         shnum, stab, strtab );
3916          if (!ok) return ok;
3917       }
3918    }
3919
3920 #if defined(powerpc_HOST_ARCH)
3921    ocFlushInstructionCache( oc );
3922 #endif
3923
3924    return 1;
3925 }
3926
3927 /*
3928  * PowerPC & X86_64 ELF specifics
3929  */
3930
3931 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
3932
3933 static int ocAllocateSymbolExtras_ELF( ObjectCode *oc )
3934 {
3935   Elf_Ehdr *ehdr;
3936   Elf_Shdr* shdr;
3937   int i;
3938
3939   ehdr = (Elf_Ehdr *) oc->image;
3940   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
3941
3942   for( i = 0; i < ehdr->e_shnum; i++ )
3943     if( shdr[i].sh_type == SHT_SYMTAB )
3944       break;
3945
3946   if( i == ehdr->e_shnum )
3947   {
3948     errorBelch( "This ELF file contains no symtab" );
3949     return 0;
3950   }
3951
3952   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
3953   {
3954     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
3955       (int) shdr[i].sh_entsize, (int) sizeof( Elf_Sym ) );
3956
3957     return 0;
3958   }
3959
3960   return ocAllocateSymbolExtras( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
3961 }
3962
3963 #endif /* powerpc */
3964
3965 #endif /* ELF */
3966
3967 /* --------------------------------------------------------------------------
3968  * Mach-O specifics
3969  * ------------------------------------------------------------------------*/
3970
3971 #if defined(OBJFORMAT_MACHO)
3972
3973 /*
3974   Support for MachO linking on Darwin/MacOS X
3975   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
3976
3977   I hereby formally apologize for the hackish nature of this code.
3978   Things that need to be done:
3979   *) implement ocVerifyImage_MachO
3980   *) add still more sanity checks.
3981 */
3982
3983 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
3984 #define mach_header mach_header_64
3985 #define segment_command segment_command_64
3986 #define section section_64
3987 #define nlist nlist_64
3988 #endif
3989
3990 #ifdef powerpc_HOST_ARCH
3991 static int ocAllocateSymbolExtras_MachO(ObjectCode* oc)
3992 {
3993     struct mach_header *header = (struct mach_header *) oc->image;
3994     struct load_command *lc = (struct load_command *) (header + 1);
3995     unsigned i;
3996
3997     for( i = 0; i < header->ncmds; i++ )
3998     {
3999         if( lc->cmd == LC_SYMTAB )
4000         {
4001                 // Find out the first and last undefined external
4002                 // symbol, so we don't have to allocate too many
4003                 // jump islands.
4004             struct symtab_command *symLC = (struct symtab_command *) lc;
4005             unsigned min = symLC->nsyms, max = 0;
4006             struct nlist *nlist =
4007                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
4008                       : NULL;
4009             for(i=0;i<symLC->nsyms;i++)
4010             {
4011                 if(nlist[i].n_type & N_STAB)
4012                     ;
4013                 else if(nlist[i].n_type & N_EXT)
4014                 {
4015                     if((nlist[i].n_type & N_TYPE) == N_UNDF
4016                         && (nlist[i].n_value == 0))
4017                     {
4018                         if(i < min)
4019                             min = i;
4020                         if(i > max)
4021                             max = i;
4022                     }
4023                 }
4024             }
4025             if(max >= min)
4026                 return ocAllocateSymbolExtras(oc, max - min + 1, min);
4027
4028             break;
4029         }
4030
4031         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4032     }
4033     return ocAllocateSymbolExtras(oc,0,0);
4034 }
4035 #endif
4036 #ifdef x86_64_HOST_ARCH
4037 static int ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4038 {
4039     struct mach_header *header = (struct mach_header *) oc->image;
4040     struct load_command *lc = (struct load_command *) (header + 1);
4041     unsigned i;
4042
4043     for( i = 0; i < header->ncmds; i++ )
4044     {
4045         if( lc->cmd == LC_SYMTAB )
4046         {
4047                 // Just allocate one entry for every symbol
4048             struct symtab_command *symLC = (struct symtab_command *) lc;
4049
4050             return ocAllocateSymbolExtras(oc, symLC->nsyms, 0);
4051         }
4052
4053         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4054     }
4055     return ocAllocateSymbolExtras(oc,0,0);
4056 }
4057 #endif
4058
4059 static int ocVerifyImage_MachO(ObjectCode* oc)
4060 {
4061     char *image = (char*) oc->image;
4062     struct mach_header *header = (struct mach_header*) image;
4063
4064 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4065     if(header->magic != MH_MAGIC_64) {
4066         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4067                    oc->fileName, MH_MAGIC_64, header->magic);
4068         return 0;
4069     }
4070 #else
4071     if(header->magic != MH_MAGIC) {
4072         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4073                    oc->fileName, MH_MAGIC, header->magic);
4074         return 0;
4075     }
4076 #endif
4077     // FIXME: do some more verifying here
4078     return 1;
4079 }
4080
4081 static int resolveImports(
4082     ObjectCode* oc,
4083     char *image,
4084     struct symtab_command *symLC,
4085     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
4086     unsigned long *indirectSyms,
4087     struct nlist *nlist)
4088 {
4089     unsigned i;
4090     size_t itemSize = 4;
4091
4092 #if i386_HOST_ARCH
4093     int isJumpTable = 0;
4094     if(!strcmp(sect->sectname,"__jump_table"))
4095     {
4096         isJumpTable = 1;
4097         itemSize = 5;
4098         ASSERT(sect->reserved2 == itemSize);
4099     }
4100 #endif
4101
4102     for(i=0; i*itemSize < sect->size;i++)
4103     {
4104         // according to otool, reserved1 contains the first index into the indirect symbol table
4105         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
4106         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4107         void *addr = NULL;
4108
4109         if((symbol->n_type & N_TYPE) == N_UNDF
4110             && (symbol->n_type & N_EXT) && (symbol->n_value != 0))
4111             addr = (void*) (symbol->n_value);
4112         else
4113             addr = lookupSymbol(nm);
4114         if(!addr)
4115         {
4116             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
4117             return 0;
4118         }
4119         ASSERT(addr);
4120
4121 #if i386_HOST_ARCH
4122         if(isJumpTable)
4123         {
4124             checkProddableBlock(oc,image + sect->offset + i*itemSize);
4125             *(image + sect->offset + i*itemSize) = 0xe9; // jmp
4126             *(unsigned*)(image + sect->offset + i*itemSize + 1)
4127                 = (char*)addr - (image + sect->offset + i*itemSize + 5);
4128         }
4129         else
4130 #endif
4131         {
4132             checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
4133             ((void**)(image + sect->offset))[i] = addr;
4134         }
4135     }
4136
4137     return 1;
4138 }
4139
4140 static unsigned long relocateAddress(
4141     ObjectCode* oc,
4142     int nSections,
4143     struct section* sections,
4144     unsigned long address)
4145 {
4146     int i;
4147     for(i = 0; i < nSections; i++)
4148     {
4149         if(sections[i].addr <= address
4150             && address < sections[i].addr + sections[i].size)
4151         {
4152             return (unsigned long)oc->image
4153                     + sections[i].offset + address - sections[i].addr;
4154         }
4155     }
4156     barf("Invalid Mach-O file:"
4157          "Address out of bounds while relocating object file");
4158     return 0;
4159 }
4160
4161 static int relocateSection(
4162     ObjectCode* oc,
4163     char *image,
4164     struct symtab_command *symLC, struct nlist *nlist,
4165     int nSections, struct section* sections, struct section *sect)
4166 {
4167     struct relocation_info *relocs;
4168     int i,n;
4169
4170     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
4171         return 1;
4172     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
4173         return 1;
4174     else if(!strcmp(sect->sectname,"__la_sym_ptr2"))
4175         return 1;
4176     else if(!strcmp(sect->sectname,"__la_sym_ptr3"))
4177         return 1;
4178
4179     n = sect->nreloc;
4180     relocs = (struct relocation_info*) (image + sect->reloff);
4181
4182     for(i=0;i<n;i++)
4183     {
4184 #ifdef x86_64_HOST_ARCH
4185         struct relocation_info *reloc = &relocs[i];
4186
4187         char    *thingPtr = image + sect->offset + reloc->r_address;
4188         uint64_t thing;
4189         /* We shouldn't need to initialise this, but gcc on OS X 64 bit
4190            complains that it may be used uninitialized if we don't */
4191         uint64_t value = 0;
4192         uint64_t baseValue;
4193         int type = reloc->r_type;
4194
4195         checkProddableBlock(oc,thingPtr);
4196         switch(reloc->r_length)
4197         {
4198             case 0:
4199                 thing = *(uint8_t*)thingPtr;
4200                 baseValue = (uint64_t)thingPtr + 1;
4201                 break;
4202             case 1:
4203                 thing = *(uint16_t*)thingPtr;
4204                 baseValue = (uint64_t)thingPtr + 2;
4205                 break;
4206             case 2:
4207                 thing = *(uint32_t*)thingPtr;
4208                 baseValue = (uint64_t)thingPtr + 4;
4209                 break;
4210             case 3:
4211                 thing = *(uint64_t*)thingPtr;
4212                 baseValue = (uint64_t)thingPtr + 8;
4213                 break;
4214             default:
4215                 barf("Unknown size.");
4216         }
4217
4218         if(type == X86_64_RELOC_GOT
4219            || type == X86_64_RELOC_GOT_LOAD)
4220         {
4221             ASSERT(reloc->r_extern);
4222             value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)->addr;
4223
4224             type = X86_64_RELOC_SIGNED;
4225         }
4226         else if(reloc->r_extern)
4227         {
4228             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4229             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4230             if(symbol->n_value == 0)
4231                 value = (uint64_t) lookupSymbol(nm);
4232             else
4233                 value = relocateAddress(oc, nSections, sections,
4234                                         symbol->n_value);
4235         }
4236         else
4237         {
4238             value = sections[reloc->r_symbolnum-1].offset
4239                   - sections[reloc->r_symbolnum-1].addr
4240                   + (uint64_t) image;
4241         }
4242
4243         if(type == X86_64_RELOC_BRANCH)
4244         {
4245             if((int32_t)(value - baseValue) != (int64_t)(value - baseValue))
4246             {
4247                 ASSERT(reloc->r_extern);
4248                 value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)
4249                                         -> jumpIsland;
4250             }
4251             ASSERT((int32_t)(value - baseValue) == (int64_t)(value - baseValue));
4252             type = X86_64_RELOC_SIGNED;
4253         }
4254
4255         switch(type)
4256         {
4257             case X86_64_RELOC_UNSIGNED:
4258                 ASSERT(!reloc->r_pcrel);
4259                 thing += value;
4260                 break;
4261             case X86_64_RELOC_SIGNED:
4262             case X86_64_RELOC_SIGNED_1:
4263             case X86_64_RELOC_SIGNED_2:
4264             case X86_64_RELOC_SIGNED_4:
4265                 ASSERT(reloc->r_pcrel);
4266                 thing += value - baseValue;
4267                 break;
4268             case X86_64_RELOC_SUBTRACTOR:
4269                 ASSERT(!reloc->r_pcrel);
4270                 thing -= value;
4271                 break;
4272             default:
4273                 barf("unkown relocation");
4274         }
4275
4276         switch(reloc->r_length)
4277         {
4278             case 0:
4279                 *(uint8_t*)thingPtr = thing;
4280                 break;
4281             case 1:
4282                 *(uint16_t*)thingPtr = thing;
4283                 break;
4284             case 2:
4285                 *(uint32_t*)thingPtr = thing;
4286                 break;
4287             case 3:
4288                 *(uint64_t*)thingPtr = thing;
4289                 break;
4290         }
4291 #else
4292         if(relocs[i].r_address & R_SCATTERED)
4293         {
4294             struct scattered_relocation_info *scat =
4295                 (struct scattered_relocation_info*) &relocs[i];
4296
4297             if(!scat->r_pcrel)
4298             {
4299                 if(scat->r_length == 2)
4300                 {
4301                     unsigned long word = 0;
4302                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
4303                     checkProddableBlock(oc,wordPtr);
4304
4305                     // Note on relocation types:
4306                     // i386 uses the GENERIC_RELOC_* types,
4307                     // while ppc uses special PPC_RELOC_* types.
4308                     // *_RELOC_VANILLA and *_RELOC_PAIR have the same value
4309                     // in both cases, all others are different.
4310                     // Therefore, we use GENERIC_RELOC_VANILLA
4311                     // and GENERIC_RELOC_PAIR instead of the PPC variants,
4312                     // and use #ifdefs for the other types.
4313
4314                     // Step 1: Figure out what the relocated value should be
4315                     if(scat->r_type == GENERIC_RELOC_VANILLA)
4316                     {
4317                         word = *wordPtr + (unsigned long) relocateAddress(
4318                                                                 oc,
4319                                                                 nSections,
4320                                                                 sections,
4321                                                                 scat->r_value)
4322                                         - scat->r_value;
4323                     }
4324 #ifdef powerpc_HOST_ARCH
4325                     else if(scat->r_type == PPC_RELOC_SECTDIFF
4326                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
4327                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
4328                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF
4329                         || scat->r_type == PPC_RELOC_LOCAL_SECTDIFF)
4330 #else
4331                     else if(scat->r_type == GENERIC_RELOC_SECTDIFF
4332                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
4333 #endif
4334                     {
4335                         struct scattered_relocation_info *pair =
4336                                 (struct scattered_relocation_info*) &relocs[i+1];
4337
4338                         if(!pair->r_scattered || pair->r_type != GENERIC_RELOC_PAIR)
4339                             barf("Invalid Mach-O file: "
4340                                  "RELOC_*_SECTDIFF not followed by RELOC_PAIR");
4341
4342                         word = (unsigned long)
4343                                (relocateAddress(oc, nSections, sections, scat->r_value)
4344                               - relocateAddress(oc, nSections, sections, pair->r_value));
4345                         i++;
4346                     }
4347 #ifdef powerpc_HOST_ARCH
4348                     else if(scat->r_type == PPC_RELOC_HI16
4349                          || scat->r_type == PPC_RELOC_LO16
4350                          || scat->r_type == PPC_RELOC_HA16
4351                          || scat->r_type == PPC_RELOC_LO14)
4352                     {   // these are generated by label+offset things
4353                         struct relocation_info *pair = &relocs[i+1];
4354                         if((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR)
4355                             barf("Invalid Mach-O file: "
4356                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
4357
4358                         if(scat->r_type == PPC_RELOC_LO16)
4359                         {
4360                             word = ((unsigned short*) wordPtr)[1];
4361                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4362                         }
4363                         else if(scat->r_type == PPC_RELOC_LO14)
4364                         {
4365                             barf("Unsupported Relocation: PPC_RELOC_LO14");
4366                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
4367                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4368                         }
4369                         else if(scat->r_type == PPC_RELOC_HI16)
4370                         {
4371                             word = ((unsigned short*) wordPtr)[1] << 16;
4372                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4373                         }
4374                         else if(scat->r_type == PPC_RELOC_HA16)
4375                         {
4376                             word = ((unsigned short*) wordPtr)[1] << 16;
4377                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
4378                         }
4379
4380
4381                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
4382                                                 - scat->r_value;
4383
4384                         i++;
4385                     }
4386  #endif
4387                     else
4388                     {
4389                         barf ("Don't know how to handle this Mach-O "
4390                               "scattered relocation entry: "
4391                               "object file %s; entry type %ld; "
4392                               "address %#lx\n",
4393                               oc->fileName, scat->r_type, scat->r_address);
4394                         return 0;
4395                      }
4396
4397 #ifdef powerpc_HOST_ARCH
4398                     if(scat->r_type == GENERIC_RELOC_VANILLA
4399                         || scat->r_type == PPC_RELOC_SECTDIFF)
4400 #else
4401                     if(scat->r_type == GENERIC_RELOC_VANILLA
4402                         || scat->r_type == GENERIC_RELOC_SECTDIFF
4403                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
4404 #endif
4405                     {
4406                         *wordPtr = word;
4407                     }
4408 #ifdef powerpc_HOST_ARCH
4409                     else if(scat->r_type == PPC_RELOC_LO16_SECTDIFF || scat->r_type == PPC_RELOC_LO16)
4410                     {
4411                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
4412                     }
4413                     else if(scat->r_type == PPC_RELOC_HI16_SECTDIFF || scat->r_type == PPC_RELOC_HI16)
4414                     {
4415                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
4416                     }
4417                     else if(scat->r_type == PPC_RELOC_HA16_SECTDIFF || scat->r_type == PPC_RELOC_HA16)
4418                     {
4419                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
4420                             + ((word & (1<<15)) ? 1 : 0);
4421                     }
4422 #endif
4423                 }
4424                 else
4425                 {
4426                     barf("Can't handle Mach-O scattered relocation entry "
4427                          "with this r_length tag: "
4428                          "object file %s; entry type %ld; "
4429                          "r_length tag %ld; address %#lx\n",
4430                          oc->fileName, scat->r_type, scat->r_length,
4431                          scat->r_address);
4432                     return 0;
4433                 }
4434             }
4435             else /* scat->r_pcrel */
4436             {
4437                 barf("Don't know how to handle *PC-relative* Mach-O "
4438                      "scattered relocation entry: "
4439                      "object file %s; entry type %ld; address %#lx\n",
4440                      oc->fileName, scat->r_type, scat->r_address);
4441                return 0;
4442             }
4443
4444         }
4445         else /* !(relocs[i].r_address & R_SCATTERED) */
4446         {
4447             struct relocation_info *reloc = &relocs[i];
4448             if(reloc->r_pcrel && !reloc->r_extern)
4449                 continue;
4450
4451             if(reloc->r_length == 2)
4452             {
4453                 unsigned long word = 0;
4454 #ifdef powerpc_HOST_ARCH
4455                 unsigned long jumpIsland = 0;
4456                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
4457                                                       // to avoid warning and to catch
4458                                                       // bugs.
4459 #endif
4460
4461                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
4462                 checkProddableBlock(oc,wordPtr);
4463
4464                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
4465                 {
4466                     word = *wordPtr;
4467                 }
4468 #ifdef powerpc_HOST_ARCH
4469                 else if(reloc->r_type == PPC_RELOC_LO16)
4470                 {
4471                     word = ((unsigned short*) wordPtr)[1];
4472                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4473                 }
4474                 else if(reloc->r_type == PPC_RELOC_HI16)
4475                 {
4476                     word = ((unsigned short*) wordPtr)[1] << 16;
4477                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4478                 }
4479                 else if(reloc->r_type == PPC_RELOC_HA16)
4480                 {
4481                     word = ((unsigned short*) wordPtr)[1] << 16;
4482                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
4483                 }
4484                 else if(reloc->r_type == PPC_RELOC_BR24)
4485                 {
4486                     word = *wordPtr;
4487                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
4488                 }
4489 #endif
4490                 else
4491                 {
4492                     barf("Can't handle this Mach-O relocation entry "
4493                          "(not scattered): "
4494                          "object file %s; entry type %ld; address %#lx\n",
4495                          oc->fileName, reloc->r_type, reloc->r_address);
4496                     return 0;
4497                 }
4498
4499                 if(!reloc->r_extern)
4500                 {
4501                     long delta =
4502                         sections[reloc->r_symbolnum-1].offset
4503                         - sections[reloc->r_symbolnum-1].addr
4504                         + ((long) image);
4505
4506                     word += delta;
4507                 }
4508                 else
4509                 {
4510                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
4511                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4512                     void *symbolAddress = lookupSymbol(nm);
4513                     if(!symbolAddress)
4514                     {
4515                         errorBelch("\nunknown symbol `%s'", nm);
4516                         return 0;
4517                     }
4518
4519                     if(reloc->r_pcrel)
4520                     {
4521 #ifdef powerpc_HOST_ARCH
4522                             // In the .o file, this should be a relative jump to NULL
4523                             // and we'll change it to a relative jump to the symbol
4524                         ASSERT(word + reloc->r_address == 0);
4525                         jumpIsland = (unsigned long)
4526                                         &makeSymbolExtra(oc,
4527                                                          reloc->r_symbolnum,
4528                                                          (unsigned long) symbolAddress)
4529                                          -> jumpIsland;
4530                         if(jumpIsland != 0)
4531                         {
4532                             offsetToJumpIsland = word + jumpIsland
4533                                 - (((long)image) + sect->offset - sect->addr);
4534                         }
4535 #endif
4536                         word += (unsigned long) symbolAddress
4537                                 - (((long)image) + sect->offset - sect->addr);
4538                     }
4539                     else
4540                     {
4541                         word += (unsigned long) symbolAddress;
4542                     }
4543                 }
4544
4545                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
4546                 {
4547                     *wordPtr = word;
4548                     continue;
4549                 }
4550 #ifdef powerpc_HOST_ARCH
4551                 else if(reloc->r_type == PPC_RELOC_LO16)
4552                 {
4553                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
4554                     i++; continue;
4555                 }
4556                 else if(reloc->r_type == PPC_RELOC_HI16)
4557                 {
4558                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
4559                     i++; continue;
4560                 }
4561                 else if(reloc->r_type == PPC_RELOC_HA16)
4562                 {
4563                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
4564                         + ((word & (1<<15)) ? 1 : 0);
4565                     i++; continue;
4566                 }
4567                 else if(reloc->r_type == PPC_RELOC_BR24)
4568                 {
4569                     if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
4570                     {
4571                         // The branch offset is too large.
4572                         // Therefore, we try to use a jump island.
4573                         if(jumpIsland == 0)
4574                         {
4575                             barf("unconditional relative branch out of range: "
4576                                  "no jump island available");
4577                         }
4578
4579                         word = offsetToJumpIsland;
4580                         if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
4581                             barf("unconditional relative branch out of range: "
4582                                  "jump island out of range");
4583                     }
4584                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
4585                     continue;
4586                 }
4587 #endif
4588             }
4589             else
4590             {
4591                  barf("Can't handle Mach-O relocation entry (not scattered) "
4592                       "with this r_length tag: "
4593                       "object file %s; entry type %ld; "
4594                       "r_length tag %ld; address %#lx\n",
4595                       oc->fileName, reloc->r_type, reloc->r_length,
4596                       reloc->r_address);
4597                  return 0;
4598             }
4599         }
4600 #endif
4601     }
4602     return 1;
4603 }
4604
4605 static int ocGetNames_MachO(ObjectCode* oc)
4606 {
4607     char *image = (char*) oc->image;
4608     struct mach_header *header = (struct mach_header*) image;
4609     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
4610     unsigned i,curSymbol = 0;
4611     struct segment_command *segLC = NULL;
4612     struct section *sections;
4613     struct symtab_command *symLC = NULL;
4614     struct nlist *nlist;
4615     unsigned long commonSize = 0;
4616     char    *commonStorage = NULL;
4617     unsigned long commonCounter;
4618
4619     for(i=0;i<header->ncmds;i++)
4620     {
4621         if(lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64)
4622             segLC = (struct segment_command*) lc;
4623         else if(lc->cmd == LC_SYMTAB)
4624             symLC = (struct symtab_command*) lc;
4625         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
4626     }
4627
4628     sections = (struct section*) (segLC+1);
4629     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
4630                   : NULL;
4631
4632     if(!segLC)
4633         barf("ocGetNames_MachO: no segment load command");
4634
4635     for(i=0;i<segLC->nsects;i++)
4636     {
4637         if(sections[i].size == 0)
4638             continue;
4639
4640         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
4641         {
4642             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
4643                                       "ocGetNames_MachO(common symbols)");
4644             sections[i].offset = zeroFillArea - image;
4645         }
4646
4647         if(!strcmp(sections[i].sectname,"__text"))
4648             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
4649                 (void*) (image + sections[i].offset),
4650                 (void*) (image + sections[i].offset + sections[i].size));
4651         else if(!strcmp(sections[i].sectname,"__const"))
4652             addSection(oc, SECTIONKIND_RWDATA,
4653                 (void*) (image + sections[i].offset),
4654                 (void*) (image + sections[i].offset + sections[i].size));
4655         else if(!strcmp(sections[i].sectname,"__data"))
4656             addSection(oc, SECTIONKIND_RWDATA,
4657                 (void*) (image + sections[i].offset),
4658                 (void*) (image + sections[i].offset + sections[i].size));
4659         else if(!strcmp(sections[i].sectname,"__bss")
4660                 || !strcmp(sections[i].sectname,"__common"))
4661             addSection(oc, SECTIONKIND_RWDATA,
4662                 (void*) (image + sections[i].offset),
4663                 (void*) (image + sections[i].offset + sections[i].size));
4664
4665         addProddableBlock(oc, (void*) (image + sections[i].offset),
4666                                         sections[i].size);
4667     }
4668
4669         // count external symbols defined here
4670     oc->n_symbols = 0;
4671     if(symLC)
4672     {
4673         for(i=0;i<symLC->nsyms;i++)
4674         {
4675             if(nlist[i].n_type & N_STAB)
4676                 ;
4677             else if(nlist[i].n_type & N_EXT)
4678             {
4679                 if((nlist[i].n_type & N_TYPE) == N_UNDF
4680                     && (nlist[i].n_value != 0))
4681                 {
4682                     commonSize += nlist[i].n_value;
4683                     oc->n_symbols++;
4684                 }
4685                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
4686                     oc->n_symbols++;
4687             }
4688         }
4689     }
4690     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
4691                                    "ocGetNames_MachO(oc->symbols)");
4692
4693     if(symLC)
4694     {
4695         for(i=0;i<symLC->nsyms;i++)
4696         {
4697             if(nlist[i].n_type & N_STAB)
4698                 ;
4699             else if((nlist[i].n_type & N_TYPE) == N_SECT)
4700             {
4701                 if(nlist[i].n_type & N_EXT)
4702                 {
4703                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
4704                     if((nlist[i].n_desc & N_WEAK_DEF) && lookupSymbol(nm))
4705                         ; // weak definition, and we already have a definition
4706                     else
4707                     {
4708                             ghciInsertStrHashTable(oc->fileName, symhash, nm,
4709                                                     image
4710                                                     + sections[nlist[i].n_sect-1].offset
4711                                                     - sections[nlist[i].n_sect-1].addr
4712                                                     + nlist[i].n_value);
4713                             oc->symbols[curSymbol++] = nm;
4714                     }
4715                 }
4716             }
4717         }
4718     }
4719
4720     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
4721     commonCounter = (unsigned long)commonStorage;
4722     if(symLC)
4723     {
4724         for(i=0;i<symLC->nsyms;i++)
4725         {
4726             if((nlist[i].n_type & N_TYPE) == N_UNDF
4727                     && (nlist[i].n_type & N_EXT) && (nlist[i].n_value != 0))
4728             {
4729                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
4730                 unsigned long sz = nlist[i].n_value;
4731
4732                 nlist[i].n_value = commonCounter;
4733
4734                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
4735                                        (void*)commonCounter);
4736                 oc->symbols[curSymbol++] = nm;
4737
4738                 commonCounter += sz;
4739             }
4740         }
4741     }
4742     return 1;
4743 }
4744
4745 static int ocResolve_MachO(ObjectCode* oc)
4746 {
4747     char *image = (char*) oc->image;
4748     struct mach_header *header = (struct mach_header*) image;
4749     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
4750     unsigned i;
4751     struct segment_command *segLC = NULL;
4752     struct section *sections;
4753     struct symtab_command *symLC = NULL;
4754     struct dysymtab_command *dsymLC = NULL;
4755     struct nlist *nlist;
4756
4757     for(i=0;i<header->ncmds;i++)
4758     {
4759         if(lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64)
4760             segLC = (struct segment_command*) lc;
4761         else if(lc->cmd == LC_SYMTAB)
4762             symLC = (struct symtab_command*) lc;
4763         else if(lc->cmd == LC_DYSYMTAB)
4764             dsymLC = (struct dysymtab_command*) lc;
4765         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
4766     }
4767
4768     sections = (struct section*) (segLC+1);
4769     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
4770                   : NULL;
4771
4772     if(dsymLC)
4773     {
4774         unsigned long *indirectSyms
4775             = (unsigned long*) (image + dsymLC->indirectsymoff);
4776
4777         for(i=0;i<segLC->nsects;i++)
4778         {
4779             if(    !strcmp(sections[i].sectname,"__la_symbol_ptr")
4780                 || !strcmp(sections[i].sectname,"__la_sym_ptr2")
4781                 || !strcmp(sections[i].sectname,"__la_sym_ptr3"))
4782             {
4783                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
4784                     return 0;
4785             }
4786             else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr")
4787                 ||  !strcmp(sections[i].sectname,"__pointers"))
4788             {
4789                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
4790                     return 0;
4791             }
4792             else if(!strcmp(sections[i].sectname,"__jump_table"))
4793             {
4794                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
4795                     return 0;
4796             }
4797         }
4798     }
4799
4800     for(i=0;i<segLC->nsects;i++)
4801     {
4802         if(!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
4803             return 0;
4804     }
4805
4806 #if defined (powerpc_HOST_ARCH)
4807     ocFlushInstructionCache( oc );
4808 #endif
4809
4810     return 1;
4811 }
4812
4813 #ifdef powerpc_HOST_ARCH
4814 /*
4815  * The Mach-O object format uses leading underscores. But not everywhere.
4816  * There is a small number of runtime support functions defined in
4817  * libcc_dynamic.a whose name does not have a leading underscore.
4818  * As a consequence, we can't get their address from C code.
4819  * We have to use inline assembler just to take the address of a function.
4820  * Yuck.
4821  */
4822
4823 extern void* symbolsWithoutUnderscore[];
4824
4825 static void machoInitSymbolsWithoutUnderscore()
4826 {
4827     void **p = symbolsWithoutUnderscore;
4828     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
4829
4830 #undef SymI_NeedsProto
4831 #define SymI_NeedsProto(x)  \
4832     __asm__ volatile(".long " # x);
4833
4834     RTS_MACHO_NOUNDERLINE_SYMBOLS
4835
4836     __asm__ volatile(".text");
4837
4838 #undef SymI_NeedsProto
4839 #define SymI_NeedsProto(x)  \
4840     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
4841
4842     RTS_MACHO_NOUNDERLINE_SYMBOLS
4843
4844 #undef SymI_NeedsProto
4845 }
4846 #endif
4847
4848 #ifndef USE_MMAP
4849 /*
4850  * Figure out by how much to shift the entire Mach-O file in memory
4851  * when loading so that its single segment ends up 16-byte-aligned
4852  */
4853 static int machoGetMisalignment( FILE * f )
4854 {
4855     struct mach_header header;
4856     int misalignment;
4857
4858     fread(&header, sizeof(header), 1, f);
4859     rewind(f);
4860
4861 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4862     if(header.magic != MH_MAGIC_64) {
4863         errorBelch("Bad magic. Expected: %08x, got: %08x.\n",
4864                    MH_MAGIC_64, header->magic);
4865         return 0;
4866     }
4867 #else
4868     if(header.magic != MH_MAGIC) {
4869         errorBelch("Bad magic. Expected: %08x, got: %08x.\n",
4870                    MH_MAGIC, header->magic);
4871         return 0;
4872     }
4873 #endif
4874
4875     misalignment = (header.sizeofcmds + sizeof(header))
4876                     & 0xF;
4877
4878     return misalignment ? (16 - misalignment) : 0;
4879 }
4880 #endif
4881
4882 #endif