rts/Linker.c

   1 /* -----------------------------------------------------------------------------
   2  *
   3  * (c) The GHC Team, 2000-2004
   4  *
   5  * RTS Object Linker
   6  *
   7  * ---------------------------------------------------------------------------*/
   8
   9 #if 0
  10 #include "PosixSource.h"
  11 #endif
  12
  13 /* Linux needs _GNU_SOURCE to get RTLD_DEFAULT from <dlfcn.h> and
  14    MREMAP_MAYMOVE from <sys/mman.h>.
  15  */
  16 #if defined(__linux__)  || defined(__GLIBC__)
  17 #define _GNU_SOURCE 1
  18 #endif
  19
  20 #include "Rts.h"
  21 #include "HsFFI.h"
  22
  23 #include "sm/Storage.h"
  24 #include "Stats.h"
  25 #include "Hash.h"
  26 #include "LinkerInternals.h"
  27 #include "RtsUtils.h"
  28 #include "Trace.h"
  29 #include "StgPrimFloat.h" // for __int_encodeFloat etc.
  30 #include "Stable.h"
  31
  32 #if !defined(mingw32_HOST_OS)
  33 #include "posix/Signals.h"
  34 #endif
  35
  36 // get protos for is*()
  37 #include <ctype.h>
  38
  39 #ifdef HAVE_SYS_TYPES_H
  40 #include <sys/types.h>
  41 #endif
  42
  43 #include <inttypes.h>
  44 #include <stdlib.h>
  45 #include <string.h>
  46 #include <stdio.h>
  47 #include <assert.h>
  48
  49 #ifdef HAVE_SYS_STAT_H
  50 #include <sys/stat.h>
  51 #endif
  52
  53 #if defined(HAVE_DLFCN_H)
  54 #include <dlfcn.h>
  55 #endif
  56
  57 #if defined(cygwin32_HOST_OS)
  58 #ifdef HAVE_DIRENT_H
  59 #include <dirent.h>
  60 #endif
  61
  62 #ifdef HAVE_SYS_TIME_H
  63 #include <sys/time.h>
  64 #endif
  65 #include <regex.h>
  66 #include <sys/fcntl.h>
  67 #include <sys/termios.h>
  68 #include <sys/utime.h>
  69 #include <sys/utsname.h>
  70 #include <sys/wait.h>
  71 #endif
  72
  73 #if defined(linux_HOST_OS    ) || defined(freebsd_HOST_OS) || \
  74     defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS ) || \
  75     defined(openbsd_HOST_OS  ) || \
  76     ( defined(darwin_HOST_OS ) && !defined(powerpc_HOST_ARCH) ) || \
  77     defined(kfreebsdgnu_HOST_OS)
  78 /* Don't use mmap on powerpc-apple-darwin as mmap doesn't support
  79  * reallocating but we need to allocate jump islands just after each
  80  * object images. Otherwise relative branches to jump islands can fail
  81  * due to 24-bits displacement overflow.
  82  */
  83 #define USE_MMAP
  84 #include <fcntl.h>
  85 #include <sys/mman.h>
  86
  87 #ifdef HAVE_UNISTD_H
  88 #include <unistd.h>
  89 #endif
  90
  91 #endif
  92
  93 #if defined(linux_HOST_OS) || defined(solaris2_HOST_OS) || defined(freebsd_HOST_OS) || defined(kfreebsdgnu_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  94 #  define OBJFORMAT_ELF
  95 #  include <regex.h>    // regex is already used by dlopen() so this is OK
  96                         // to use here without requiring an additional lib
  97 #elif defined(cygwin32_HOST_OS) || defined (mingw32_HOST_OS)
  98 #  define OBJFORMAT_PEi386
  99 #  include <windows.h>
 100 #  include <math.h>
 101 #elif defined(darwin_HOST_OS)
 102 #  define OBJFORMAT_MACHO
 103 #  include <regex.h>
 104 #  include <mach/machine.h>
 105 #  include <mach-o/fat.h>
 106 #  include <mach-o/loader.h>
 107 #  include <mach-o/nlist.h>
 108 #  include <mach-o/reloc.h>
 109 #if !defined(HAVE_DLFCN_H)
 110 #  include <mach-o/dyld.h>
 111 #endif
 112 #if defined(powerpc_HOST_ARCH)
 113 #  include <mach-o/ppc/reloc.h>
 114 #endif
 115 #if defined(x86_64_HOST_ARCH)
 116 #  include <mach-o/x86_64/reloc.h>
 117 #endif
 118 #endif
 119
 120 #if defined(x86_64_HOST_ARCH) && defined(darwin_HOST_OS)
 121 #define ALWAYS_PIC
 122 #endif
 123
 124 /* Hash table mapping symbol names to Symbol */
 125 static /*Str*/HashTable *symhash;
 126
 127 /* Hash table mapping symbol names to StgStablePtr */
 128 static /*Str*/HashTable *stablehash;
 129
 130 /* List of currently loaded objects */
 131 ObjectCode *objects = NULL;     /* initially empty */
 132
 133 static HsInt loadOc( ObjectCode* oc );
 134 static ObjectCode* mkOc( char *path, char *image, int imageSize,
 135                          char *archiveMemberName
 136 #ifndef USE_MMAP
 137 #ifdef darwin_HOST_OS
 138                        , int misalignment
 139 #endif
 140 #endif
 141                        );
 142
 143 #if defined(OBJFORMAT_ELF)
 144 static int ocVerifyImage_ELF    ( ObjectCode* oc );
 145 static int ocGetNames_ELF       ( ObjectCode* oc );
 146 static int ocResolve_ELF        ( ObjectCode* oc );
 147 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 148 static int ocAllocateSymbolExtras_ELF ( ObjectCode* oc );
 149 #endif
 150 #elif defined(OBJFORMAT_PEi386)
 151 static int ocVerifyImage_PEi386 ( ObjectCode* oc );
 152 static int ocGetNames_PEi386    ( ObjectCode* oc );
 153 static int ocResolve_PEi386     ( ObjectCode* oc );
 154 static void *lookupSymbolInDLLs ( unsigned char *lbl );
 155 static void zapTrailingAtSign   ( unsigned char *sym );
 156 #elif defined(OBJFORMAT_MACHO)
 157 static int ocVerifyImage_MachO    ( ObjectCode* oc );
 158 static int ocGetNames_MachO       ( ObjectCode* oc );
 159 static int ocResolve_MachO        ( ObjectCode* oc );
 160
 161 #ifndef USE_MMAP
 162 static int machoGetMisalignment( FILE * );
 163 #endif
 164 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 165 static int ocAllocateSymbolExtras_MachO ( ObjectCode* oc );
 166 #endif
 167 #ifdef powerpc_HOST_ARCH
 168 static void machoInitSymbolsWithoutUnderscore( void );
 169 #endif
 170 #endif
 171
 172 /* on x86_64 we have a problem with relocating symbol references in
 173  * code that was compiled without -fPIC.  By default, the small memory
 174  * model is used, which assumes that symbol references can fit in a
 175  * 32-bit slot.  The system dynamic linker makes this work for
 176  * references to shared libraries by either (a) allocating a jump
 177  * table slot for code references, or (b) moving the symbol at load
 178  * time (and copying its contents, if necessary) for data references.
 179  *
 180  * We unfortunately can't tell whether symbol references are to code
 181  * or data.  So for now we assume they are code (the vast majority
 182  * are), and allocate jump-table slots.  Unfortunately this will
 183  * SILENTLY generate crashing code for data references.  This hack is
 184  * enabled by X86_64_ELF_NONPIC_HACK.
 185  *
 186  * One workaround is to use shared Haskell libraries.  This is
 187  * coming.  Another workaround is to keep the static libraries but
 188  * compile them with -fPIC, because that will generate PIC references
 189  * to data which can be relocated.  The PIC code is still too green to
 190  * do this systematically, though.
 191  *
 192  * See bug #781
 193  * See thread http://www.haskell.org/pipermail/cvs-ghc/2007-September/038458.html
 194  *
 195  * Naming Scheme for Symbol Macros
 196  *
 197  * SymI_*: symbol is internal to the RTS. It resides in an object
 198  *         file/library that is statically.
 199  * SymE_*: symbol is external to the RTS library. It might be linked
 200  *         dynamically.
 201  *
 202  * Sym*_HasProto  : the symbol prototype is imported in an include file
 203  *                  or defined explicitly
 204  * Sym*_NeedsProto: the symbol is undefined and we add a dummy
 205  *                  default proto extern void sym(void);
 206  */
 207 #define X86_64_ELF_NONPIC_HACK 1
 208
 209 /* Link objects into the lower 2Gb on x86_64.  GHC assumes the
 210  * small memory model on this architecture (see gcc docs,
 211  * -mcmodel=small).
 212  *
 213  * MAP_32BIT not available on OpenBSD/amd64
 214  */
 215 #if defined(x86_64_HOST_ARCH) && defined(MAP_32BIT)
 216 #define TRY_MAP_32BIT MAP_32BIT
 217 #else
 218 #define TRY_MAP_32BIT 0
 219 #endif
 220
 221 /*
 222  * Due to the small memory model (see above), on x86_64 we have to map
 223  * all our non-PIC object files into the low 2Gb of the address space
 224  * (why 2Gb and not 4Gb?  Because all addresses must be reachable
 225  * using a 32-bit signed PC-relative offset). On Linux we can do this
 226  * using the MAP_32BIT flag to mmap(), however on other OSs
 227  * (e.g. *BSD, see #2063, and also on Linux inside Xen, see #2512), we
 228  * can't do this.  So on these systems, we have to pick a base address
 229  * in the low 2Gb of the address space and try to allocate memory from
 230  * there.
 231  *
 232  * We pick a default address based on the OS, but also make this
 233  * configurable via an RTS flag (+RTS -xm)
 234  */
 235 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
 236
 237 #if defined(MAP_32BIT)
 238 // Try to use MAP_32BIT
 239 #define MMAP_32BIT_BASE_DEFAULT 0
 240 #else
 241 // A guess: 1Gb.
 242 #define MMAP_32BIT_BASE_DEFAULT 0x40000000
 243 #endif
 244
 245 static void *mmap_32bit_base = (void *)MMAP_32BIT_BASE_DEFAULT;
 246 #endif
 247
 248 /* MAP_ANONYMOUS is MAP_ANON on some systems, e.g. OpenBSD */
 249 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
 250 #define MAP_ANONYMOUS MAP_ANON
 251 #endif
 252
 253 /* -----------------------------------------------------------------------------
 254  * Built-in symbols from the RTS
 255  */
 256
 257 typedef struct _RtsSymbolVal {
 258     char   *lbl;
 259     void   *addr;
 260 } RtsSymbolVal;
 261
 262 #define Maybe_Stable_Names      SymI_HasProto(stg_mkWeakzh)                     \
 263                                 SymI_HasProto(stg_mkWeakForeignEnvzh)           \
 264                                 SymI_HasProto(stg_makeStableNamezh)             \
 265                                 SymI_HasProto(stg_finalizzeWeakzh)
 266
 267 #if !defined (mingw32_HOST_OS)
 268 #define RTS_POSIX_ONLY_SYMBOLS                  \
 269       SymI_HasProto(__hscore_get_saved_termios) \
 270       SymI_HasProto(__hscore_set_saved_termios) \
 271       SymI_HasProto(shutdownHaskellAndSignal)   \
 272       SymI_HasProto(lockFile)                   \
 273       SymI_HasProto(unlockFile)                 \
 274       SymI_HasProto(signal_handlers)            \
 275       SymI_HasProto(stg_sig_install)            \
 276       SymI_HasProto(rtsTimerSignal)             \
 277       SymI_HasProto(atexit)                     \
 278       SymI_NeedsProto(nocldstop)
 279 #endif
 280
 281 #if defined (cygwin32_HOST_OS)
 282 #define RTS_MINGW_ONLY_SYMBOLS /**/
 283 /* Don't have the ability to read import libs / archives, so
 284  * we have to stupidly list a lot of what libcygwin.a
 285  * exports; sigh.
 286  */
 287 #define RTS_CYGWIN_ONLY_SYMBOLS                          \
 288       SymI_HasProto(regfree)                             \
 289       SymI_HasProto(regexec)                             \
 290       SymI_HasProto(regerror)                            \
 291       SymI_HasProto(regcomp)                             \
 292       SymI_HasProto(__errno)                             \
 293       SymI_HasProto(access)                              \
 294       SymI_HasProto(chmod)                               \
 295       SymI_HasProto(chdir)                               \
 296       SymI_HasProto(close)                               \
 297       SymI_HasProto(creat)                               \
 298       SymI_HasProto(dup)                                 \
 299       SymI_HasProto(dup2)                                \
 300       SymI_HasProto(fstat)                               \
 301       SymI_HasProto(fcntl)                               \
 302       SymI_HasProto(getcwd)                              \
 303       SymI_HasProto(getenv)                              \
 304       SymI_HasProto(lseek)                               \
 305       SymI_HasProto(open)                                \
 306       SymI_HasProto(fpathconf)                           \
 307       SymI_HasProto(pathconf)                            \
 308       SymI_HasProto(stat)                                \
 309       SymI_HasProto(pow)                                 \
 310       SymI_HasProto(tanh)                                \
 311       SymI_HasProto(cosh)                                \
 312       SymI_HasProto(sinh)                                \
 313       SymI_HasProto(atan)                                \
 314       SymI_HasProto(acos)                                \
 315       SymI_HasProto(asin)                                \
 316       SymI_HasProto(tan)                                 \
 317       SymI_HasProto(cos)                                 \
 318       SymI_HasProto(sin)                                 \
 319       SymI_HasProto(exp)                                 \
 320       SymI_HasProto(log)                                 \
 321       SymI_HasProto(sqrt)                                \
 322       SymI_HasProto(localtime_r)                         \
 323       SymI_HasProto(gmtime_r)                            \
 324       SymI_HasProto(mktime)                              \
 325       SymI_NeedsProto(_imp___tzname)                     \
 326       SymI_HasProto(gettimeofday)                        \
 327       SymI_HasProto(timezone)                            \
 328       SymI_HasProto(tcgetattr)                           \
 329       SymI_HasProto(tcsetattr)                           \
 330       SymI_HasProto(memcpy)                              \
 331       SymI_HasProto(memmove)                             \
 332       SymI_HasProto(realloc)                             \
 333       SymI_HasProto(malloc)                              \
 334       SymI_HasProto(free)                                \
 335       SymI_HasProto(fork)                                \
 336       SymI_HasProto(lstat)                               \
 337       SymI_HasProto(isatty)                              \
 338       SymI_HasProto(mkdir)                               \
 339       SymI_HasProto(opendir)                             \
 340       SymI_HasProto(readdir)                             \
 341       SymI_HasProto(rewinddir)                           \
 342       SymI_HasProto(closedir)                            \
 343       SymI_HasProto(link)                                \
 344       SymI_HasProto(mkfifo)                              \
 345       SymI_HasProto(pipe)                                \
 346       SymI_HasProto(read)                                \
 347       SymI_HasProto(rename)                              \
 348       SymI_HasProto(rmdir)                               \
 349       SymI_HasProto(select)                              \
 350       SymI_HasProto(system)                              \
 351       SymI_HasProto(write)                               \
 352       SymI_HasProto(strcmp)                              \
 353       SymI_HasProto(strcpy)                              \
 354       SymI_HasProto(strncpy)                             \
 355       SymI_HasProto(strerror)                            \
 356       SymI_HasProto(sigaddset)                           \
 357       SymI_HasProto(sigemptyset)                         \
 358       SymI_HasProto(sigprocmask)                         \
 359       SymI_HasProto(umask)                               \
 360       SymI_HasProto(uname)                               \
 361       SymI_HasProto(unlink)                              \
 362       SymI_HasProto(utime)                               \
 363       SymI_HasProto(waitpid)
 364
 365 #elif !defined(mingw32_HOST_OS)
 366 #define RTS_MINGW_ONLY_SYMBOLS /**/
 367 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 368 #else /* defined(mingw32_HOST_OS) */
 369 #define RTS_POSIX_ONLY_SYMBOLS  /**/
 370 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 371
 372 #if HAVE_GETTIMEOFDAY
 373 #define RTS_MINGW_GETTIMEOFDAY_SYM SymI_NeedsProto(gettimeofday)
 374 #else
 375 #define RTS_MINGW_GETTIMEOFDAY_SYM /**/
 376 #endif
 377
 378 #if HAVE___MINGW_VFPRINTF
 379 #define RTS___MINGW_VFPRINTF_SYM SymI_HasProto(__mingw_vfprintf)
 380 #else
 381 #define RTS___MINGW_VFPRINTF_SYM /**/
 382 #endif
 383
 384 /* These are statically linked from the mingw libraries into the ghc
 385    executable, so we have to employ this hack. */
 386 #define RTS_MINGW_ONLY_SYMBOLS                           \
 387       SymI_HasProto(stg_asyncReadzh)                     \
 388       SymI_HasProto(stg_asyncWritezh)                    \
 389       SymI_HasProto(stg_asyncDoProczh)                   \
 390       SymI_HasProto(memset)                              \
 391       SymI_HasProto(inet_ntoa)                           \
 392       SymI_HasProto(inet_addr)                           \
 393       SymI_HasProto(htonl)                               \
 394       SymI_HasProto(recvfrom)                            \
 395       SymI_HasProto(listen)                              \
 396       SymI_HasProto(bind)                                \
 397       SymI_HasProto(shutdown)                            \
 398       SymI_HasProto(connect)                             \
 399       SymI_HasProto(htons)                               \
 400       SymI_HasProto(ntohs)                               \
 401       SymI_HasProto(getservbyname)                       \
 402       SymI_HasProto(getservbyport)                       \
 403       SymI_HasProto(getprotobynumber)                    \
 404       SymI_HasProto(getprotobyname)                      \
 405       SymI_HasProto(gethostbyname)                       \
 406       SymI_HasProto(gethostbyaddr)                       \
 407       SymI_HasProto(gethostname)                         \
 408       SymI_HasProto(strcpy)                              \
 409       SymI_HasProto(strncpy)                             \
 410       SymI_HasProto(abort)                               \
 411       SymI_NeedsProto(_alloca)                           \
 412       SymI_HasProto(isxdigit)                            \
 413       SymI_HasProto(isupper)                             \
 414       SymI_HasProto(ispunct)                             \
 415       SymI_HasProto(islower)                             \
 416       SymI_HasProto(isspace)                             \
 417       SymI_HasProto(isprint)                             \
 418       SymI_HasProto(isdigit)                             \
 419       SymI_HasProto(iscntrl)                             \
 420       SymI_HasProto(isalpha)                             \
 421       SymI_HasProto(isalnum)                             \
 422       SymI_HasProto(isascii)                             \
 423       RTS___MINGW_VFPRINTF_SYM                           \
 424       SymI_HasProto(strcmp)                              \
 425       SymI_HasProto(memmove)                             \
 426       SymI_HasProto(realloc)                             \
 427       SymI_HasProto(malloc)                              \
 428       SymI_HasProto(pow)                                 \
 429       SymI_HasProto(tanh)                                \
 430       SymI_HasProto(cosh)                                \
 431       SymI_HasProto(sinh)                                \
 432       SymI_HasProto(atan)                                \
 433       SymI_HasProto(acos)                                \
 434       SymI_HasProto(asin)                                \
 435       SymI_HasProto(tan)                                 \
 436       SymI_HasProto(cos)                                 \
 437       SymI_HasProto(sin)                                 \
 438       SymI_HasProto(exp)                                 \
 439       SymI_HasProto(log)                                 \
 440       SymI_HasProto(sqrt)                                \
 441       SymI_HasProto(powf)                                \
 442       SymI_HasProto(tanhf)                               \
 443       SymI_HasProto(coshf)                               \
 444       SymI_HasProto(sinhf)                               \
 445       SymI_HasProto(atanf)                               \
 446       SymI_HasProto(acosf)                               \
 447       SymI_HasProto(asinf)                               \
 448       SymI_HasProto(tanf)                                \
 449       SymI_HasProto(cosf)                                \
 450       SymI_HasProto(sinf)                                \
 451       SymI_HasProto(expf)                                \
 452       SymI_HasProto(logf)                                \
 453       SymI_HasProto(sqrtf)                               \
 454       SymI_HasProto(erf)                                 \
 455       SymI_HasProto(erfc)                                \
 456       SymI_HasProto(erff)                                \
 457       SymI_HasProto(erfcf)                               \
 458       SymI_HasProto(memcpy)                              \
 459       SymI_HasProto(rts_InstallConsoleEvent)             \
 460       SymI_HasProto(rts_ConsoleHandlerDone)              \
 461       SymI_NeedsProto(mktime)                            \
 462       SymI_NeedsProto(_imp___timezone)                   \
 463       SymI_NeedsProto(_imp___tzname)                     \
 464       SymI_NeedsProto(_imp__tzname)                      \
 465       SymI_NeedsProto(_imp___iob)                        \
 466       SymI_NeedsProto(_imp___osver)                      \
 467       SymI_NeedsProto(localtime)                         \
 468       SymI_NeedsProto(gmtime)                            \
 469       SymI_NeedsProto(opendir)                           \
 470       SymI_NeedsProto(readdir)                           \
 471       SymI_NeedsProto(rewinddir)                         \
 472       SymI_NeedsProto(_imp____mb_cur_max)                \
 473       SymI_NeedsProto(_imp___pctype)                     \
 474       SymI_NeedsProto(__chkstk)                          \
 475       RTS_MINGW_GETTIMEOFDAY_SYM                         \
 476       SymI_NeedsProto(closedir)
 477 #endif
 478
 479
 480 #if defined(darwin_HOST_OS) && HAVE_PRINTF_LDBLSTUB
 481 #define RTS_DARWIN_ONLY_SYMBOLS                             \
 482      SymI_NeedsProto(asprintf$LDBLStub)                     \
 483      SymI_NeedsProto(err$LDBLStub)                          \
 484      SymI_NeedsProto(errc$LDBLStub)                         \
 485      SymI_NeedsProto(errx$LDBLStub)                         \
 486      SymI_NeedsProto(fprintf$LDBLStub)                      \
 487      SymI_NeedsProto(fscanf$LDBLStub)                       \
 488      SymI_NeedsProto(fwprintf$LDBLStub)                     \
 489      SymI_NeedsProto(fwscanf$LDBLStub)                      \
 490      SymI_NeedsProto(printf$LDBLStub)                       \
 491      SymI_NeedsProto(scanf$LDBLStub)                        \
 492      SymI_NeedsProto(snprintf$LDBLStub)                     \
 493      SymI_NeedsProto(sprintf$LDBLStub)                      \
 494      SymI_NeedsProto(sscanf$LDBLStub)                       \
 495      SymI_NeedsProto(strtold$LDBLStub)                      \
 496      SymI_NeedsProto(swprintf$LDBLStub)                     \
 497      SymI_NeedsProto(swscanf$LDBLStub)                      \
 498      SymI_NeedsProto(syslog$LDBLStub)                       \
 499      SymI_NeedsProto(vasprintf$LDBLStub)                    \
 500      SymI_NeedsProto(verr$LDBLStub)                         \
 501      SymI_NeedsProto(verrc$LDBLStub)                        \
 502      SymI_NeedsProto(verrx$LDBLStub)                        \
 503      SymI_NeedsProto(vfprintf$LDBLStub)                     \
 504      SymI_NeedsProto(vfscanf$LDBLStub)                      \
 505      SymI_NeedsProto(vfwprintf$LDBLStub)                    \
 506      SymI_NeedsProto(vfwscanf$LDBLStub)                     \
 507      SymI_NeedsProto(vprintf$LDBLStub)                      \
 508      SymI_NeedsProto(vscanf$LDBLStub)                       \
 509      SymI_NeedsProto(vsnprintf$LDBLStub)                    \
 510      SymI_NeedsProto(vsprintf$LDBLStub)                     \
 511      SymI_NeedsProto(vsscanf$LDBLStub)                      \
 512      SymI_NeedsProto(vswprintf$LDBLStub)                    \
 513      SymI_NeedsProto(vswscanf$LDBLStub)                     \
 514      SymI_NeedsProto(vsyslog$LDBLStub)                      \
 515      SymI_NeedsProto(vwarn$LDBLStub)                        \
 516      SymI_NeedsProto(vwarnc$LDBLStub)                       \
 517      SymI_NeedsProto(vwarnx$LDBLStub)                       \
 518      SymI_NeedsProto(vwprintf$LDBLStub)                     \
 519      SymI_NeedsProto(vwscanf$LDBLStub)                      \
 520      SymI_NeedsProto(warn$LDBLStub)                         \
 521      SymI_NeedsProto(warnc$LDBLStub)                        \
 522      SymI_NeedsProto(warnx$LDBLStub)                        \
 523      SymI_NeedsProto(wcstold$LDBLStub)                      \
 524      SymI_NeedsProto(wprintf$LDBLStub)                      \
 525      SymI_NeedsProto(wscanf$LDBLStub)
 526 #else
 527 #define RTS_DARWIN_ONLY_SYMBOLS
 528 #endif
 529
 530 #ifndef SMP
 531 # define MAIN_CAP_SYM SymI_HasProto(MainCapability)
 532 #else
 533 # define MAIN_CAP_SYM
 534 #endif
 535
 536 #if !defined(mingw32_HOST_OS)
 537 #define RTS_USER_SIGNALS_SYMBOLS        \
 538    SymI_HasProto(setIOManagerControlFd) \
 539    SymI_HasProto(setIOManagerWakeupFd)  \
 540    SymI_HasProto(ioManagerWakeup)       \
 541    SymI_HasProto(blockUserSignals)      \
 542    SymI_HasProto(unblockUserSignals)
 543 #else
 544 #define RTS_USER_SIGNALS_SYMBOLS        \
 545    SymI_HasProto(ioManagerWakeup)       \
 546    SymI_HasProto(sendIOManagerEvent)    \
 547    SymI_HasProto(readIOManagerEvent)    \
 548    SymI_HasProto(getIOManagerEvent)     \
 549    SymI_HasProto(console_handler)
 550 #endif
 551
 552 #define RTS_LIBFFI_SYMBOLS                                  \
 553      SymE_NeedsProto(ffi_prep_cif)                          \
 554      SymE_NeedsProto(ffi_call)                              \
 555      SymE_NeedsProto(ffi_type_void)                         \
 556      SymE_NeedsProto(ffi_type_float)                        \
 557      SymE_NeedsProto(ffi_type_double)                       \
 558      SymE_NeedsProto(ffi_type_sint64)                       \
 559      SymE_NeedsProto(ffi_type_uint64)                       \
 560      SymE_NeedsProto(ffi_type_sint32)                       \
 561      SymE_NeedsProto(ffi_type_uint32)                       \
 562      SymE_NeedsProto(ffi_type_sint16)                       \
 563      SymE_NeedsProto(ffi_type_uint16)                       \
 564      SymE_NeedsProto(ffi_type_sint8)                        \
 565      SymE_NeedsProto(ffi_type_uint8)                        \
 566      SymE_NeedsProto(ffi_type_pointer)
 567
 568 #ifdef TABLES_NEXT_TO_CODE
 569 #define RTS_RET_SYMBOLS /* nothing */
 570 #else
 571 #define RTS_RET_SYMBOLS                                 \
 572       SymI_HasProto(stg_enter_ret)                      \
 573       SymI_HasProto(stg_gc_fun_ret)                     \
 574       SymI_HasProto(stg_ap_v_ret)                       \
 575       SymI_HasProto(stg_ap_f_ret)                       \
 576       SymI_HasProto(stg_ap_d_ret)                       \
 577       SymI_HasProto(stg_ap_l_ret)                       \
 578       SymI_HasProto(stg_ap_n_ret)                       \
 579       SymI_HasProto(stg_ap_p_ret)                       \
 580       SymI_HasProto(stg_ap_pv_ret)                      \
 581       SymI_HasProto(stg_ap_pp_ret)                      \
 582       SymI_HasProto(stg_ap_ppv_ret)                     \
 583       SymI_HasProto(stg_ap_ppp_ret)                     \
 584       SymI_HasProto(stg_ap_pppv_ret)                    \
 585       SymI_HasProto(stg_ap_pppp_ret)                    \
 586       SymI_HasProto(stg_ap_ppppp_ret)                   \
 587       SymI_HasProto(stg_ap_pppppp_ret)
 588 #endif
 589
 590 /* Modules compiled with -ticky may mention ticky counters */
 591 /* This list should marry up with the one in $(TOP)/includes/stg/Ticky.h */
 592 #define RTS_TICKY_SYMBOLS                               \
 593       SymI_NeedsProto(ticky_entry_ctrs)                 \
 594       SymI_NeedsProto(top_ct)                           \
 595                                                         \
 596       SymI_HasProto(ENT_VIA_NODE_ctr)                   \
 597       SymI_HasProto(ENT_STATIC_THK_ctr)                 \
 598       SymI_HasProto(ENT_DYN_THK_ctr)                    \
 599       SymI_HasProto(ENT_STATIC_FUN_DIRECT_ctr)          \
 600       SymI_HasProto(ENT_DYN_FUN_DIRECT_ctr)             \
 601       SymI_HasProto(ENT_STATIC_CON_ctr)                 \
 602       SymI_HasProto(ENT_DYN_CON_ctr)                    \
 603       SymI_HasProto(ENT_STATIC_IND_ctr)                 \
 604       SymI_HasProto(ENT_DYN_IND_ctr)                    \
 605       SymI_HasProto(ENT_PERM_IND_ctr)                   \
 606       SymI_HasProto(ENT_PAP_ctr)                        \
 607       SymI_HasProto(ENT_AP_ctr)                         \
 608       SymI_HasProto(ENT_AP_STACK_ctr)                   \
 609       SymI_HasProto(ENT_BH_ctr)                         \
 610       SymI_HasProto(UNKNOWN_CALL_ctr)                   \
 611       SymI_HasProto(SLOW_CALL_v_ctr)                    \
 612       SymI_HasProto(SLOW_CALL_f_ctr)                    \
 613       SymI_HasProto(SLOW_CALL_d_ctr)                    \
 614       SymI_HasProto(SLOW_CALL_l_ctr)                    \
 615       SymI_HasProto(SLOW_CALL_n_ctr)                    \
 616       SymI_HasProto(SLOW_CALL_p_ctr)                    \
 617       SymI_HasProto(SLOW_CALL_pv_ctr)                   \
 618       SymI_HasProto(SLOW_CALL_pp_ctr)                   \
 619       SymI_HasProto(SLOW_CALL_ppv_ctr)                  \
 620       SymI_HasProto(SLOW_CALL_ppp_ctr)                  \
 621       SymI_HasProto(SLOW_CALL_pppv_ctr)                 \
 622       SymI_HasProto(SLOW_CALL_pppp_ctr)                 \
 623       SymI_HasProto(SLOW_CALL_ppppp_ctr)                \
 624       SymI_HasProto(SLOW_CALL_pppppp_ctr)               \
 625       SymI_HasProto(SLOW_CALL_OTHER_ctr)                \
 626       SymI_HasProto(ticky_slow_call_unevald)            \
 627       SymI_HasProto(SLOW_CALL_ctr)                      \
 628       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_ctr)          \
 629       SymI_HasProto(MULTI_CHUNK_SLOW_CALL_CHUNKS_ctr)   \
 630       SymI_HasProto(KNOWN_CALL_ctr)                     \
 631       SymI_HasProto(KNOWN_CALL_TOO_FEW_ARGS_ctr)        \
 632       SymI_HasProto(KNOWN_CALL_EXTRA_ARGS_ctr)          \
 633       SymI_HasProto(SLOW_CALL_FUN_TOO_FEW_ctr)          \
 634       SymI_HasProto(SLOW_CALL_FUN_CORRECT_ctr)          \
 635       SymI_HasProto(SLOW_CALL_FUN_TOO_MANY_ctr)         \
 636       SymI_HasProto(SLOW_CALL_PAP_TOO_FEW_ctr)          \
 637       SymI_HasProto(SLOW_CALL_PAP_CORRECT_ctr)          \
 638       SymI_HasProto(SLOW_CALL_PAP_TOO_MANY_ctr)         \
 639       SymI_HasProto(SLOW_CALL_UNEVALD_ctr)              \
 640       SymI_HasProto(UPDF_OMITTED_ctr)                   \
 641       SymI_HasProto(UPDF_PUSHED_ctr)                    \
 642       SymI_HasProto(CATCHF_PUSHED_ctr)                  \
 643       SymI_HasProto(UPDF_RCC_PUSHED_ctr)                \
 644       SymI_HasProto(UPDF_RCC_OMITTED_ctr)               \
 645       SymI_HasProto(UPD_SQUEEZED_ctr)                   \
 646       SymI_HasProto(UPD_CON_IN_NEW_ctr)                 \
 647       SymI_HasProto(UPD_CON_IN_PLACE_ctr)               \
 648       SymI_HasProto(UPD_PAP_IN_NEW_ctr)                 \
 649       SymI_HasProto(UPD_PAP_IN_PLACE_ctr)               \
 650       SymI_HasProto(ALLOC_HEAP_ctr)                     \
 651       SymI_HasProto(ALLOC_HEAP_tot)                     \
 652       SymI_HasProto(ALLOC_FUN_ctr)                      \
 653       SymI_HasProto(ALLOC_FUN_adm)                      \
 654       SymI_HasProto(ALLOC_FUN_gds)                      \
 655       SymI_HasProto(ALLOC_FUN_slp)                      \
 656       SymI_HasProto(UPD_NEW_IND_ctr)                    \
 657       SymI_HasProto(UPD_NEW_PERM_IND_ctr)               \
 658       SymI_HasProto(UPD_OLD_IND_ctr)                    \
 659       SymI_HasProto(UPD_OLD_PERM_IND_ctr)               \
 660       SymI_HasProto(UPD_BH_UPDATABLE_ctr)               \
 661       SymI_HasProto(UPD_BH_SINGLE_ENTRY_ctr)            \
 662       SymI_HasProto(UPD_CAF_BH_UPDATABLE_ctr)           \
 663       SymI_HasProto(UPD_CAF_BH_SINGLE_ENTRY_ctr)        \
 664       SymI_HasProto(GC_SEL_ABANDONED_ctr)               \
 665       SymI_HasProto(GC_SEL_MINOR_ctr)                   \
 666       SymI_HasProto(GC_SEL_MAJOR_ctr)                   \
 667       SymI_HasProto(GC_FAILED_PROMOTION_ctr)            \
 668       SymI_HasProto(ALLOC_UP_THK_ctr)                   \
 669       SymI_HasProto(ALLOC_SE_THK_ctr)                   \
 670       SymI_HasProto(ALLOC_THK_adm)                      \
 671       SymI_HasProto(ALLOC_THK_gds)                      \
 672       SymI_HasProto(ALLOC_THK_slp)                      \
 673       SymI_HasProto(ALLOC_CON_ctr)                      \
 674       SymI_HasProto(ALLOC_CON_adm)                      \
 675       SymI_HasProto(ALLOC_CON_gds)                      \
 676       SymI_HasProto(ALLOC_CON_slp)                      \
 677       SymI_HasProto(ALLOC_TUP_ctr)                      \
 678       SymI_HasProto(ALLOC_TUP_adm)                      \
 679       SymI_HasProto(ALLOC_TUP_gds)                      \
 680       SymI_HasProto(ALLOC_TUP_slp)                      \
 681       SymI_HasProto(ALLOC_BH_ctr)                       \
 682       SymI_HasProto(ALLOC_BH_adm)                       \
 683       SymI_HasProto(ALLOC_BH_gds)                       \
 684       SymI_HasProto(ALLOC_BH_slp)                       \
 685       SymI_HasProto(ALLOC_PRIM_ctr)                     \
 686       SymI_HasProto(ALLOC_PRIM_adm)                     \
 687       SymI_HasProto(ALLOC_PRIM_gds)                     \
 688       SymI_HasProto(ALLOC_PRIM_slp)                     \
 689       SymI_HasProto(ALLOC_PAP_ctr)                      \
 690       SymI_HasProto(ALLOC_PAP_adm)                      \
 691       SymI_HasProto(ALLOC_PAP_gds)                      \
 692       SymI_HasProto(ALLOC_PAP_slp)                      \
 693       SymI_HasProto(ALLOC_TSO_ctr)                      \
 694       SymI_HasProto(ALLOC_TSO_adm)                      \
 695       SymI_HasProto(ALLOC_TSO_gds)                      \
 696       SymI_HasProto(ALLOC_TSO_slp)                      \
 697       SymI_HasProto(RET_NEW_ctr)                        \
 698       SymI_HasProto(RET_OLD_ctr)                        \
 699       SymI_HasProto(RET_UNBOXED_TUP_ctr)                \
 700       SymI_HasProto(RET_SEMI_loads_avoided)
 701
 702
 703 // On most platforms, the garbage collector rewrites references
 704 //      to small integer and char objects to a set of common, shared ones.
 705 //
 706 // We don't do this when compiling to Windows DLLs at the moment because
 707 //      it doesn't support cross package data references well.
 708 //
 709 #if defined(__PIC__) && defined(mingw32_HOST_OS)
 710 #define RTS_INTCHAR_SYMBOLS
 711 #else
 712 #define RTS_INTCHAR_SYMBOLS                             \
 713       SymI_HasProto(stg_CHARLIKE_closure)               \
 714       SymI_HasProto(stg_INTLIKE_closure)
 715 #endif
 716
 717
 718 #define RTS_SYMBOLS                                     \
 719       Maybe_Stable_Names                                \
 720       RTS_TICKY_SYMBOLS                                 \
 721       SymI_HasProto(StgReturn)                          \
 722       SymI_HasProto(stg_enter_info)                     \
 723       SymI_HasProto(stg_gc_void_info)                   \
 724       SymI_HasProto(__stg_gc_enter_1)                   \
 725       SymI_HasProto(stg_gc_noregs)                      \
 726       SymI_HasProto(stg_gc_unpt_r1_info)                \
 727       SymI_HasProto(stg_gc_unpt_r1)                     \
 728       SymI_HasProto(stg_gc_unbx_r1_info)                \
 729       SymI_HasProto(stg_gc_unbx_r1)                     \
 730       SymI_HasProto(stg_gc_f1_info)                     \
 731       SymI_HasProto(stg_gc_f1)                          \
 732       SymI_HasProto(stg_gc_d1_info)                     \
 733       SymI_HasProto(stg_gc_d1)                          \
 734       SymI_HasProto(stg_gc_l1_info)                     \
 735       SymI_HasProto(stg_gc_l1)                          \
 736       SymI_HasProto(__stg_gc_fun)                       \
 737       SymI_HasProto(stg_gc_fun_info)                    \
 738       SymI_HasProto(stg_gc_gen)                         \
 739       SymI_HasProto(stg_gc_gen_info)                    \
 740       SymI_HasProto(stg_gc_gen_hp)                      \
 741       SymI_HasProto(stg_gc_ut)                          \
 742       SymI_HasProto(stg_gen_yield)                      \
 743       SymI_HasProto(stg_yield_noregs)                   \
 744       SymI_HasProto(stg_yield_to_interpreter)           \
 745       SymI_HasProto(stg_gen_block)                      \
 746       SymI_HasProto(stg_block_noregs)                   \
 747       SymI_HasProto(stg_block_1)                        \
 748       SymI_HasProto(stg_block_takemvar)                 \
 749       SymI_HasProto(stg_block_putmvar)                  \
 750       MAIN_CAP_SYM                                      \
 751       SymI_HasProto(MallocFailHook)                     \
 752       SymI_HasProto(OnExitHook)                         \
 753       SymI_HasProto(OutOfHeapHook)                      \
 754       SymI_HasProto(StackOverflowHook)                  \
 755       SymI_HasProto(addDLL)                             \
 756       SymI_HasProto(__int_encodeDouble)                 \
 757       SymI_HasProto(__word_encodeDouble)                \
 758       SymI_HasProto(__2Int_encodeDouble)                \
 759       SymI_HasProto(__int_encodeFloat)                  \
 760       SymI_HasProto(__word_encodeFloat)                 \
 761       SymI_HasProto(stg_atomicallyzh)                   \
 762       SymI_HasProto(barf)                               \
 763       SymI_HasProto(debugBelch)                         \
 764       SymI_HasProto(errorBelch)                         \
 765       SymI_HasProto(sysErrorBelch)                      \
 766       SymI_HasProto(stg_getMaskingStatezh)              \
 767       SymI_HasProto(stg_maskAsyncExceptionszh)          \
 768       SymI_HasProto(stg_maskUninterruptiblezh)          \
 769       SymI_HasProto(stg_catchzh)                        \
 770       SymI_HasProto(stg_catchRetryzh)                   \
 771       SymI_HasProto(stg_catchSTMzh)                     \
 772       SymI_HasProto(stg_checkzh)                        \
 773       SymI_HasProto(closure_flags)                      \
 774       SymI_HasProto(cmp_thread)                         \
 775       SymI_HasProto(createAdjustor)                     \
 776       SymI_HasProto(stg_decodeDoublezu2Intzh)           \
 777       SymI_HasProto(stg_decodeFloatzuIntzh)             \
 778       SymI_HasProto(defaultsHook)                       \
 779       SymI_HasProto(stg_delayzh)                        \
 780       SymI_HasProto(stg_deRefWeakzh)                    \
 781       SymI_HasProto(stg_deRefStablePtrzh)               \
 782       SymI_HasProto(dirty_MUT_VAR)                      \
 783       SymI_HasProto(stg_forkzh)                         \
 784       SymI_HasProto(stg_forkOnzh)                       \
 785       SymI_HasProto(forkProcess)                        \
 786       SymI_HasProto(forkOS_createThread)                \
 787       SymI_HasProto(freeHaskellFunctionPtr)             \
 788       SymI_HasProto(getOrSetTypeableStore)              \
 789       SymI_HasProto(getOrSetGHCConcSignalSignalHandlerStore)            \
 790       SymI_HasProto(getOrSetGHCConcWindowsPendingDelaysStore)           \
 791       SymI_HasProto(getOrSetGHCConcWindowsIOManagerThreadStore)         \
 792       SymI_HasProto(getOrSetGHCConcWindowsProddingStore)                \
 793       SymI_HasProto(getOrSetSystemEventThreadEventManagerStore)         \
 794       SymI_HasProto(getOrSetSystemEventThreadIOManagerThreadStore)      \
 795       SymI_HasProto(genSymZh)                           \
 796       SymI_HasProto(genericRaise)                       \
 797       SymI_HasProto(getProgArgv)                        \
 798       SymI_HasProto(getFullProgArgv)                    \
 799       SymI_HasProto(getStablePtr)                       \
 800       SymI_HasProto(hs_init)                            \
 801       SymI_HasProto(hs_exit)                            \
 802       SymI_HasProto(hs_set_argv)                        \
 803       SymI_HasProto(hs_add_root)                        \
 804       SymI_HasProto(hs_perform_gc)                      \
 805       SymI_HasProto(hs_free_stable_ptr)                 \
 806       SymI_HasProto(hs_free_fun_ptr)                    \
 807       SymI_HasProto(hs_hpc_rootModule)                  \
 808       SymI_HasProto(hs_hpc_module)                      \
 809       SymI_HasProto(initLinker)                         \
 810       SymI_HasProto(stg_unpackClosurezh)                \
 811       SymI_HasProto(stg_getApStackValzh)                \
 812       SymI_HasProto(stg_getSparkzh)                     \
 813       SymI_HasProto(stg_numSparkszh)                    \
 814       SymI_HasProto(stg_isCurrentThreadBoundzh)         \
 815       SymI_HasProto(stg_isEmptyMVarzh)                  \
 816       SymI_HasProto(stg_killThreadzh)                   \
 817       SymI_HasProto(loadArchive)                        \
 818       SymI_HasProto(loadObj)                            \
 819       SymI_HasProto(insertStableSymbol)                 \
 820       SymI_HasProto(insertSymbol)                       \
 821       SymI_HasProto(lookupSymbol)                       \
 822       SymI_HasProto(stg_makeStablePtrzh)                \
 823       SymI_HasProto(stg_mkApUpd0zh)                     \
 824       SymI_HasProto(stg_myThreadIdzh)                   \
 825       SymI_HasProto(stg_labelThreadzh)                  \
 826       SymI_HasProto(stg_newArrayzh)                     \
 827       SymI_HasProto(stg_newBCOzh)                       \
 828       SymI_HasProto(stg_newByteArrayzh)                 \
 829       SymI_HasProto_redirect(newCAF, newDynCAF)         \
 830       SymI_HasProto(stg_newMVarzh)                      \
 831       SymI_HasProto(stg_newMutVarzh)                    \
 832       SymI_HasProto(stg_newTVarzh)                      \
 833       SymI_HasProto(stg_noDuplicatezh)                  \
 834       SymI_HasProto(stg_atomicModifyMutVarzh)           \
 835       SymI_HasProto(stg_casMutVarzh)                    \
 836       SymI_HasProto(stg_newPinnedByteArrayzh)           \
 837       SymI_HasProto(stg_newAlignedPinnedByteArrayzh)    \
 838       SymI_HasProto(newSpark)                           \
 839       SymI_HasProto(performGC)                          \
 840       SymI_HasProto(performMajorGC)                     \
 841       SymI_HasProto(prog_argc)                          \
 842       SymI_HasProto(prog_argv)                          \
 843       SymI_HasProto(stg_putMVarzh)                      \
 844       SymI_HasProto(stg_raisezh)                        \
 845       SymI_HasProto(stg_raiseIOzh)                      \
 846       SymI_HasProto(stg_readTVarzh)                     \
 847       SymI_HasProto(stg_readTVarIOzh)                   \
 848       SymI_HasProto(resumeThread)                       \
 849       SymI_HasProto(resolveObjs)                        \
 850       SymI_HasProto(stg_retryzh)                        \
 851       SymI_HasProto(rts_apply)                          \
 852       SymI_HasProto(rts_checkSchedStatus)               \
 853       SymI_HasProto(rts_eval)                           \
 854       SymI_HasProto(rts_evalIO)                         \
 855       SymI_HasProto(rts_evalLazyIO)                     \
 856       SymI_HasProto(rts_evalStableIO)                   \
 857       SymI_HasProto(rts_eval_)                          \
 858       SymI_HasProto(rts_getBool)                        \
 859       SymI_HasProto(rts_getChar)                        \
 860       SymI_HasProto(rts_getDouble)                      \
 861       SymI_HasProto(rts_getFloat)                       \
 862       SymI_HasProto(rts_getInt)                         \
 863       SymI_HasProto(rts_getInt8)                        \
 864       SymI_HasProto(rts_getInt16)                       \
 865       SymI_HasProto(rts_getInt32)                       \
 866       SymI_HasProto(rts_getInt64)                       \
 867       SymI_HasProto(rts_getPtr)                         \
 868       SymI_HasProto(rts_getFunPtr)                      \
 869       SymI_HasProto(rts_getStablePtr)                   \
 870       SymI_HasProto(rts_getThreadId)                    \
 871       SymI_HasProto(rts_getWord)                        \
 872       SymI_HasProto(rts_getWord8)                       \
 873       SymI_HasProto(rts_getWord16)                      \
 874       SymI_HasProto(rts_getWord32)                      \
 875       SymI_HasProto(rts_getWord64)                      \
 876       SymI_HasProto(rts_lock)                           \
 877       SymI_HasProto(rts_mkBool)                         \
 878       SymI_HasProto(rts_mkChar)                         \
 879       SymI_HasProto(rts_mkDouble)                       \
 880       SymI_HasProto(rts_mkFloat)                        \
 881       SymI_HasProto(rts_mkInt)                          \
 882       SymI_HasProto(rts_mkInt8)                         \
 883       SymI_HasProto(rts_mkInt16)                        \
 884       SymI_HasProto(rts_mkInt32)                        \
 885       SymI_HasProto(rts_mkInt64)                        \
 886       SymI_HasProto(rts_mkPtr)                          \
 887       SymI_HasProto(rts_mkFunPtr)                       \
 888       SymI_HasProto(rts_mkStablePtr)                    \
 889       SymI_HasProto(rts_mkString)                       \
 890       SymI_HasProto(rts_mkWord)                         \
 891       SymI_HasProto(rts_mkWord8)                        \
 892       SymI_HasProto(rts_mkWord16)                       \
 893       SymI_HasProto(rts_mkWord32)                       \
 894       SymI_HasProto(rts_mkWord64)                       \
 895       SymI_HasProto(rts_unlock)                         \
 896       SymI_HasProto(rts_unsafeGetMyCapability)          \
 897       SymI_HasProto(rtsSupportsBoundThreads)            \
 898       SymI_HasProto(rts_isProfiled)                     \
 899       SymI_HasProto(setProgArgv)                        \
 900       SymI_HasProto(startupHaskell)                     \
 901       SymI_HasProto(shutdownHaskell)                    \
 902       SymI_HasProto(shutdownHaskellAndExit)             \
 903       SymI_HasProto(stable_ptr_table)                   \
 904       SymI_HasProto(stackOverflow)                      \
 905       SymI_HasProto(stg_CAF_BLACKHOLE_info)             \
 906       SymI_HasProto(stg_BLACKHOLE_info)                 \
 907       SymI_HasProto(__stg_EAGER_BLACKHOLE_info)         \
 908       SymI_HasProto(stg_BLOCKING_QUEUE_CLEAN_info)      \
 909       SymI_HasProto(stg_BLOCKING_QUEUE_DIRTY_info)      \
 910       SymI_HasProto(startTimer)                         \
 911       SymI_HasProto(stg_MVAR_CLEAN_info)                \
 912       SymI_HasProto(stg_MVAR_DIRTY_info)                \
 913       SymI_HasProto(stg_IND_STATIC_info)                \
 914       SymI_HasProto(stg_ARR_WORDS_info)                 \
 915       SymI_HasProto(stg_MUT_ARR_PTRS_DIRTY_info)        \
 916       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN_info)       \
 917       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN0_info)      \
 918       SymI_HasProto(stg_WEAK_info)                      \
 919       SymI_HasProto(stg_ap_v_info)                      \
 920       SymI_HasProto(stg_ap_f_info)                      \
 921       SymI_HasProto(stg_ap_d_info)                      \
 922       SymI_HasProto(stg_ap_l_info)                      \
 923       SymI_HasProto(stg_ap_n_info)                      \
 924       SymI_HasProto(stg_ap_p_info)                      \
 925       SymI_HasProto(stg_ap_pv_info)                     \
 926       SymI_HasProto(stg_ap_pp_info)                     \
 927       SymI_HasProto(stg_ap_ppv_info)                    \
 928       SymI_HasProto(stg_ap_ppp_info)                    \
 929       SymI_HasProto(stg_ap_pppv_info)                   \
 930       SymI_HasProto(stg_ap_pppp_info)                   \
 931       SymI_HasProto(stg_ap_ppppp_info)                  \
 932       SymI_HasProto(stg_ap_pppppp_info)                 \
 933       SymI_HasProto(stg_ap_0_fast)                      \
 934       SymI_HasProto(stg_ap_v_fast)                      \
 935       SymI_HasProto(stg_ap_f_fast)                      \
 936       SymI_HasProto(stg_ap_d_fast)                      \
 937       SymI_HasProto(stg_ap_l_fast)                      \
 938       SymI_HasProto(stg_ap_n_fast)                      \
 939       SymI_HasProto(stg_ap_p_fast)                      \
 940       SymI_HasProto(stg_ap_pv_fast)                     \
 941       SymI_HasProto(stg_ap_pp_fast)                     \
 942       SymI_HasProto(stg_ap_ppv_fast)                    \
 943       SymI_HasProto(stg_ap_ppp_fast)                    \
 944       SymI_HasProto(stg_ap_pppv_fast)                   \
 945       SymI_HasProto(stg_ap_pppp_fast)                   \
 946       SymI_HasProto(stg_ap_ppppp_fast)                  \
 947       SymI_HasProto(stg_ap_pppppp_fast)                 \
 948       SymI_HasProto(stg_ap_1_upd_info)                  \
 949       SymI_HasProto(stg_ap_2_upd_info)                  \
 950       SymI_HasProto(stg_ap_3_upd_info)                  \
 951       SymI_HasProto(stg_ap_4_upd_info)                  \
 952       SymI_HasProto(stg_ap_5_upd_info)                  \
 953       SymI_HasProto(stg_ap_6_upd_info)                  \
 954       SymI_HasProto(stg_ap_7_upd_info)                  \
 955       SymI_HasProto(stg_exit)                           \
 956       SymI_HasProto(stg_sel_0_upd_info)                 \
 957       SymI_HasProto(stg_sel_10_upd_info)                \
 958       SymI_HasProto(stg_sel_11_upd_info)                \
 959       SymI_HasProto(stg_sel_12_upd_info)                \
 960       SymI_HasProto(stg_sel_13_upd_info)                \
 961       SymI_HasProto(stg_sel_14_upd_info)                \
 962       SymI_HasProto(stg_sel_15_upd_info)                \
 963       SymI_HasProto(stg_sel_1_upd_info)                 \
 964       SymI_HasProto(stg_sel_2_upd_info)                 \
 965       SymI_HasProto(stg_sel_3_upd_info)                 \
 966       SymI_HasProto(stg_sel_4_upd_info)                 \
 967       SymI_HasProto(stg_sel_5_upd_info)                 \
 968       SymI_HasProto(stg_sel_6_upd_info)                 \
 969       SymI_HasProto(stg_sel_7_upd_info)                 \
 970       SymI_HasProto(stg_sel_8_upd_info)                 \
 971       SymI_HasProto(stg_sel_9_upd_info)                 \
 972       SymI_HasProto(stg_upd_frame_info)                 \
 973       SymI_HasProto(stg_bh_upd_frame_info)              \
 974       SymI_HasProto(suspendThread)                      \
 975       SymI_HasProto(stg_takeMVarzh)                     \
 976       SymI_HasProto(stg_threadStatuszh)                 \
 977       SymI_HasProto(stg_tryPutMVarzh)                   \
 978       SymI_HasProto(stg_tryTakeMVarzh)                  \
 979       SymI_HasProto(stg_unmaskAsyncExceptionszh)        \
 980       SymI_HasProto(unloadObj)                          \
 981       SymI_HasProto(stg_unsafeThawArrayzh)              \
 982       SymI_HasProto(stg_waitReadzh)                     \
 983       SymI_HasProto(stg_waitWritezh)                    \
 984       SymI_HasProto(stg_writeTVarzh)                    \
 985       SymI_HasProto(stg_yieldzh)                        \
 986       SymI_NeedsProto(stg_interp_constr_entry)          \
 987       SymI_HasProto(stg_arg_bitmaps)                    \
 988       SymI_HasProto(large_alloc_lim)                    \
 989       SymI_HasProto(g0)                                 \
 990       SymI_HasProto(allocate)                           \
 991       SymI_HasProto(allocateExec)                       \
 992       SymI_HasProto(freeExec)                           \
 993       SymI_HasProto(getAllocations)                     \
 994       SymI_HasProto(revertCAFs)                         \
 995       SymI_HasProto(RtsFlags)                           \
 996       SymI_NeedsProto(rts_breakpoint_io_action)         \
 997       SymI_NeedsProto(rts_stop_next_breakpoint)         \
 998       SymI_NeedsProto(rts_stop_on_exception)            \
 999       SymI_HasProto(stopTimer)                          \
1000       SymI_HasProto(n_capabilities)                     \
1001       SymI_HasProto(stg_traceCcszh)                     \
1002       SymI_HasProto(stg_traceEventzh)                   \
1003       RTS_USER_SIGNALS_SYMBOLS                          \
1004       RTS_INTCHAR_SYMBOLS
1005
1006
1007 // 64-bit support functions in libgcc.a
1008 #if defined(__GNUC__) && SIZEOF_VOID_P <= 4
1009 #define RTS_LIBGCC_SYMBOLS                             \
1010       SymI_NeedsProto(__divdi3)                        \
1011       SymI_NeedsProto(__udivdi3)                       \
1012       SymI_NeedsProto(__moddi3)                        \
1013       SymI_NeedsProto(__umoddi3)                       \
1014       SymI_NeedsProto(__muldi3)                        \
1015       SymI_NeedsProto(__ashldi3)                       \
1016       SymI_NeedsProto(__ashrdi3)                       \
1017       SymI_NeedsProto(__lshrdi3)
1018 #else
1019 #define RTS_LIBGCC_SYMBOLS
1020 #endif
1021
1022 #if defined(darwin_HOST_OS) && defined(powerpc_HOST_ARCH)
1023       // Symbols that don't have a leading underscore
1024       // on Mac OS X. They have to receive special treatment,
1025       // see machoInitSymbolsWithoutUnderscore()
1026 #define RTS_MACHO_NOUNDERLINE_SYMBOLS                   \
1027       SymI_NeedsProto(saveFP)                           \
1028       SymI_NeedsProto(restFP)
1029 #endif
1030
1031 /* entirely bogus claims about types of these symbols */
1032 #define SymI_NeedsProto(vvv)  extern void vvv(void);
1033 #if defined(__PIC__) && defined(mingw32_HOST_OS)
1034 #define SymE_HasProto(vvv)    SymE_HasProto(vvv);
1035 #define SymE_NeedsProto(vvv)    extern void _imp__ ## vvv (void);
1036 #else
1037 #define SymE_NeedsProto(vvv)  SymI_NeedsProto(vvv);
1038 #define SymE_HasProto(vvv)    SymI_HasProto(vvv)
1039 #endif
1040 #define SymI_HasProto(vvv) /**/
1041 #define SymI_HasProto_redirect(vvv,xxx) /**/
1042 RTS_SYMBOLS
1043 RTS_RET_SYMBOLS
1044 RTS_POSIX_ONLY_SYMBOLS
1045 RTS_MINGW_ONLY_SYMBOLS
1046 RTS_CYGWIN_ONLY_SYMBOLS
1047 RTS_DARWIN_ONLY_SYMBOLS
1048 RTS_LIBGCC_SYMBOLS
1049 RTS_LIBFFI_SYMBOLS
1050 #undef SymI_NeedsProto
1051 #undef SymI_HasProto
1052 #undef SymI_HasProto_redirect
1053 #undef SymE_HasProto
1054 #undef SymE_NeedsProto
1055
1056 #ifdef LEADING_UNDERSCORE
1057 #define MAYBE_LEADING_UNDERSCORE_STR(s) ("_" s)
1058 #else
1059 #define MAYBE_LEADING_UNDERSCORE_STR(s) (s)
1060 #endif
1061
1062 #define SymI_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1063                     (void*)(&(vvv)) },
1064 #define SymE_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1065             (void*)DLL_IMPORT_DATA_REF(vvv) },
1066
1067 #define SymI_NeedsProto(vvv) SymI_HasProto(vvv)
1068 #define SymE_NeedsProto(vvv) SymE_HasProto(vvv)
1069
1070 // SymI_HasProto_redirect allows us to redirect references to one symbol to
1071 // another symbol.  See newCAF/newDynCAF for an example.
1072 #define SymI_HasProto_redirect(vvv,xxx)   \
1073     { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
1074       (void*)(&(xxx)) },
1075
1076 static RtsSymbolVal rtsSyms[] = {
1077       RTS_SYMBOLS
1078       RTS_RET_SYMBOLS
1079       RTS_POSIX_ONLY_SYMBOLS
1080       RTS_MINGW_ONLY_SYMBOLS
1081       RTS_CYGWIN_ONLY_SYMBOLS
1082       RTS_DARWIN_ONLY_SYMBOLS
1083       RTS_LIBGCC_SYMBOLS
1084       RTS_LIBFFI_SYMBOLS
1085 #if defined(darwin_HOST_OS) && defined(i386_HOST_ARCH)
1086       // dyld stub code contains references to this,
1087       // but it should never be called because we treat
1088       // lazy pointers as nonlazy.
1089       { "dyld_stub_binding_helper", (void*)0xDEADBEEF },
1090 #endif
1091       { 0, 0 } /* sentinel */
1092 };
1093
1094
1095
1096 /* -----------------------------------------------------------------------------
1097  * Insert symbols into hash tables, checking for duplicates.
1098  */
1099
1100 static void ghciInsertStrHashTable ( char* obj_name,
1101                                      HashTable *table,
1102                                      char* key,
1103                                      void *data
1104                                    )
1105 {
1106    if (lookupHashTable(table, (StgWord)key) == NULL)
1107    {
1108       insertStrHashTable(table, (StgWord)key, data);
1109       return;
1110    }
1111    debugBelch(
1112       "\n\n"
1113       "GHCi runtime linker: fatal error: I found a duplicate definition for symbol\n"
1114       "   %s\n"
1115       "whilst processing object file\n"
1116       "   %s\n"
1117       "This could be caused by:\n"
1118       "   * Loading two different object files which export the same symbol\n"
1119       "   * Specifying the same object file twice on the GHCi command line\n"
1120       "   * An incorrect `package.conf' entry, causing some object to be\n"
1121       "     loaded twice.\n"
1122       "GHCi cannot safely continue in this situation.  Exiting now.  Sorry.\n"
1123       "\n",
1124       (char*)key,
1125       obj_name
1126    );
1127    stg_exit(1);
1128 }
1129 /* -----------------------------------------------------------------------------
1130  * initialize the object linker
1131  */
1132
1133
1134 static int linker_init_done = 0 ;
1135
1136 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1137 static void *dl_prog_handle;
1138 static regex_t re_invalid;
1139 static regex_t re_realso;
1140 #ifdef THREADED_RTS
1141 static Mutex dl_mutex; // mutex to protect dlopen/dlerror critical section
1142 #endif
1143 #endif
1144
1145 void
1146 initLinker( void )
1147 {
1148     RtsSymbolVal *sym;
1149 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1150     int compileResult;
1151 #endif
1152
1153     IF_DEBUG(linker, debugBelch("initLinker: start\n"));
1154
1155     /* Make initLinker idempotent, so we can call it
1156        before evey relevant operation; that means we
1157        don't need to initialise the linker separately */
1158     if (linker_init_done == 1) {
1159         IF_DEBUG(linker, debugBelch("initLinker: idempotent return\n"));
1160         return;
1161     } else {
1162         linker_init_done = 1;
1163     }
1164
1165 #if defined(THREADED_RTS) && (defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO))
1166     initMutex(&dl_mutex);
1167 #endif
1168     stablehash = allocStrHashTable();
1169     symhash = allocStrHashTable();
1170
1171     /* populate the symbol table with stuff from the RTS */
1172     for (sym = rtsSyms; sym->lbl != NULL; sym++) {
1173         ghciInsertStrHashTable("(GHCi built-in symbols)",
1174                                symhash, sym->lbl, sym->addr);
1175         IF_DEBUG(linker, debugBelch("initLinker: inserting rts symbol %s, %p\n", sym->lbl, sym->addr));
1176     }
1177 #   if defined(OBJFORMAT_MACHO) && defined(powerpc_HOST_ARCH)
1178     machoInitSymbolsWithoutUnderscore();
1179 #   endif
1180
1181 #   if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1182 #   if defined(RTLD_DEFAULT)
1183     dl_prog_handle = RTLD_DEFAULT;
1184 #   else
1185     dl_prog_handle = dlopen(NULL, RTLD_LAZY);
1186 #   endif /* RTLD_DEFAULT */
1187
1188     compileResult = regcomp(&re_invalid,
1189            "(([^ \t()])+\\.so([^ \t:()])*):([ \t])*invalid ELF header",
1190            REG_EXTENDED);
1191     ASSERT( compileResult == 0 );
1192     compileResult = regcomp(&re_realso,
1193            "GROUP *\\( *(([^ )])+)",
1194            REG_EXTENDED);
1195     ASSERT( compileResult == 0 );
1196 #   endif
1197
1198 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1199     if (RtsFlags.MiscFlags.linkerMemBase != 0) {
1200         // User-override for mmap_32bit_base
1201         mmap_32bit_base = (void*)RtsFlags.MiscFlags.linkerMemBase;
1202     }
1203 #endif
1204
1205 #if defined(mingw32_HOST_OS)
1206     /*
1207      * These two libraries cause problems when added to the static link,
1208      * but are necessary for resolving symbols in GHCi, hence we load
1209      * them manually here.
1210      */
1211     addDLL("msvcrt");
1212     addDLL("kernel32");
1213 #endif
1214
1215     IF_DEBUG(linker, debugBelch("initLinker: done\n"));
1216     return;
1217 }
1218
1219 void
1220 exitLinker( void ) {
1221 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1222    if (linker_init_done == 1) {
1223       regfree(&re_invalid);
1224       regfree(&re_realso);
1225 #ifdef THREADED_RTS
1226       closeMutex(&dl_mutex);
1227 #endif
1228    }
1229 #endif
1230 }
1231
1232 /* -----------------------------------------------------------------------------
1233  *                  Loading DLL or .so dynamic libraries
1234  * -----------------------------------------------------------------------------
1235  *
1236  * Add a DLL from which symbols may be found.  In the ELF case, just
1237  * do RTLD_GLOBAL-style add, so no further messing around needs to
1238  * happen in order that symbols in the loaded .so are findable --
1239  * lookupSymbol() will subsequently see them by dlsym on the program's
1240  * dl-handle.  Returns NULL if success, otherwise ptr to an err msg.
1241  *
1242  * In the PEi386 case, open the DLLs and put handles to them in a
1243  * linked list.  When looking for a symbol, try all handles in the
1244  * list.  This means that we need to load even DLLs that are guaranteed
1245  * to be in the ghc.exe image already, just so we can get a handle
1246  * to give to loadSymbol, so that we can find the symbols.  For such
1247  * libraries, the LoadLibrary call should be a no-op except for returning
1248  * the handle.
1249  *
1250  */
1251
1252 #if defined(OBJFORMAT_PEi386)
1253 /* A record for storing handles into DLLs. */
1254
1255 typedef
1256    struct _OpenedDLL {
1257       char*              name;
1258       struct _OpenedDLL* next;
1259       HINSTANCE instance;
1260    }
1261    OpenedDLL;
1262
1263 /* A list thereof. */
1264 static OpenedDLL* opened_dlls = NULL;
1265 #endif
1266
1267 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1268
1269 static const char *
1270 internal_dlopen(const char *dll_name)
1271 {
1272    void *hdl;
1273    const char *errmsg;
1274    char *errmsg_copy;
1275
1276    // omitted: RTLD_NOW
1277    // see http://www.haskell.org/pipermail/cvs-ghc/2007-September/038570.html
1278    IF_DEBUG(linker,
1279       debugBelch("internal_dlopen: dll_name = '%s'\n", dll_name));
1280
1281    //-------------- Begin critical section ------------------
1282    // This critical section is necessary because dlerror() is not
1283    // required to be reentrant (see POSIX -- IEEE Std 1003.1-2008)
1284    // Also, the error message returned must be copied to preserve it
1285    // (see POSIX also)
1286
1287    ACQUIRE_LOCK(&dl_mutex);
1288    hdl = dlopen(dll_name, RTLD_LAZY | RTLD_GLOBAL);
1289
1290    errmsg = NULL;
1291    if (hdl == NULL) {
1292       /* dlopen failed; return a ptr to the error msg. */
1293       errmsg = dlerror();
1294       if (errmsg == NULL) errmsg = "addDLL: unknown error";
1295       errmsg_copy = stgMallocBytes(strlen(errmsg)+1, "addDLL");
1296       strcpy(errmsg_copy, errmsg);
1297       errmsg = errmsg_copy;
1298    }
1299    RELEASE_LOCK(&dl_mutex);
1300    //--------------- End critical section -------------------
1301
1302    return errmsg;
1303 }
1304 #  endif
1305
1306 const char *
1307 addDLL( char *dll_name )
1308 {
1309 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1310    /* ------------------- ELF DLL loader ------------------- */
1311
1312 #define NMATCH 5
1313    regmatch_t match[NMATCH];
1314    const char *errmsg;
1315    FILE* fp;
1316    size_t match_length;
1317 #define MAXLINE 1000
1318    char line[MAXLINE];
1319    int result;
1320
1321    initLinker();
1322
1323    IF_DEBUG(linker, debugBelch("addDLL: dll_name = '%s'\n", dll_name));
1324    errmsg = internal_dlopen(dll_name);
1325
1326    if (errmsg == NULL) {
1327       return NULL;
1328    }
1329
1330    // GHC Trac ticket #2615
1331    // On some systems (e.g., Gentoo Linux) dynamic files (e.g. libc.so)
1332    // contain linker scripts rather than ELF-format object code. This
1333    // code handles the situation by recognizing the real object code
1334    // file name given in the linker script.
1335    //
1336    // If an "invalid ELF header" error occurs, it is assumed that the
1337    // .so file contains a linker script instead of ELF object code.
1338    // In this case, the code looks for the GROUP ( ... ) linker
1339    // directive. If one is found, the first file name inside the
1340    // parentheses is treated as the name of a dynamic library and the
1341    // code attempts to dlopen that file. If this is also unsuccessful,
1342    // an error message is returned.
1343
1344    // see if the error message is due to an invalid ELF header
1345    IF_DEBUG(linker, debugBelch("errmsg = '%s'\n", errmsg));
1346    result = regexec(&re_invalid, errmsg, (size_t) NMATCH, match, 0);
1347    IF_DEBUG(linker, debugBelch("result = %i\n", result));
1348    if (result == 0) {
1349       // success -- try to read the named file as a linker script
1350       match_length = (size_t) stg_min((match[1].rm_eo - match[1].rm_so),
1351                                  MAXLINE-1);
1352       strncpy(line, (errmsg+(match[1].rm_so)),match_length);
1353       line[match_length] = '\0'; // make sure string is null-terminated
1354       IF_DEBUG(linker, debugBelch ("file name = '%s'\n", line));
1355       if ((fp = fopen(line, "r")) == NULL) {
1356          return errmsg; // return original error if open fails
1357       }
1358       // try to find a GROUP ( ... ) command
1359       while (fgets(line, MAXLINE, fp) != NULL) {
1360          IF_DEBUG(linker, debugBelch("input line = %s", line));
1361          if (regexec(&re_realso, line, (size_t) NMATCH, match, 0) == 0) {
1362             // success -- try to dlopen the first named file
1363             IF_DEBUG(linker, debugBelch("match%s\n",""));
1364             line[match[1].rm_eo] = '\0';
1365             errmsg = internal_dlopen(line+match[1].rm_so);
1366             break;
1367          }
1368          // if control reaches here, no GROUP ( ... ) directive was found
1369          // and the original error message is returned to the caller
1370       }
1371       fclose(fp);
1372    }
1373    return errmsg;
1374
1375 #  elif defined(OBJFORMAT_PEi386)
1376    /* ------------------- Win32 DLL loader ------------------- */
1377
1378    char*      buf;
1379    OpenedDLL* o_dll;
1380    HINSTANCE  instance;
1381
1382    initLinker();
1383
1384    /* debugBelch("\naddDLL; dll_name = `%s'\n", dll_name); */
1385
1386    /* See if we've already got it, and ignore if so. */
1387    for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
1388       if (0 == strcmp(o_dll->name, dll_name))
1389          return NULL;
1390    }
1391
1392    /* The file name has no suffix (yet) so that we can try
1393       both foo.dll and foo.drv
1394
1395       The documentation for LoadLibrary says:
1396         If no file name extension is specified in the lpFileName
1397         parameter, the default library extension .dll is
1398         appended. However, the file name string can include a trailing
1399         point character (.) to indicate that the module name has no
1400         extension. */
1401
1402    buf = stgMallocBytes(strlen(dll_name) + 10, "addDLL");
1403    sprintf(buf, "%s.DLL", dll_name);
1404    instance = LoadLibrary(buf);
1405    if (instance == NULL) {
1406        if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1407        // KAA: allow loading of drivers (like winspool.drv)
1408        sprintf(buf, "%s.DRV", dll_name);
1409        instance = LoadLibrary(buf);
1410        if (instance == NULL) {
1411            if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1412            // #1883: allow loading of unix-style libfoo.dll DLLs
1413            sprintf(buf, "lib%s.DLL", dll_name);
1414            instance = LoadLibrary(buf);
1415            if (instance == NULL) {
1416                goto error;
1417            }
1418        }
1419    }
1420    stgFree(buf);
1421
1422    /* Add this DLL to the list of DLLs in which to search for symbols. */
1423    o_dll = stgMallocBytes( sizeof(OpenedDLL), "addDLL" );
1424    o_dll->name     = stgMallocBytes(1+strlen(dll_name), "addDLL");
1425    strcpy(o_dll->name, dll_name);
1426    o_dll->instance = instance;
1427    o_dll->next     = opened_dlls;
1428    opened_dlls     = o_dll;
1429
1430    return NULL;
1431
1432 error:
1433    stgFree(buf);
1434    sysErrorBelch(dll_name);
1435
1436    /* LoadLibrary failed; return a ptr to the error msg. */
1437    return "addDLL: could not load DLL";
1438
1439 #  else
1440    barf("addDLL: not implemented on this platform");
1441 #  endif
1442 }
1443
1444 /* -----------------------------------------------------------------------------
1445  * insert a stable symbol in the hash table
1446  */
1447
1448 void
1449 insertStableSymbol(char* obj_name, char* key, StgPtr p)
1450 {
1451   ghciInsertStrHashTable(obj_name, stablehash, key, getStablePtr(p));
1452 }
1453
1454
1455 /* -----------------------------------------------------------------------------
1456  * insert a symbol in the hash table
1457  */
1458 void
1459 insertSymbol(char* obj_name, char* key, void* data)
1460 {
1461   ghciInsertStrHashTable(obj_name, symhash, key, data);
1462 }
1463
1464 /* -----------------------------------------------------------------------------
1465  * lookup a symbol in the hash table
1466  */
1467 void *
1468 lookupSymbol( char *lbl )
1469 {
1470     void *val;
1471     IF_DEBUG(linker, debugBelch("lookupSymbol: looking up %s\n", lbl));
1472     initLinker() ;
1473     ASSERT(symhash != NULL);
1474     val = lookupStrHashTable(symhash, lbl);
1475
1476     if (val == NULL) {
1477         IF_DEBUG(linker, debugBelch("lookupSymbol: symbol not found\n"));
1478 #       if defined(OBJFORMAT_ELF)
1479         return dlsym(dl_prog_handle, lbl);
1480 #       elif defined(OBJFORMAT_MACHO)
1481 #       if HAVE_DLFCN_H
1482         /* On OS X 10.3 and later, we use dlsym instead of the old legacy
1483            interface.
1484
1485            HACK: On OS X, global symbols are prefixed with an underscore.
1486                  However, dlsym wants us to omit the leading underscore from the
1487                  symbol name. For now, we simply strip it off here (and ONLY
1488                  here).
1489         */
1490         IF_DEBUG(linker, debugBelch("lookupSymbol: looking up %s with dlsym\n", lbl));
1491         ASSERT(lbl[0] == '_');
1492         return dlsym(dl_prog_handle, lbl+1);
1493 #       else
1494         if(NSIsSymbolNameDefined(lbl)) {
1495             NSSymbol symbol = NSLookupAndBindSymbol(lbl);
1496             return NSAddressOfSymbol(symbol);
1497         } else {
1498             return NULL;
1499         }
1500 #       endif /* HAVE_DLFCN_H */
1501 #       elif defined(OBJFORMAT_PEi386)
1502         void* sym;
1503
1504         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1505         if (sym != NULL) { return sym; };
1506
1507         // Also try looking up the symbol without the @N suffix.  Some
1508         // DLLs have the suffixes on their symbols, some don't.
1509         zapTrailingAtSign ( (unsigned char*)lbl );
1510         sym = lookupSymbolInDLLs((unsigned char*)lbl);
1511         if (sym != NULL) { return sym; };
1512         return NULL;
1513
1514 #       else
1515         ASSERT(2+2 == 5);
1516         return NULL;
1517 #       endif
1518     } else {
1519         IF_DEBUG(linker, debugBelch("lookupSymbol: value of %s is %p\n", lbl, val));
1520         return val;
1521     }
1522 }
1523
1524 /* -----------------------------------------------------------------------------
1525  * Debugging aid: look in GHCi's object symbol tables for symbols
1526  * within DELTA bytes of the specified address, and show their names.
1527  */
1528 #ifdef DEBUG
1529 void ghci_enquire ( char* addr );
1530
1531 void ghci_enquire ( char* addr )
1532 {
1533    int   i;
1534    char* sym;
1535    char* a;
1536    const int DELTA = 64;
1537    ObjectCode* oc;
1538
1539    initLinker();
1540
1541    for (oc = objects; oc; oc = oc->next) {
1542       for (i = 0; i < oc->n_symbols; i++) {
1543          sym = oc->symbols[i];
1544          if (sym == NULL) continue;
1545          a = NULL;
1546          if (a == NULL) {
1547             a = lookupStrHashTable(symhash, sym);
1548          }
1549          if (a == NULL) {
1550              // debugBelch("ghci_enquire: can't find %s\n", sym);
1551          }
1552          else if (addr-DELTA <= a && a <= addr+DELTA) {
1553             debugBelch("%p + %3d  ==  `%s'\n", addr, (int)(a - addr), sym);
1554          }
1555       }
1556    }
1557 }
1558 #endif
1559
1560 #ifdef USE_MMAP
1561 #define ROUND_UP(x,size) ((x + size - 1) & ~(size - 1))
1562
1563 static void *
1564 mmapForLinker (size_t bytes, nat flags, int fd)
1565 {
1566    void *map_addr = NULL;
1567    void *result;
1568    int pagesize, size;
1569    static nat fixed = 0;
1570
1571    IF_DEBUG(linker, debugBelch("mmapForLinker: start\n"));
1572    pagesize = getpagesize();
1573    size = ROUND_UP(bytes, pagesize);
1574
1575 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1576 mmap_again:
1577
1578    if (mmap_32bit_base != 0) {
1579        map_addr = mmap_32bit_base;
1580    }
1581 #endif
1582
1583    IF_DEBUG(linker, debugBelch("mmapForLinker: \tprotection %#0x\n", PROT_EXEC | PROT_READ | PROT_WRITE));
1584    IF_DEBUG(linker, debugBelch("mmapForLinker: \tflags      %#0x\n", MAP_PRIVATE | TRY_MAP_32BIT | fixed | flags));
1585    result = mmap(map_addr, size, PROT_EXEC|PROT_READ|PROT_WRITE,
1586                     MAP_PRIVATE|TRY_MAP_32BIT|fixed|flags, fd, 0);
1587
1588    if (result == MAP_FAILED) {
1589        sysErrorBelch("mmap %lu bytes at %p",(lnat)size,map_addr);
1590        errorBelch("Try specifying an address with +RTS -xm<addr> -RTS");
1591        stg_exit(EXIT_FAILURE);
1592    }
1593
1594 #if !defined(ALWAYS_PIC) && defined(x86_64_HOST_ARCH)
1595    if (mmap_32bit_base != 0) {
1596        if (result == map_addr) {
1597            mmap_32bit_base = (StgWord8*)map_addr + size;
1598        } else {
1599            if ((W_)result > 0x80000000) {
1600                // oops, we were given memory over 2Gb
1601 #if defined(freebsd_HOST_OS)  || defined(kfreebsdgnu_HOST_OS) || defined(dragonfly_HOST_OS)
1602                // Some platforms require MAP_FIXED.  This is normally
1603                // a bad idea, because MAP_FIXED will overwrite
1604                // existing mappings.
1605                munmap(result,size);
1606                fixed = MAP_FIXED;
1607                goto mmap_again;
1608 #else
1609                barf("loadObj: failed to mmap() memory below 2Gb; asked for %lu bytes at %p.  Try specifying an address with +RTS -xm<addr> -RTS", size, map_addr, result);
1610 #endif
1611            } else {
1612                // hmm, we were given memory somewhere else, but it's
1613                // still under 2Gb so we can use it.  Next time, ask
1614                // for memory right after the place we just got some
1615                mmap_32bit_base = (StgWord8*)result + size;
1616            }
1617        }
1618    } else {
1619        if ((W_)result > 0x80000000) {
1620            // oops, we were given memory over 2Gb
1621            // ... try allocating memory somewhere else?;
1622            debugTrace(DEBUG_linker,"MAP_32BIT didn't work; gave us %lu bytes at 0x%p", bytes, result);
1623            munmap(result, size);
1624
1625            // Set a base address and try again... (guess: 1Gb)
1626            mmap_32bit_base = (void*)0x40000000;
1627            goto mmap_again;
1628        }
1629    }
1630 #endif
1631
1632    IF_DEBUG(linker, debugBelch("mmapForLinker: mapped %lu bytes starting at %p\n", (lnat)size, result));
1633    IF_DEBUG(linker, debugBelch("mmapForLinker: done\n"));
1634    return result;
1635 }
1636 #endif // USE_MMAP
1637
1638 static ObjectCode*
1639 mkOc( char *path, char *image, int imageSize,
1640       char *archiveMemberName
1641 #ifndef USE_MMAP
1642 #ifdef darwin_HOST_OS
1643     , int misalignment
1644 #endif
1645 #endif
1646     ) {
1647    ObjectCode* oc;
1648
1649    IF_DEBUG(linker, debugBelch("mkOc: start\n"));
1650    oc = stgMallocBytes(sizeof(ObjectCode), "loadArchive(oc)");
1651
1652 #  if defined(OBJFORMAT_ELF)
1653    oc->formatName = "ELF";
1654 #  elif defined(OBJFORMAT_PEi386)
1655    oc->formatName = "PEi386";
1656 #  elif defined(OBJFORMAT_MACHO)
1657    oc->formatName = "Mach-O";
1658 #  else
1659    stgFree(oc);
1660    barf("loadObj: not implemented on this platform");
1661 #  endif
1662
1663    oc->image = image;
1664    /* sigh, strdup() isn't a POSIX function, so do it the long way */
1665    oc->fileName = stgMallocBytes( strlen(path)+1, "loadObj" );
1666    strcpy(oc->fileName, path);
1667
1668    if (archiveMemberName) {
1669        oc->archiveMemberName = stgMallocBytes( strlen(archiveMemberName)+1, "loadObj" );
1670        strcpy(oc->archiveMemberName, archiveMemberName);
1671    }
1672    else {
1673        oc->archiveMemberName = NULL;
1674    }
1675
1676    oc->fileSize          = imageSize;
1677    oc->symbols           = NULL;
1678    oc->sections          = NULL;
1679    oc->proddables        = NULL;
1680
1681 #ifndef USE_MMAP
1682 #ifdef darwin_HOST_OS
1683    oc->misalignment = misalignment;
1684 #endif
1685 #endif
1686
1687    /* chain it onto the list of objects */
1688    oc->next              = objects;
1689    objects               = oc;
1690
1691    IF_DEBUG(linker, debugBelch("mkOc: done\n"));
1692    return oc;
1693 }
1694
1695 HsInt
1696 loadArchive( char *path )
1697 {
1698     ObjectCode* oc;
1699     char *image;
1700     int memberSize;
1701     FILE *f;
1702     int n;
1703     size_t thisFileNameSize;
1704     char *fileName;
1705     size_t fileNameSize;
1706     int isObject, isGnuIndex;
1707     char tmp[20];
1708     char *gnuFileIndex;
1709     int gnuFileIndexSize;
1710 #if defined(darwin_HOST_OS)
1711     int i;
1712     uint32_t nfat_arch, nfat_offset, cputype, cpusubtype;
1713 #if defined(i386_HOST_ARCH)
1714     const uint32_t mycputype = CPU_TYPE_X86;
1715     const uint32_t mycpusubtype = CPU_SUBTYPE_X86_ALL;
1716 #elif defined(x86_64_HOST_ARCH)
1717     const uint32_t mycputype = CPU_TYPE_X86_64;
1718     const uint32_t mycpusubtype = CPU_SUBTYPE_X86_64_ALL;
1719 #elif defined(powerpc_HOST_ARCH)
1720     const uint32_t mycputype = CPU_TYPE_POWERPC;
1721     const uint32_t mycpusubtype = CPU_SUBTYPE_POWERPC_ALL;
1722 #elif defined(powerpc64_HOST_ARCH)
1723     const uint32_t mycputype = CPU_TYPE_POWERPC64;
1724     const uint32_t mycpusubtype = CPU_SUBTYPE_POWERPC_ALL;
1725 #else
1726 #error Unknown Darwin architecture
1727 #endif
1728 #if !defined(USE_MMAP)
1729     int misalignment;
1730 #endif
1731 #endif
1732
1733     IF_DEBUG(linker, debugBelch("loadArchive: start\n"));
1734     IF_DEBUG(linker, debugBelch("loadArchive: Loading archive `%s'\n", path));
1735
1736     gnuFileIndex = NULL;
1737     gnuFileIndexSize = 0;
1738
1739     fileNameSize = 32;
1740     fileName = stgMallocBytes(fileNameSize, "loadArchive(fileName)");
1741
1742     f = fopen(path, "rb");
1743     if (!f)
1744         barf("loadObj: can't read `%s'", path);
1745
1746     /* Check if this is an archive by looking for the magic "!<arch>\n"
1747      * string.  Usually, if this fails, we barf and quit.  On Darwin however,
1748      * we may have a fat archive, which contains archives for more than
1749      * one architecture.  Fat archives start with the magic number 0xcafebabe,
1750      * always stored big endian.  If we find a fat_header, we scan through
1751      * the fat_arch structs, searching through for one for our host
1752      * architecture.  If a matching struct is found, we read the offset
1753      * of our archive data (nfat_offset) and seek forward nfat_offset bytes
1754      * from the start of the file.
1755      *
1756      * A subtlety is that all of the members of the fat_header and fat_arch
1757      * structs are stored big endian, so we need to call byte order
1758      * conversion functions.
1759      *
1760      * If we find the appropriate architecture in a fat archive, we gobble
1761      * its magic "!<arch>\n" string and continue processing just as if
1762      * we had a single architecture archive.
1763      */
1764
1765     n = fread ( tmp, 1, 8, f );
1766     if (n != 8)
1767         barf("loadArchive: Failed reading header from `%s'", path);
1768     if (strncmp(tmp, "!<arch>\n", 8) != 0) {
1769
1770 #if defined(darwin_HOST_OS)
1771         /* Not a standard archive, look for a fat archive magic number: */
1772         if (ntohl(*(uint32_t *)tmp) == FAT_MAGIC) {
1773             nfat_arch = ntohl(*(uint32_t *)(tmp + 4));
1774             IF_DEBUG(linker, debugBelch("loadArchive: found a fat archive containing %d architectures\n", nfat_arch));
1775             nfat_offset = 0;
1776
1777             for (i = 0; i < (int)nfat_arch; i++) {
1778                 /* search for the right arch */
1779                 n = fread( tmp, 1, 20, f );
1780                 if (n != 8)
1781                     barf("loadArchive: Failed reading arch from `%s'", path);
1782                 cputype = ntohl(*(uint32_t *)tmp);
1783                 cpusubtype = ntohl(*(uint32_t *)(tmp + 4));
1784
1785                 if (cputype == mycputype && cpusubtype == mycpusubtype) {
1786                     IF_DEBUG(linker, debugBelch("loadArchive: found my archive in a fat archive\n"));
1787                     nfat_offset = ntohl(*(uint32_t *)(tmp + 8));
1788                     break;
1789                 }
1790             }
1791
1792             if (nfat_offset == 0) {
1793                barf ("loadArchive: searched %d architectures, but no host arch found", (int)nfat_arch);
1794             }
1795             else {
1796                 n = fseek( f, nfat_offset, SEEK_SET );
1797                 if (n != 0)
1798                     barf("loadArchive: Failed to seek to arch in `%s'", path);
1799                 n = fread ( tmp, 1, 8, f );
1800                 if (n != 8)
1801                     barf("loadArchive: Failed reading header from `%s'", path);
1802                 if (strncmp(tmp, "!<arch>\n", 8) != 0) {
1803                     barf("loadArchive: couldn't find archive in `%s' at offset %d", path, nfat_offset);
1804                 }
1805             }
1806         }
1807         else {
1808             barf("loadArchive: Neither an archive, nor a fat archive: `%s'", path);
1809         }
1810
1811 #else
1812         barf("loadArchive: Not an archive: `%s'", path);
1813 #endif
1814     }
1815
1816     IF_DEBUG(linker, debugBelch("loadArchive: loading archive contents\n"));
1817
1818     while(1) {
1819         n = fread ( fileName, 1, 16, f );
1820         if (n != 16) {
1821             if (feof(f)) {
1822                 IF_DEBUG(linker, debugBelch("loadArchive: EOF while reading from '%s'\n", path));
1823                 break;
1824             }
1825             else {
1826                 barf("loadArchive: Failed reading file name from `%s'", path);
1827             }
1828         }
1829
1830 #if defined(darwin_HOST_OS)
1831         if (strncmp(fileName, "!<arch>\n", 8) == 0) {
1832             IF_DEBUG(linker, debugBelch("loadArchive: found the start of another archive, breaking\n"));
1833             break;
1834         }
1835 #endif
1836
1837         n = fread ( tmp, 1, 12, f );
1838         if (n != 12)
1839             barf("loadArchive: Failed reading mod time from `%s'", path);
1840         n = fread ( tmp, 1, 6, f );
1841         if (n != 6)
1842             barf("loadArchive: Failed reading owner from `%s'", path);
1843         n = fread ( tmp, 1, 6, f );
1844         if (n != 6)
1845             barf("loadArchive: Failed reading group from `%s'", path);
1846         n = fread ( tmp, 1, 8, f );
1847         if (n != 8)
1848             barf("loadArchive: Failed reading mode from `%s'", path);
1849         n = fread ( tmp, 1, 10, f );
1850         if (n != 10)
1851             barf("loadArchive: Failed reading size from `%s'", path);
1852         tmp[10] = '\0';
1853         for (n = 0; isdigit(tmp[n]); n++);
1854         tmp[n] = '\0';
1855         memberSize = atoi(tmp);
1856
1857         IF_DEBUG(linker, debugBelch("loadArchive: size of this archive member is %d\n", memberSize));
1858         n = fread ( tmp, 1, 2, f );
1859         if (n != 2)
1860             barf("loadArchive: Failed reading magic from `%s'", path);
1861         if (strncmp(tmp, "\x60\x0A", 2) != 0)
1862             barf("loadArchive: Failed reading magic from `%s' at %ld. Got %c%c",
1863                  path, ftell(f), tmp[0], tmp[1]);
1864
1865         isGnuIndex = 0;
1866         /* Check for BSD-variant large filenames */
1867         if (0 == strncmp(fileName, "#1/", 3)) {
1868             fileName[16] = '\0';
1869             if (isdigit(fileName[3])) {
1870                 for (n = 4; isdigit(fileName[n]); n++);
1871                 fileName[n] = '\0';
1872                 thisFileNameSize = atoi(fileName + 3);
1873                 memberSize -= thisFileNameSize;
1874                 if (thisFileNameSize >= fileNameSize) {
1875                     /* Double it to avoid potentially continually
1876                        increasing it by 1 */
1877                     fileNameSize = thisFileNameSize * 2;
1878                     fileName = stgReallocBytes(fileName, fileNameSize, "loadArchive(fileName)");
1879                 }
1880                 n = fread ( fileName, 1, thisFileNameSize, f );
1881                 if (n != (int)thisFileNameSize) {
1882                     barf("loadArchive: Failed reading filename from `%s'",
1883                          path);
1884                 }
1885                 fileName[thisFileNameSize] = 0;
1886
1887                 /* On OS X at least, thisFileNameSize is the size of the
1888                    fileName field, not the length of the fileName
1889                    itself. */
1890                 thisFileNameSize = strlen(fileName);
1891             }
1892             else {
1893                 barf("loadArchive: BSD-variant filename size not found while reading filename from `%s'", path);
1894             }
1895         }
1896         /* Check for GNU file index file */
1897         else if (0 == strncmp(fileName, "//", 2)) {
1898             fileName[0] = '\0';
1899             thisFileNameSize = 0;
1900             isGnuIndex = 1;
1901         }
1902         /* Check for a file in the GNU file index */
1903         else if (fileName[0] == '/') {
1904             if (isdigit(fileName[1])) {
1905                 int i;
1906
1907                 for (n = 2; isdigit(fileName[n]); n++);
1908                 fileName[n] = '\0';
1909                 n = atoi(fileName + 1);
1910
1911                 if (gnuFileIndex == NULL) {
1912                     barf("loadArchive: GNU-variant filename without an index while reading from `%s'", path);
1913                 }
1914                 if (n < 0 || n > gnuFileIndexSize) {
1915                     barf("loadArchive: GNU-variant filename offset %d out of range [0..%d] while reading filename from `%s'", n, gnuFileIndexSize, path);
1916                 }
1917                 if (n != 0 && gnuFileIndex[n - 1] != '\n') {
1918                     barf("loadArchive: GNU-variant filename offset %d invalid (range [0..%d]) while reading filename from `%s'", n, gnuFileIndexSize, path);
1919                 }
1920                 for (i = n; gnuFileIndex[i] != '/'; i++);
1921                 thisFileNameSize = i - n;
1922                 if (thisFileNameSize >= fileNameSize) {
1923                     /* Double it to avoid potentially continually
1924                        increasing it by 1 */
1925                     fileNameSize = thisFileNameSize * 2;
1926                     fileName = stgReallocBytes(fileName, fileNameSize, "loadArchive(fileName)");
1927                 }
1928                 memcpy(fileName, gnuFileIndex + n, thisFileNameSize);
1929                 fileName[thisFileNameSize] = '\0';
1930             }
1931             else if (fileName[1] == ' ') {
1932                 fileName[0] = '\0';
1933                 thisFileNameSize = 0;
1934             }
1935             else {
1936                 barf("loadArchive: GNU-variant filename offset not found while reading filename from `%s'", path);
1937             }
1938         }
1939         /* Finally, the case where the filename field actually contains
1940            the filename */
1941         else {
1942             /* GNU ar terminates filenames with a '/', this allowing
1943                spaces in filenames. So first look to see if there is a
1944                terminating '/'. */
1945             for (thisFileNameSize = 0;
1946                  thisFileNameSize < 16;
1947                  thisFileNameSize++) {
1948                 if (fileName[thisFileNameSize] == '/') {
1949                     fileName[thisFileNameSize] = '\0';
1950                     break;
1951                 }
1952             }
1953             /* If we didn't find a '/', then a space teminates the
1954                filename. Note that if we don't find one, then
1955                thisFileNameSize ends up as 16, and we already have the
1956                '\0' at the end. */
1957             if (thisFileNameSize == 16) {
1958                 for (thisFileNameSize = 0;
1959                      thisFileNameSize < 16;
1960                      thisFileNameSize++) {
1961                     if (fileName[thisFileNameSize] == ' ') {
1962                         fileName[thisFileNameSize] = '\0';
1963                         break;
1964                     }
1965                 }
1966             }
1967         }
1968
1969         IF_DEBUG(linker,
1970                  debugBelch("loadArchive: Found member file `%s'\n", fileName));
1971
1972         isObject = thisFileNameSize >= 2
1973                 && fileName[thisFileNameSize - 2] == '.'
1974                 && fileName[thisFileNameSize - 1] == 'o';
1975
1976         IF_DEBUG(linker, debugBelch("loadArchive: \tthisFileNameSize = %d\n", (int)thisFileNameSize));
1977         IF_DEBUG(linker, debugBelch("loadArchive: \tisObject = %d\n", isObject));
1978
1979         if (isObject) {
1980             char *archiveMemberName;
1981
1982             IF_DEBUG(linker, debugBelch("loadArchive: Member is an object file...loading...\n"));
1983
1984             /* We can't mmap from the archive directly, as object
1985                files need to be 8-byte aligned but files in .ar
1986                archives are 2-byte aligned. When possible we use mmap
1987                to get some anonymous memory, as on 64-bit platforms if
1988                we use malloc then we can be given memory above 2^32.
1989                In the mmap case we're probably wasting lots of space;
1990                we could do better. */
1991 #if defined(USE_MMAP)
1992             image = mmapForLinker(memberSize, MAP_ANONYMOUS, -1);
1993 #elif defined(darwin_HOST_OS)
1994             /* See loadObj() */
1995             misalignment = machoGetMisalignment(f);
1996             image = stgMallocBytes(memberSize + misalignment, "loadArchive(image)");
1997             image += misalignment;
1998 #else
1999             image = stgMallocBytes(memberSize, "loadArchive(image)");
2000 #endif
2001             n = fread ( image, 1, memberSize, f );
2002             if (n != memberSize) {
2003                 barf("loadArchive: error whilst reading `%s'", path);
2004             }
2005
2006             archiveMemberName = stgMallocBytes(strlen(path) + thisFileNameSize + 3,
2007                                                "loadArchive(file)");
2008             sprintf(archiveMemberName, "%s(%.*s)",
2009                     path, (int)thisFileNameSize, fileName);
2010
2011             oc = mkOc(path, image, memberSize, archiveMemberName
2012 #ifndef USE_MMAP
2013 #ifdef darwin_HOST_OS
2014                      , misalignment
2015 #endif
2016 #endif
2017                      );
2018
2019             stgFree(archiveMemberName);
2020
2021             if (0 == loadOc(oc)) {
2022                 stgFree(fileName);
2023                 return 0;
2024             }
2025         }
2026         else if (isGnuIndex) {
2027             if (gnuFileIndex != NULL) {
2028                 barf("loadArchive: GNU-variant index found, but already have an index, while reading filename from `%s'", path);
2029             }
2030             IF_DEBUG(linker, debugBelch("loadArchive: Found GNU-variant file index\n"));
2031 #ifdef USE_MMAP
2032             gnuFileIndex = mmapForLinker(memberSize + 1, MAP_ANONYMOUS, -1);
2033 #else
2034             gnuFileIndex = stgMallocBytes(memberSize + 1, "loadArchive(image)");
2035 #endif
2036             n = fread ( gnuFileIndex, 1, memberSize, f );
2037             if (n != memberSize) {
2038                 barf("loadArchive: error whilst reading `%s'", path);
2039             }
2040             gnuFileIndex[memberSize] = '/';
2041             gnuFileIndexSize = memberSize;
2042         }
2043         else {
2044             IF_DEBUG(linker, debugBelch("loadArchive: '%s' does not appear to be an object file\n", fileName));
2045             n = fseek(f, memberSize, SEEK_CUR);
2046             if (n != 0)
2047                 barf("loadArchive: error whilst seeking by %d in `%s'",
2048                      memberSize, path);
2049         }
2050
2051         /* .ar files are 2-byte aligned */
2052         if (memberSize % 2) {
2053             IF_DEBUG(linker, debugBelch("loadArchive: trying to read one pad byte\n"));
2054             n = fread ( tmp, 1, 1, f );
2055             if (n != 1) {
2056                 if (feof(f)) {
2057                     IF_DEBUG(linker, debugBelch("loadArchive: found EOF while reading one pad byte\n"));
2058                     break;
2059                 }
2060                 else {
2061                     barf("loadArchive: Failed reading padding from `%s'", path);
2062                 }
2063             }
2064             IF_DEBUG(linker, debugBelch("loadArchive: successfully read one pad byte\n"));
2065         }
2066         IF_DEBUG(linker, debugBelch("loadArchive: reached end of archive loading while loop\n"));
2067     }
2068
2069     fclose(f);
2070
2071     stgFree(fileName);
2072     if (gnuFileIndex != NULL) {
2073 #ifdef USE_MMAP
2074         munmap(gnuFileIndex, gnuFileIndexSize + 1);
2075 #else
2076         stgFree(gnuFileIndex);
2077 #endif
2078     }
2079
2080     IF_DEBUG(linker, debugBelch("loadArchive: done\n"));
2081     return 1;
2082 }
2083
2084 /* -----------------------------------------------------------------------------
2085  * Load an obj (populate the global symbol table, but don't resolve yet)
2086  *
2087  * Returns: 1 if ok, 0 on error.
2088  */
2089 HsInt
2090 loadObj( char *path )
2091 {
2092    ObjectCode* oc;
2093    char *image;
2094    int fileSize;
2095    struct stat st;
2096    int r;
2097 #ifdef USE_MMAP
2098    int fd;
2099 #else
2100    FILE *f;
2101 #  if defined(darwin_HOST_OS)
2102    int misalignment;
2103 #  endif
2104 #endif
2105    IF_DEBUG(linker, debugBelch("loadObj %s\n", path));
2106
2107    initLinker();
2108
2109    /* debugBelch("loadObj %s\n", path ); */
2110
2111    /* Check that we haven't already loaded this object.
2112       Ignore requests to load multiple times */
2113    {
2114        ObjectCode *o;
2115        int is_dup = 0;
2116        for (o = objects; o; o = o->next) {
2117           if (0 == strcmp(o->fileName, path)) {
2118              is_dup = 1;
2119              break; /* don't need to search further */
2120           }
2121        }
2122        if (is_dup) {
2123           IF_DEBUG(linker, debugBelch(
2124             "GHCi runtime linker: warning: looks like you're trying to load the\n"
2125             "same object file twice:\n"
2126             "   %s\n"
2127             "GHCi will ignore this, but be warned.\n"
2128             , path));
2129           return 1; /* success */
2130        }
2131    }
2132
2133    r = stat(path, &st);
2134    if (r == -1) {
2135        IF_DEBUG(linker, debugBelch("File doesn't exist\n"));
2136        return 0;
2137    }
2138
2139    fileSize = st.st_size;
2140
2141 #ifdef USE_MMAP
2142    /* On many architectures malloc'd memory isn't executable, so we need to use mmap. */
2143
2144 #if defined(openbsd_HOST_OS)
2145    fd = open(path, O_RDONLY, S_IRUSR);
2146 #else
2147    fd = open(path, O_RDONLY);
2148 #endif
2149    if (fd == -1)
2150       barf("loadObj: can't open `%s'", path);
2151
2152    image = mmapForLinker(fileSize, 0, fd);
2153
2154    close(fd);
2155
2156 #else /* !USE_MMAP */
2157    /* load the image into memory */
2158    f = fopen(path, "rb");
2159    if (!f)
2160        barf("loadObj: can't read `%s'", path);
2161
2162 #   if defined(mingw32_HOST_OS)
2163         // TODO: We would like to use allocateExec here, but allocateExec
2164         //       cannot currently allocate blocks large enough.
2165     image = VirtualAlloc(NULL, fileSize, MEM_RESERVE | MEM_COMMIT,
2166                              PAGE_EXECUTE_READWRITE);
2167 #   elif defined(darwin_HOST_OS)
2168     // In a Mach-O .o file, all sections can and will be misaligned
2169     // if the total size of the headers is not a multiple of the
2170     // desired alignment. This is fine for .o files that only serve
2171     // as input for the static linker, but it's not fine for us,
2172     // as SSE (used by gcc for floating point) and Altivec require
2173     // 16-byte alignment.
2174     // We calculate the correct alignment from the header before
2175     // reading the file, and then we misalign image on purpose so
2176     // that the actual sections end up aligned again.
2177    misalignment = machoGetMisalignment(f);
2178    image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
2179    image += misalignment;
2180 #  else
2181    image = stgMallocBytes(fileSize, "loadObj(image)");
2182 #  endif
2183
2184    {
2185        int n;
2186        n = fread ( image, 1, fileSize, f );
2187        if (n != fileSize)
2188            barf("loadObj: error whilst reading `%s'", path);
2189    }
2190    fclose(f);
2191 #endif /* USE_MMAP */
2192
2193    oc = mkOc(path, image, fileSize, NULL
2194 #ifndef USE_MMAP
2195 #ifdef darwin_HOST_OS
2196             , misalignment
2197 #endif
2198 #endif
2199             );
2200
2201    return loadOc(oc);
2202 }
2203
2204 static HsInt
2205 loadOc( ObjectCode* oc ) {
2206    int r;
2207
2208    IF_DEBUG(linker, debugBelch("loadOc: start\n"));
2209
2210 #  if defined(OBJFORMAT_MACHO) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
2211    r = ocAllocateSymbolExtras_MachO ( oc );
2212    if (!r) {
2213        IF_DEBUG(linker, debugBelch("loadOc: ocAllocateSymbolExtras_MachO failed\n"));
2214        return r;
2215    }
2216 #  elif defined(OBJFORMAT_ELF) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
2217    r = ocAllocateSymbolExtras_ELF ( oc );
2218    if (!r) {
2219        IF_DEBUG(linker, debugBelch("loadOc: ocAllocateSymbolExtras_ELF failed\n"));
2220        return r;
2221    }
2222 #endif
2223
2224    /* verify the in-memory image */
2225 #  if defined(OBJFORMAT_ELF)
2226    r = ocVerifyImage_ELF ( oc );
2227 #  elif defined(OBJFORMAT_PEi386)
2228    r = ocVerifyImage_PEi386 ( oc );
2229 #  elif defined(OBJFORMAT_MACHO)
2230    r = ocVerifyImage_MachO ( oc );
2231 #  else
2232    barf("loadObj: no verify method");
2233 #  endif
2234    if (!r) {
2235        IF_DEBUG(linker, debugBelch("loadOc: ocVerifyImage_* failed\n"));
2236        return r;
2237    }
2238
2239    /* build the symbol list for this image */
2240 #  if defined(OBJFORMAT_ELF)
2241    r = ocGetNames_ELF ( oc );
2242 #  elif defined(OBJFORMAT_PEi386)
2243    r = ocGetNames_PEi386 ( oc );
2244 #  elif defined(OBJFORMAT_MACHO)
2245    r = ocGetNames_MachO ( oc );
2246 #  else
2247    barf("loadObj: no getNames method");
2248 #  endif
2249    if (!r) {
2250        IF_DEBUG(linker, debugBelch("loadOc: ocGetNames_* failed\n"));
2251        return r;
2252    }
2253
2254    /* loaded, but not resolved yet */
2255    oc->status = OBJECT_LOADED;
2256    IF_DEBUG(linker, debugBelch("loadOc: done.\n"));
2257
2258    return 1;
2259 }
2260
2261 /* -----------------------------------------------------------------------------
2262  * resolve all the currently unlinked objects in memory
2263  *
2264  * Returns: 1 if ok, 0 on error.
2265  */
2266 HsInt
2267 resolveObjs( void )
2268 {
2269     ObjectCode *oc;
2270     int r;
2271
2272     IF_DEBUG(linker, debugBelch("resolveObjs: start\n"));
2273     initLinker();
2274
2275     for (oc = objects; oc; oc = oc->next) {
2276         if (oc->status != OBJECT_RESOLVED) {
2277 #           if defined(OBJFORMAT_ELF)
2278             r = ocResolve_ELF ( oc );
2279 #           elif defined(OBJFORMAT_PEi386)
2280             r = ocResolve_PEi386 ( oc );
2281 #           elif defined(OBJFORMAT_MACHO)
2282             r = ocResolve_MachO ( oc );
2283 #           else
2284             barf("resolveObjs: not implemented on this platform");
2285 #           endif
2286             if (!r) { return r; }
2287             oc->status = OBJECT_RESOLVED;
2288         }
2289     }
2290     IF_DEBUG(linker, debugBelch("resolveObjs: done\n"));
2291     return 1;
2292 }
2293
2294 /* -----------------------------------------------------------------------------
2295  * delete an object from the pool
2296  */
2297 HsInt
2298 unloadObj( char *path )
2299 {
2300     ObjectCode *oc, *prev;
2301     HsBool unloadedAnyObj = HS_BOOL_FALSE;
2302
2303     ASSERT(symhash != NULL);
2304     ASSERT(objects != NULL);
2305
2306     initLinker();
2307
2308     prev = NULL;
2309     for (oc = objects; oc; prev = oc, oc = oc->next) {
2310         if (!strcmp(oc->fileName,path)) {
2311
2312             /* Remove all the mappings for the symbols within this
2313              * object..
2314              */
2315             {
2316                 int i;
2317                 for (i = 0; i < oc->n_symbols; i++) {
2318                    if (oc->symbols[i] != NULL) {
2319                        removeStrHashTable(symhash, oc->symbols[i], NULL);
2320                    }
2321                 }
2322             }
2323
2324             if (prev == NULL) {
2325                 objects = oc->next;
2326             } else {
2327                 prev->next = oc->next;
2328             }
2329
2330             // We're going to leave this in place, in case there are
2331             // any pointers from the heap into it:
2332                 // #ifdef mingw32_HOST_OS
2333                 //  VirtualFree(oc->image);
2334                 // #else
2335             //  stgFree(oc->image);
2336             // #endif
2337             stgFree(oc->fileName);
2338             stgFree(oc->symbols);
2339             stgFree(oc->sections);
2340             stgFree(oc);
2341
2342             /* This could be a member of an archive so continue
2343              * unloading other members. */
2344             unloadedAnyObj = HS_BOOL_TRUE;
2345         }
2346     }
2347
2348     if (unloadedAnyObj) {
2349         return 1;
2350     }
2351     else {
2352         errorBelch("unloadObj: can't find `%s' to unload", path);
2353         return 0;
2354     }
2355 }
2356
2357 /* -----------------------------------------------------------------------------
2358  * Sanity checking.  For each ObjectCode, maintain a list of address ranges
2359  * which may be prodded during relocation, and abort if we try and write
2360  * outside any of these.
2361  */
2362 static void
2363 addProddableBlock ( ObjectCode* oc, void* start, int size )
2364 {
2365    ProddableBlock* pb
2366       = stgMallocBytes(sizeof(ProddableBlock), "addProddableBlock");
2367
2368    IF_DEBUG(linker, debugBelch("addProddableBlock: %p %p %d\n", oc, start, size));
2369    ASSERT(size > 0);
2370    pb->start      = start;
2371    pb->size       = size;
2372    pb->next       = oc->proddables;
2373    oc->proddables = pb;
2374 }
2375
2376 static void
2377 checkProddableBlock (ObjectCode *oc, void *addr )
2378 {
2379    ProddableBlock* pb;
2380
2381    for (pb = oc->proddables; pb != NULL; pb = pb->next) {
2382       char* s = (char*)(pb->start);
2383       char* e = s + pb->size - 1;
2384       char* a = (char*)addr;
2385       /* Assumes that the biggest fixup involves a 4-byte write.  This
2386          probably needs to be changed to 8 (ie, +7) on 64-bit
2387          plats. */
2388       if (a >= s && (a+3) <= e) return;
2389    }
2390    barf("checkProddableBlock: invalid fixup in runtime linker");
2391 }
2392
2393 /* -----------------------------------------------------------------------------
2394  * Section management.
2395  */
2396 static void
2397 addSection ( ObjectCode* oc, SectionKind kind,
2398                          void* start, void* end )
2399 {
2400    Section* s   = stgMallocBytes(sizeof(Section), "addSection");
2401    s->start     = start;
2402    s->end       = end;
2403    s->kind      = kind;
2404    s->next      = oc->sections;
2405    oc->sections = s;
2406
2407    IF_DEBUG(linker, debugBelch("addSection: %p-%p (size %ld), kind %d\n",
2408                                start, ((char*)end)-1, (long)end - (long)start + 1, kind ));
2409 }
2410
2411
2412 /* --------------------------------------------------------------------------
2413  * Symbol Extras.
2414  * This is about allocating a small chunk of memory for every symbol in the
2415  * object file. We make sure that the SymboLExtras are always "in range" of
2416  * limited-range PC-relative instructions on various platforms by allocating
2417  * them right next to the object code itself.
2418  */
2419
2420 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
2421
2422 /*
2423   ocAllocateSymbolExtras
2424
2425   Allocate additional space at the end of the object file image to make room
2426   for jump islands (powerpc, x86_64) and GOT entries (x86_64).
2427
2428   PowerPC relative branch instructions have a 24 bit displacement field.
2429   As PPC code is always 4-byte-aligned, this yields a +-32MB range.
2430   If a particular imported symbol is outside this range, we have to redirect
2431   the jump to a short piece of new code that just loads the 32bit absolute
2432   address and jumps there.
2433   On x86_64, PC-relative jumps and PC-relative accesses to the GOT are limited
2434   to 32 bits (+-2GB).
2435
2436   This function just allocates space for one SymbolExtra for every
2437   undefined symbol in the object file. The code for the jump islands is
2438   filled in by makeSymbolExtra below.
2439 */
2440
2441 static int ocAllocateSymbolExtras( ObjectCode* oc, int count, int first )
2442 {
2443 #ifdef USE_MMAP
2444   int pagesize, n, m;
2445 #endif
2446   int aligned;
2447 #ifndef USE_MMAP
2448   int misalignment = 0;
2449 #ifdef darwin_HOST_OS
2450   misalignment = oc->misalignment;
2451 #endif
2452 #endif
2453
2454   if( count > 0 )
2455   {
2456     // round up to the nearest 4
2457     aligned = (oc->fileSize + 3) & ~3;
2458
2459 #ifdef USE_MMAP
2460     pagesize = getpagesize();
2461     n = ROUND_UP( oc->fileSize, pagesize );
2462     m = ROUND_UP( aligned + sizeof (SymbolExtra) * count, pagesize );
2463
2464     /* we try to use spare space at the end of the last page of the
2465      * image for the jump islands, but if there isn't enough space
2466      * then we have to map some (anonymously, remembering MAP_32BIT).
2467      */
2468     if( m > n ) // we need to allocate more pages
2469     {
2470         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
2471                                           MAP_ANONYMOUS, -1);
2472     }
2473     else
2474     {
2475         oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
2476     }
2477 #else
2478     oc->image -= misalignment;
2479     oc->image = stgReallocBytes( oc->image,
2480                                  misalignment +
2481                                  aligned + sizeof (SymbolExtra) * count,
2482                                  "ocAllocateSymbolExtras" );
2483     oc->image += misalignment;
2484
2485     oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
2486 #endif /* USE_MMAP */
2487
2488     memset( oc->symbol_extras, 0, sizeof (SymbolExtra) * count );
2489   }
2490   else
2491     oc->symbol_extras = NULL;
2492
2493   oc->first_symbol_extra = first;
2494   oc->n_symbol_extras = count;
2495
2496   return 1;
2497 }
2498
2499 static SymbolExtra* makeSymbolExtra( ObjectCode* oc,
2500                                      unsigned long symbolNumber,
2501                                      unsigned long target )
2502 {
2503   SymbolExtra *extra;
2504
2505   ASSERT( symbolNumber >= oc->first_symbol_extra
2506         && symbolNumber - oc->first_symbol_extra < oc->n_symbol_extras);
2507
2508   extra = &oc->symbol_extras[symbolNumber - oc->first_symbol_extra];
2509
2510 #ifdef powerpc_HOST_ARCH
2511   // lis r12, hi16(target)
2512   extra->jumpIsland.lis_r12     = 0x3d80;
2513   extra->jumpIsland.hi_addr     = target >> 16;
2514
2515   // ori r12, r12, lo16(target)
2516   extra->jumpIsland.ori_r12_r12 = 0x618c;
2517   extra->jumpIsland.lo_addr     = target & 0xffff;
2518
2519   // mtctr r12
2520   extra->jumpIsland.mtctr_r12   = 0x7d8903a6;
2521
2522   // bctr
2523   extra->jumpIsland.bctr        = 0x4e800420;
2524 #endif
2525 #ifdef x86_64_HOST_ARCH
2526         // jmp *-14(%rip)
2527   static uint8_t jmp[] = { 0xFF, 0x25, 0xF2, 0xFF, 0xFF, 0xFF };
2528   extra->addr = target;
2529   memcpy(extra->jumpIsland, jmp, 6);
2530 #endif
2531
2532   return extra;
2533 }
2534
2535 #endif
2536
2537 /* --------------------------------------------------------------------------
2538  * PowerPC specifics (instruction cache flushing)
2539  * ------------------------------------------------------------------------*/
2540
2541 #ifdef powerpc_HOST_ARCH
2542 /*
2543    ocFlushInstructionCache
2544
2545    Flush the data & instruction caches.
2546    Because the PPC has split data/instruction caches, we have to
2547    do that whenever we modify code at runtime.
2548  */
2549
2550 static void
2551 ocFlushInstructionCacheFrom(void* begin, size_t length)
2552 {
2553     size_t         n = (length + 3) / 4;
2554     unsigned long* p = begin;
2555
2556     while (n--)
2557     {
2558         __asm__ volatile ( "dcbf 0,%0\n\t"
2559                            "sync\n\t"
2560                            "icbi 0,%0"
2561                            :
2562                            : "r" (p)
2563                          );
2564         p++;
2565     }
2566     __asm__ volatile ( "sync\n\t"
2567                        "isync"
2568                      );
2569 }
2570
2571 static void
2572 ocFlushInstructionCache( ObjectCode *oc )
2573 {
2574     /* The main object code */
2575     ocFlushInstructionCacheFrom(oc->image + oc->misalignment, oc->fileSize);
2576
2577     /* Jump Islands */
2578     ocFlushInstructionCacheFrom(oc->symbol_extras, sizeof(SymbolExtra) * oc->n_symbol_extras);
2579 }
2580 #endif /* powerpc_HOST_ARCH */
2581
2582
2583 /* --------------------------------------------------------------------------
2584  * PEi386 specifics (Win32 targets)
2585  * ------------------------------------------------------------------------*/
2586
2587 /* The information for this linker comes from
2588       Microsoft Portable Executable
2589       and Common Object File Format Specification
2590       revision 5.1 January 1998
2591    which SimonM says comes from the MS Developer Network CDs.
2592
2593    It can be found there (on older CDs), but can also be found
2594    online at:
2595
2596       http://www.microsoft.com/hwdev/hardware/PECOFF.asp
2597
2598    (this is Rev 6.0 from February 1999).
2599
2600    Things move, so if that fails, try searching for it via
2601
2602       http://www.google.com/search?q=PE+COFF+specification
2603
2604    The ultimate reference for the PE format is the Winnt.h
2605    header file that comes with the Platform SDKs; as always,
2606    implementations will drift wrt their documentation.
2607
2608    A good background article on the PE format is Matt Pietrek's
2609    March 1994 article in Microsoft System Journal (MSJ)
2610    (Vol.9, No. 3): "Peering Inside the PE: A Tour of the
2611    Win32 Portable Executable File Format." The info in there
2612    has recently been updated in a two part article in
2613    MSDN magazine, issues Feb and March 2002,
2614    "Inside Windows: An In-Depth Look into the Win32 Portable
2615    Executable File Format"
2616
2617    John Levine's book "Linkers and Loaders" contains useful
2618    info on PE too.
2619 */
2620
2621
2622 #if defined(OBJFORMAT_PEi386)
2623
2624
2625
2626 typedef unsigned char  UChar;
2627 typedef unsigned short UInt16;
2628 typedef unsigned int   UInt32;
2629 typedef          int   Int32;
2630
2631
2632 typedef
2633    struct {
2634       UInt16 Machine;
2635       UInt16 NumberOfSections;
2636       UInt32 TimeDateStamp;
2637       UInt32 PointerToSymbolTable;
2638       UInt32 NumberOfSymbols;
2639       UInt16 SizeOfOptionalHeader;
2640       UInt16 Characteristics;
2641    }
2642    COFF_header;
2643
2644 #define sizeof_COFF_header 20
2645
2646
2647 typedef
2648    struct {
2649       UChar  Name[8];
2650       UInt32 VirtualSize;
2651       UInt32 VirtualAddress;
2652       UInt32 SizeOfRawData;
2653       UInt32 PointerToRawData;
2654       UInt32 PointerToRelocations;
2655       UInt32 PointerToLinenumbers;
2656       UInt16 NumberOfRelocations;
2657       UInt16 NumberOfLineNumbers;
2658       UInt32 Characteristics;
2659    }
2660    COFF_section;
2661
2662 #define sizeof_COFF_section 40
2663
2664
2665 typedef
2666    struct {
2667       UChar  Name[8];
2668       UInt32 Value;
2669       UInt16 SectionNumber;
2670       UInt16 Type;
2671       UChar  StorageClass;
2672       UChar  NumberOfAuxSymbols;
2673    }
2674    COFF_symbol;
2675
2676 #define sizeof_COFF_symbol 18
2677
2678
2679 typedef
2680    struct {
2681       UInt32 VirtualAddress;
2682       UInt32 SymbolTableIndex;
2683       UInt16 Type;
2684    }
2685    COFF_reloc;
2686
2687 #define sizeof_COFF_reloc 10
2688
2689
2690 /* From PE spec doc, section 3.3.2 */
2691 /* Note use of MYIMAGE_* since IMAGE_* are already defined in
2692    windows.h -- for the same purpose, but I want to know what I'm
2693    getting, here. */
2694 #define MYIMAGE_FILE_RELOCS_STRIPPED     0x0001
2695 #define MYIMAGE_FILE_EXECUTABLE_IMAGE    0x0002
2696 #define MYIMAGE_FILE_DLL                 0x2000
2697 #define MYIMAGE_FILE_SYSTEM              0x1000
2698 #define MYIMAGE_FILE_BYTES_REVERSED_HI   0x8000
2699 #define MYIMAGE_FILE_BYTES_REVERSED_LO   0x0080
2700 #define MYIMAGE_FILE_32BIT_MACHINE       0x0100
2701
2702 /* From PE spec doc, section 5.4.2 and 5.4.4 */
2703 #define MYIMAGE_SYM_CLASS_EXTERNAL       2
2704 #define MYIMAGE_SYM_CLASS_STATIC         3
2705 #define MYIMAGE_SYM_UNDEFINED            0
2706
2707 /* From PE spec doc, section 4.1 */
2708 #define MYIMAGE_SCN_CNT_CODE             0x00000020
2709 #define MYIMAGE_SCN_CNT_INITIALIZED_DATA 0x00000040
2710 #define MYIMAGE_SCN_LNK_NRELOC_OVFL      0x01000000
2711
2712 /* From PE spec doc, section 5.2.1 */
2713 #define MYIMAGE_REL_I386_DIR32           0x0006
2714 #define MYIMAGE_REL_I386_REL32           0x0014
2715
2716
2717 /* We use myindex to calculate array addresses, rather than
2718    simply doing the normal subscript thing.  That's because
2719    some of the above structs have sizes which are not
2720    a whole number of words.  GCC rounds their sizes up to a
2721    whole number of words, which means that the address calcs
2722    arising from using normal C indexing or pointer arithmetic
2723    are just plain wrong.  Sigh.
2724 */
2725 static UChar *
2726 myindex ( int scale, void* base, int index )
2727 {
2728    return
2729       ((UChar*)base) + scale * index;
2730 }
2731
2732
2733 static void
2734 printName ( UChar* name, UChar* strtab )
2735 {
2736    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2737       UInt32 strtab_offset = * (UInt32*)(name+4);
2738       debugBelch("%s", strtab + strtab_offset );
2739    } else {
2740       int i;
2741       for (i = 0; i < 8; i++) {
2742          if (name[i] == 0) break;
2743          debugBelch("%c", name[i] );
2744       }
2745    }
2746 }
2747
2748
2749 static void
2750 copyName ( UChar* name, UChar* strtab, UChar* dst, int dstSize )
2751 {
2752    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2753       UInt32 strtab_offset = * (UInt32*)(name+4);
2754       strncpy ( (char*)dst, (char*)strtab+strtab_offset, dstSize );
2755       dst[dstSize-1] = 0;
2756    } else {
2757       int i = 0;
2758       while (1) {
2759          if (i >= 8) break;
2760          if (name[i] == 0) break;
2761          dst[i] = name[i];
2762          i++;
2763       }
2764       dst[i] = 0;
2765    }
2766 }
2767
2768
2769 static UChar *
2770 cstring_from_COFF_symbol_name ( UChar* name, UChar* strtab )
2771 {
2772    UChar* newstr;
2773    /* If the string is longer than 8 bytes, look in the
2774       string table for it -- this will be correctly zero terminated.
2775    */
2776    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2777       UInt32 strtab_offset = * (UInt32*)(name+4);
2778       return ((UChar*)strtab) + strtab_offset;
2779    }
2780    /* Otherwise, if shorter than 8 bytes, return the original,
2781       which by defn is correctly terminated.
2782    */
2783    if (name[7]==0) return name;
2784    /* The annoying case: 8 bytes.  Copy into a temporary
2785       (XXX which is never freed ...)
2786    */
2787    newstr = stgMallocBytes(9, "cstring_from_COFF_symbol_name");
2788    ASSERT(newstr);
2789    strncpy((char*)newstr,(char*)name,8);
2790    newstr[8] = 0;
2791    return newstr;
2792 }
2793
2794 /* Getting the name of a section is mildly tricky, so we make a
2795    function for it.  Sadly, in one case we have to copy the string
2796    (when it is exactly 8 bytes long there's no trailing '\0'), so for
2797    consistency we *always* copy the string; the caller must free it
2798 */
2799 static char *
2800 cstring_from_section_name (UChar* name, UChar* strtab)
2801 {
2802     char *newstr;
2803
2804     if (name[0]=='/') {
2805         int strtab_offset = strtol((char*)name+1,NULL,10);
2806         int len = strlen(((char*)strtab) + strtab_offset);
2807
2808         newstr = stgMallocBytes(len, "cstring_from_section_symbol_name");
2809         strcpy((char*)newstr, (char*)((UChar*)strtab) + strtab_offset);
2810         return newstr;
2811     }
2812     else
2813     {
2814         newstr = stgMallocBytes(9, "cstring_from_section_symbol_name");
2815         ASSERT(newstr);
2816         strncpy((char*)newstr,(char*)name,8);
2817         newstr[8] = 0;
2818         return newstr;
2819     }
2820 }
2821
2822 /* Just compares the short names (first 8 chars) */
2823 static COFF_section *
2824 findPEi386SectionCalled ( ObjectCode* oc,  UChar* name )
2825 {
2826    int i;
2827    COFF_header* hdr
2828       = (COFF_header*)(oc->image);
2829    COFF_section* sectab
2830       = (COFF_section*) (
2831            ((UChar*)(oc->image))
2832            + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2833         );
2834    for (i = 0; i < hdr->NumberOfSections; i++) {
2835       UChar* n1;
2836       UChar* n2;
2837       COFF_section* section_i
2838          = (COFF_section*)
2839            myindex ( sizeof_COFF_section, sectab, i );
2840       n1 = (UChar*) &(section_i->Name);
2841       n2 = name;
2842       if (n1[0]==n2[0] && n1[1]==n2[1] && n1[2]==n2[2] &&
2843           n1[3]==n2[3] && n1[4]==n2[4] && n1[5]==n2[5] &&
2844           n1[6]==n2[6] && n1[7]==n2[7])
2845          return section_i;
2846    }
2847
2848    return NULL;
2849 }
2850
2851
2852 static void
2853 zapTrailingAtSign ( UChar* sym )
2854 {
2855 #  define my_isdigit(c) ((c) >= '0' && (c) <= '9')
2856    int i, j;
2857    if (sym[0] == 0) return;
2858    i = 0;
2859    while (sym[i] != 0) i++;
2860    i--;
2861    j = i;
2862    while (j > 0 && my_isdigit(sym[j])) j--;
2863    if (j > 0 && sym[j] == '@' && j != i) sym[j] = 0;
2864 #  undef my_isdigit
2865 }
2866
2867 static void *
2868 lookupSymbolInDLLs ( UChar *lbl )
2869 {
2870     OpenedDLL* o_dll;
2871     void *sym;
2872
2873     for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
2874         /* debugBelch("look in %s for %s\n", o_dll->name, lbl); */
2875
2876         if (lbl[0] == '_') {
2877             /* HACK: if the name has an initial underscore, try stripping
2878                it off & look that up first. I've yet to verify whether there's
2879                a Rule that governs whether an initial '_' *should always* be
2880                stripped off when mapping from import lib name to the DLL name.
2881             */
2882             sym = GetProcAddress(o_dll->instance, (char*)(lbl+1));
2883             if (sym != NULL) {
2884                 /*debugBelch("found %s in %s\n", lbl+1,o_dll->name);*/
2885                 return sym;
2886             }
2887         }
2888         sym = GetProcAddress(o_dll->instance, (char*)lbl);
2889         if (sym != NULL) {
2890             /*debugBelch("found %s in %s\n", lbl,o_dll->name);*/
2891             return sym;
2892            }
2893     }
2894     return NULL;
2895 }
2896
2897
2898 static int
2899 ocVerifyImage_PEi386 ( ObjectCode* oc )
2900 {
2901    int i;
2902    UInt32 j, noRelocs;
2903    COFF_header*  hdr;
2904    COFF_section* sectab;
2905    COFF_symbol*  symtab;
2906    UChar*        strtab;
2907    /* debugBelch("\nLOADING %s\n", oc->fileName); */
2908    hdr = (COFF_header*)(oc->image);
2909    sectab = (COFF_section*) (
2910                ((UChar*)(oc->image))
2911                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2912             );
2913    symtab = (COFF_symbol*) (
2914                ((UChar*)(oc->image))
2915                + hdr->PointerToSymbolTable
2916             );
2917    strtab = ((UChar*)symtab)
2918             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2919
2920    if (hdr->Machine != 0x14c) {
2921       errorBelch("%s: Not x86 PEi386", oc->fileName);
2922       return 0;
2923    }
2924    if (hdr->SizeOfOptionalHeader != 0) {
2925       errorBelch("%s: PEi386 with nonempty optional header", oc->fileName);
2926       return 0;
2927    }
2928    if ( /* (hdr->Characteristics & MYIMAGE_FILE_RELOCS_STRIPPED) || */
2929         (hdr->Characteristics & MYIMAGE_FILE_EXECUTABLE_IMAGE) ||
2930         (hdr->Characteristics & MYIMAGE_FILE_DLL) ||
2931         (hdr->Characteristics & MYIMAGE_FILE_SYSTEM) ) {
2932       errorBelch("%s: Not a PEi386 object file", oc->fileName);
2933       return 0;
2934    }
2935    if ( (hdr->Characteristics & MYIMAGE_FILE_BYTES_REVERSED_HI)
2936         /* || !(hdr->Characteristics & MYIMAGE_FILE_32BIT_MACHINE) */ ) {
2937       errorBelch("%s: Invalid PEi386 word size or endiannness: %d",
2938                  oc->fileName,
2939                  (int)(hdr->Characteristics));
2940       return 0;
2941    }
2942    /* If the string table size is way crazy, this might indicate that
2943       there are more than 64k relocations, despite claims to the
2944       contrary.  Hence this test. */
2945    /* debugBelch("strtab size %d\n", * (UInt32*)strtab); */
2946 #if 0
2947    if ( (*(UInt32*)strtab) > 600000 ) {
2948       /* Note that 600k has no special significance other than being
2949          big enough to handle the almost-2MB-sized lumps that
2950          constitute HSwin32*.o. */
2951       debugBelch("PEi386 object has suspiciously large string table; > 64k relocs?");
2952       return 0;
2953    }
2954 #endif
2955
2956    /* No further verification after this point; only debug printing. */
2957    i = 0;
2958    IF_DEBUG(linker, i=1);
2959    if (i == 0) return 1;
2960
2961    debugBelch( "sectab offset = %d\n", ((UChar*)sectab) - ((UChar*)hdr) );
2962    debugBelch( "symtab offset = %d\n", ((UChar*)symtab) - ((UChar*)hdr) );
2963    debugBelch( "strtab offset = %d\n", ((UChar*)strtab) - ((UChar*)hdr) );
2964
2965    debugBelch("\n" );
2966    debugBelch( "Machine:           0x%x\n", (UInt32)(hdr->Machine) );
2967    debugBelch( "# sections:        %d\n",   (UInt32)(hdr->NumberOfSections) );
2968    debugBelch( "time/date:         0x%x\n", (UInt32)(hdr->TimeDateStamp) );
2969    debugBelch( "symtab offset:     %d\n",   (UInt32)(hdr->PointerToSymbolTable) );
2970    debugBelch( "# symbols:         %d\n",   (UInt32)(hdr->NumberOfSymbols) );
2971    debugBelch( "sz of opt hdr:     %d\n",   (UInt32)(hdr->SizeOfOptionalHeader) );
2972    debugBelch( "characteristics:   0x%x\n", (UInt32)(hdr->Characteristics) );
2973
2974    /* Print the section table. */
2975    debugBelch("\n" );
2976    for (i = 0; i < hdr->NumberOfSections; i++) {
2977       COFF_reloc* reltab;
2978       COFF_section* sectab_i
2979          = (COFF_section*)
2980            myindex ( sizeof_COFF_section, sectab, i );
2981       debugBelch(
2982                 "\n"
2983                 "section %d\n"
2984                 "     name `",
2985                 i
2986               );
2987       printName ( sectab_i->Name, strtab );
2988       debugBelch(
2989                 "'\n"
2990                 "    vsize %d\n"
2991                 "    vaddr %d\n"
2992                 "  data sz %d\n"
2993                 " data off %d\n"
2994                 "  num rel %d\n"
2995                 "  off rel %d\n"
2996                 "  ptr raw 0x%x\n",
2997                 sectab_i->VirtualSize,
2998                 sectab_i->VirtualAddress,
2999                 sectab_i->SizeOfRawData,
3000                 sectab_i->PointerToRawData,
3001                 sectab_i->NumberOfRelocations,
3002                 sectab_i->PointerToRelocations,
3003                 sectab_i->PointerToRawData
3004               );
3005       reltab = (COFF_reloc*) (
3006                   ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
3007                );
3008
3009       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
3010         /* If the relocation field (a short) has overflowed, the
3011          * real count can be found in the first reloc entry.
3012          *
3013          * See Section 4.1 (last para) of the PE spec (rev6.0).
3014          */
3015         COFF_reloc* rel = (COFF_reloc*)
3016                            myindex ( sizeof_COFF_reloc, reltab, 0 );
3017         noRelocs = rel->VirtualAddress;
3018         j = 1;
3019       } else {
3020         noRelocs = sectab_i->NumberOfRelocations;
3021         j = 0;
3022       }
3023
3024       for (; j < noRelocs; j++) {
3025          COFF_symbol* sym;
3026          COFF_reloc* rel = (COFF_reloc*)
3027                            myindex ( sizeof_COFF_reloc, reltab, j );
3028          debugBelch(
3029                    "        type 0x%-4x   vaddr 0x%-8x   name `",
3030                    (UInt32)rel->Type,
3031                    rel->VirtualAddress );
3032          sym = (COFF_symbol*)
3033                myindex ( sizeof_COFF_symbol, symtab, rel->SymbolTableIndex );
3034          /* Hmm..mysterious looking offset - what's it for? SOF */
3035          printName ( sym->Name, strtab -10 );
3036          debugBelch("'\n" );
3037       }
3038
3039       debugBelch("\n" );
3040    }
3041    debugBelch("\n" );
3042    debugBelch("string table has size 0x%x\n", * (UInt32*)strtab );
3043    debugBelch("---START of string table---\n");
3044    for (i = 4; i < *(Int32*)strtab; i++) {
3045       if (strtab[i] == 0)
3046          debugBelch("\n"); else
3047          debugBelch("%c", strtab[i] );
3048    }
3049    debugBelch("--- END  of string table---\n");
3050
3051    debugBelch("\n" );
3052    i = 0;
3053    while (1) {
3054       COFF_symbol* symtab_i;
3055       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
3056       symtab_i = (COFF_symbol*)
3057                  myindex ( sizeof_COFF_symbol, symtab, i );
3058       debugBelch(
3059                 "symbol %d\n"
3060                 "     name `",
3061                 i
3062               );
3063       printName ( symtab_i->Name, strtab );
3064       debugBelch(
3065                 "'\n"
3066                 "    value 0x%x\n"
3067                 "   1+sec# %d\n"
3068                 "     type 0x%x\n"
3069                 "   sclass 0x%x\n"
3070                 "     nAux %d\n",
3071                 symtab_i->Value,
3072                 (Int32)(symtab_i->SectionNumber),
3073                 (UInt32)symtab_i->Type,
3074                 (UInt32)symtab_i->StorageClass,
3075                 (UInt32)symtab_i->NumberOfAuxSymbols
3076               );
3077       i += symtab_i->NumberOfAuxSymbols;
3078       i++;
3079    }
3080
3081    debugBelch("\n" );
3082    return 1;
3083 }
3084
3085
3086 static int
3087 ocGetNames_PEi386 ( ObjectCode* oc )
3088 {
3089    COFF_header*  hdr;
3090    COFF_section* sectab;
3091    COFF_symbol*  symtab;
3092    UChar*        strtab;
3093
3094    UChar* sname;
3095    void*  addr;
3096    int    i;
3097
3098    hdr = (COFF_header*)(oc->image);
3099    sectab = (COFF_section*) (
3100                ((UChar*)(oc->image))
3101                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
3102             );
3103    symtab = (COFF_symbol*) (
3104                ((UChar*)(oc->image))
3105                + hdr->PointerToSymbolTable
3106             );
3107    strtab = ((UChar*)(oc->image))
3108             + hdr->PointerToSymbolTable
3109             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
3110
3111    /* Allocate space for any (local, anonymous) .bss sections. */
3112
3113    for (i = 0; i < hdr->NumberOfSections; i++) {
3114       UInt32 bss_sz;
3115       UChar* zspace;
3116       COFF_section* sectab_i
3117          = (COFF_section*)
3118            myindex ( sizeof_COFF_section, sectab, i );
3119
3120       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3121
3122       if (0 != strcmp(secname, ".bss")) {
3123           stgFree(secname);
3124           continue;
3125       }
3126
3127       stgFree(secname);
3128
3129       /* sof 10/05: the PE spec text isn't too clear regarding what
3130        * the SizeOfRawData field is supposed to hold for object
3131        * file sections containing just uninitialized data -- for executables,
3132        * it is supposed to be zero; unclear what it's supposed to be
3133        * for object files. However, VirtualSize is guaranteed to be
3134        * zero for object files, which definitely suggests that SizeOfRawData
3135        * will be non-zero (where else would the size of this .bss section be
3136        * stored?) Looking at the COFF_section info for incoming object files,
3137        * this certainly appears to be the case.
3138        *
3139        * => I suspect we've been incorrectly handling .bss sections in (relocatable)
3140        * object files up until now. This turned out to bite us with ghc-6.4.1's use
3141        * of gcc-3.4.x, which has started to emit initially-zeroed-out local 'static'
3142        * variable decls into to the .bss section. (The specific function in Q which
3143        * triggered this is libraries/base/cbits/dirUtils.c:__hscore_getFolderPath())
3144        */
3145       if (sectab_i->VirtualSize == 0 && sectab_i->SizeOfRawData == 0) continue;
3146       /* This is a non-empty .bss section.  Allocate zeroed space for
3147          it, and set its PointerToRawData field such that oc->image +
3148          PointerToRawData == addr_of_zeroed_space.  */
3149       bss_sz = sectab_i->VirtualSize;
3150       if ( bss_sz < sectab_i->SizeOfRawData) { bss_sz = sectab_i->SizeOfRawData; }
3151       zspace = stgCallocBytes(1, bss_sz, "ocGetNames_PEi386(anonymous bss)");
3152       sectab_i->PointerToRawData = ((UChar*)zspace) - ((UChar*)(oc->image));
3153       addProddableBlock(oc, zspace, bss_sz);
3154       /* debugBelch("BSS anon section at 0x%x\n", zspace); */
3155    }
3156
3157    /* Copy section information into the ObjectCode. */
3158
3159    for (i = 0; i < hdr->NumberOfSections; i++) {
3160       UChar* start;
3161       UChar* end;
3162       UInt32 sz;
3163
3164       SectionKind kind
3165          = SECTIONKIND_OTHER;
3166       COFF_section* sectab_i
3167          = (COFF_section*)
3168            myindex ( sizeof_COFF_section, sectab, i );
3169
3170       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3171
3172       IF_DEBUG(linker, debugBelch("section name = %s\n", secname ));
3173
3174 #     if 0
3175       /* I'm sure this is the Right Way to do it.  However, the
3176          alternative of testing the sectab_i->Name field seems to
3177          work ok with Cygwin.
3178       */
3179       if (sectab_i->Characteristics & MYIMAGE_SCN_CNT_CODE ||
3180           sectab_i->Characteristics & MYIMAGE_SCN_CNT_INITIALIZED_DATA)
3181          kind = SECTIONKIND_CODE_OR_RODATA;
3182 #     endif
3183
3184       if (0==strcmp(".text",(char*)secname) ||
3185           0==strcmp(".rdata",(char*)secname)||
3186           0==strcmp(".rodata",(char*)secname))
3187          kind = SECTIONKIND_CODE_OR_RODATA;
3188       if (0==strcmp(".data",(char*)secname) ||
3189           0==strcmp(".bss",(char*)secname))
3190          kind = SECTIONKIND_RWDATA;
3191
3192       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
3193       sz = sectab_i->SizeOfRawData;
3194       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
3195
3196       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
3197       end   = start + sz - 1;
3198
3199       if (kind == SECTIONKIND_OTHER
3200           /* Ignore sections called which contain stabs debugging
3201              information. */
3202           && 0 != strcmp(".stab", (char*)secname)
3203           && 0 != strcmp(".stabstr", (char*)secname)
3204           /* ignore constructor section for now */
3205           && 0 != strcmp(".ctors", (char*)secname)
3206           /* ignore section generated from .ident */
3207           && 0!= strncmp(".debug", (char*)secname, 6)
3208           /* ignore unknown section that appeared in gcc 3.4.5(?) */
3209           && 0!= strcmp(".reloc", (char*)secname)
3210           && 0 != strcmp(".rdata$zzz", (char*)secname)
3211          ) {
3212          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", secname, oc->fileName);
3213          stgFree(secname);
3214          return 0;
3215       }
3216
3217       if (kind != SECTIONKIND_OTHER && end >= start) {
3218          addSection(oc, kind, start, end);
3219          addProddableBlock(oc, start, end - start + 1);
3220       }
3221
3222       stgFree(secname);
3223    }
3224
3225    /* Copy exported symbols into the ObjectCode. */
3226
3227    oc->n_symbols = hdr->NumberOfSymbols;
3228    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
3229                                   "ocGetNames_PEi386(oc->symbols)");
3230    /* Call me paranoid; I don't care. */
3231    for (i = 0; i < oc->n_symbols; i++)
3232       oc->symbols[i] = NULL;
3233
3234    i = 0;
3235    while (1) {
3236       COFF_symbol* symtab_i;
3237       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
3238       symtab_i = (COFF_symbol*)
3239                  myindex ( sizeof_COFF_symbol, symtab, i );
3240
3241       addr  = NULL;
3242
3243       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
3244           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
3245          /* This symbol is global and defined, viz, exported */
3246          /* for MYIMAGE_SYMCLASS_EXTERNAL
3247                 && !MYIMAGE_SYM_UNDEFINED,
3248             the address of the symbol is:
3249                 address of relevant section + offset in section
3250          */
3251          COFF_section* sectabent
3252             = (COFF_section*) myindex ( sizeof_COFF_section,
3253                                         sectab,
3254                                         symtab_i->SectionNumber-1 );
3255          addr = ((UChar*)(oc->image))
3256                 + (sectabent->PointerToRawData
3257                    + symtab_i->Value);
3258       }
3259       else
3260       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
3261           && symtab_i->Value > 0) {
3262          /* This symbol isn't in any section at all, ie, global bss.
3263             Allocate zeroed space for it. */
3264          addr = stgCallocBytes(1, symtab_i->Value,
3265                                "ocGetNames_PEi386(non-anonymous bss)");
3266          addSection(oc, SECTIONKIND_RWDATA, addr,
3267                         ((UChar*)addr) + symtab_i->Value - 1);
3268          addProddableBlock(oc, addr, symtab_i->Value);
3269          /* debugBelch("BSS      section at 0x%x\n", addr); */
3270       }
3271
3272       if (addr != NULL ) {
3273          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
3274          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
3275          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
3276          ASSERT(i >= 0 && i < oc->n_symbols);
3277          /* cstring_from_COFF_symbol_name always succeeds. */
3278          oc->symbols[i] = (char*)sname;
3279          ghciInsertStrHashTable(oc->fileName, symhash, (char*)sname, addr);
3280       } else {
3281 #        if 0
3282          debugBelch(
3283                    "IGNORING symbol %d\n"
3284                    "     name `",
3285                    i
3286                  );
3287          printName ( symtab_i->Name, strtab );
3288          debugBelch(
3289                    "'\n"
3290                    "    value 0x%x\n"
3291                    "   1+sec# %d\n"
3292                    "     type 0x%x\n"
3293                    "   sclass 0x%x\n"
3294                    "     nAux %d\n",
3295                    symtab_i->Value,
3296                    (Int32)(symtab_i->SectionNumber),
3297                    (UInt32)symtab_i->Type,
3298                    (UInt32)symtab_i->StorageClass,
3299                    (UInt32)symtab_i->NumberOfAuxSymbols
3300                  );
3301 #        endif
3302       }
3303
3304       i += symtab_i->NumberOfAuxSymbols;
3305       i++;
3306    }
3307
3308    return 1;
3309 }
3310
3311
3312 static int
3313 ocResolve_PEi386 ( ObjectCode* oc )
3314 {
3315    COFF_header*  hdr;
3316    COFF_section* sectab;
3317    COFF_symbol*  symtab;
3318    UChar*        strtab;
3319
3320    UInt32        A;
3321    UInt32        S;
3322    UInt32*       pP;
3323
3324    int i;
3325    UInt32 j, noRelocs;
3326
3327    /* ToDo: should be variable-sized?  But is at least safe in the
3328       sense of buffer-overrun-proof. */
3329    UChar symbol[1000];
3330    /* debugBelch("resolving for %s\n", oc->fileName); */
3331
3332    hdr = (COFF_header*)(oc->image);
3333    sectab = (COFF_section*) (
3334                ((UChar*)(oc->image))
3335                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
3336             );
3337    symtab = (COFF_symbol*) (
3338                ((UChar*)(oc->image))
3339                + hdr->PointerToSymbolTable
3340             );
3341    strtab = ((UChar*)(oc->image))
3342             + hdr->PointerToSymbolTable
3343             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
3344
3345    for (i = 0; i < hdr->NumberOfSections; i++) {
3346       COFF_section* sectab_i
3347          = (COFF_section*)
3348            myindex ( sizeof_COFF_section, sectab, i );
3349       COFF_reloc* reltab
3350          = (COFF_reloc*) (
3351               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
3352            );
3353
3354       char *secname = cstring_from_section_name(sectab_i->Name, strtab);
3355
3356       /* Ignore sections called which contain stabs debugging
3357          information. */
3358       if (0 == strcmp(".stab", (char*)secname)
3359           || 0 == strcmp(".stabstr", (char*)secname)
3360           || 0 == strcmp(".ctors", (char*)secname)
3361           || 0 == strncmp(".debug", (char*)secname, 6)
3362           || 0 == strcmp(".rdata$zzz", (char*)secname)) {
3363           stgFree(secname);
3364           continue;
3365       }
3366
3367       stgFree(secname);
3368
3369       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
3370         /* If the relocation field (a short) has overflowed, the
3371          * real count can be found in the first reloc entry.
3372          *
3373          * See Section 4.1 (last para) of the PE spec (rev6.0).
3374          *
3375          * Nov2003 update: the GNU linker still doesn't correctly
3376          * handle the generation of relocatable object files with
3377          * overflown relocations. Hence the output to warn of potential
3378          * troubles.
3379          */
3380         COFF_reloc* rel = (COFF_reloc*)
3381                            myindex ( sizeof_COFF_reloc, reltab, 0 );
3382         noRelocs = rel->VirtualAddress;
3383
3384         /* 10/05: we now assume (and check for) a GNU ld that is capable
3385          * of handling object files with (>2^16) of relocs.
3386          */
3387 #if 0
3388         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
3389                    noRelocs);
3390 #endif
3391         j = 1;
3392       } else {
3393         noRelocs = sectab_i->NumberOfRelocations;
3394         j = 0;
3395       }
3396
3397
3398       for (; j < noRelocs; j++) {
3399          COFF_symbol* sym;
3400          COFF_reloc* reltab_j
3401             = (COFF_reloc*)
3402               myindex ( sizeof_COFF_reloc, reltab, j );
3403
3404          /* the location to patch */
3405          pP = (UInt32*)(
3406                  ((UChar*)(oc->image))
3407                  + (sectab_i->PointerToRawData
3408                     + reltab_j->VirtualAddress
3409                     - sectab_i->VirtualAddress )
3410               );
3411          /* the existing contents of pP */
3412          A = *pP;
3413          /* the symbol to connect to */
3414          sym = (COFF_symbol*)
3415                myindex ( sizeof_COFF_symbol,
3416                          symtab, reltab_j->SymbolTableIndex );
3417          IF_DEBUG(linker,
3418                   debugBelch(
3419                             "reloc sec %2d num %3d:  type 0x%-4x   "
3420                             "vaddr 0x%-8x   name `",
3421                             i, j,
3422                             (UInt32)reltab_j->Type,
3423                             reltab_j->VirtualAddress );
3424                             printName ( sym->Name, strtab );
3425                             debugBelch("'\n" ));
3426
3427          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
3428             COFF_section* section_sym
3429                = findPEi386SectionCalled ( oc, sym->Name );
3430             if (!section_sym) {
3431                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
3432                return 0;
3433             }
3434             S = ((UInt32)(oc->image))
3435                 + (section_sym->PointerToRawData
3436                    + sym->Value);
3437          } else {
3438             copyName ( sym->Name, strtab, symbol, 1000-1 );
3439             S = (UInt32) lookupSymbol( (char*)symbol );
3440             if ((void*)S != NULL) goto foundit;
3441             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3442             return 0;
3443            foundit:;
3444          }
3445          checkProddableBlock(oc, pP);
3446          switch (reltab_j->Type) {
3447             case MYIMAGE_REL_I386_DIR32:
3448                *pP = A + S;
3449                break;
3450             case MYIMAGE_REL_I386_REL32:
3451                /* Tricky.  We have to insert a displacement at
3452                   pP which, when added to the PC for the _next_
3453                   insn, gives the address of the target (S).
3454                   Problem is to know the address of the next insn
3455                   when we only know pP.  We assume that this
3456                   literal field is always the last in the insn,
3457                   so that the address of the next insn is pP+4
3458                   -- hence the constant 4.
3459                   Also I don't know if A should be added, but so
3460                   far it has always been zero.
3461
3462                   SOF 05/2005: 'A' (old contents of *pP) have been observed
3463                   to contain values other than zero (the 'wx' object file
3464                   that came with wxhaskell-0.9.4; dunno how it was compiled..).
3465                   So, add displacement to old value instead of asserting
3466                   A to be zero. Fixes wxhaskell-related crashes, and no other
3467                   ill effects have been observed.
3468
3469                   Update: the reason why we're seeing these more elaborate
3470                   relocations is due to a switch in how the NCG compiles SRTs
3471                   and offsets to them from info tables. SRTs live in .(ro)data,
3472                   while info tables live in .text, causing GAS to emit REL32/DISP32
3473                   relocations with non-zero values. Adding the displacement is
3474                   the right thing to do.
3475                */
3476                *pP = S - ((UInt32)pP) - 4 + A;
3477                break;
3478             default:
3479                debugBelch("%s: unhandled PEi386 relocation type %d",
3480                      oc->fileName, reltab_j->Type);
3481                return 0;
3482          }
3483
3484       }
3485    }
3486
3487    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
3488    return 1;
3489 }
3490
3491 #endif /* defined(OBJFORMAT_PEi386) */
3492
3493
3494 /* --------------------------------------------------------------------------
3495  * ELF specifics
3496  * ------------------------------------------------------------------------*/
3497
3498 #if defined(OBJFORMAT_ELF)
3499
3500 #define FALSE 0
3501 #define TRUE  1
3502
3503 #if defined(sparc_HOST_ARCH)
3504 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
3505 #elif defined(i386_HOST_ARCH)
3506 #  define ELF_TARGET_386    /* Used inside <elf.h> */
3507 #elif defined(x86_64_HOST_ARCH)
3508 #  define ELF_TARGET_X64_64
3509 #  define ELF_64BIT
3510 #endif
3511
3512 #if !defined(openbsd_HOST_OS)
3513 #  include <elf.h>
3514 #else
3515 /* openbsd elf has things in different places, with diff names */
3516 #  include <elf_abi.h>
3517 #  include <machine/reloc.h>
3518 #  define R_386_32    RELOC_32
3519 #  define R_386_PC32  RELOC_PC32
3520 #endif
3521
3522 /* If elf.h doesn't define it */
3523 #  ifndef R_X86_64_PC64
3524 #    define R_X86_64_PC64 24
3525 #  endif
3526
3527 /*
3528  * Define a set of types which can be used for both ELF32 and ELF64
3529  */
3530
3531 #ifdef ELF_64BIT
3532 #define ELFCLASS    ELFCLASS64
3533 #define Elf_Addr    Elf64_Addr
3534 #define Elf_Word    Elf64_Word
3535 #define Elf_Sword   Elf64_Sword
3536 #define Elf_Ehdr    Elf64_Ehdr
3537 #define Elf_Phdr    Elf64_Phdr
3538 #define Elf_Shdr    Elf64_Shdr
3539 #define Elf_Sym     Elf64_Sym
3540 #define Elf_Rel     Elf64_Rel
3541 #define Elf_Rela    Elf64_Rela
3542 #ifndef ELF_ST_TYPE
3543 #define ELF_ST_TYPE ELF64_ST_TYPE
3544 #endif
3545 #ifndef ELF_ST_BIND
3546 #define ELF_ST_BIND ELF64_ST_BIND
3547 #endif
3548 #ifndef ELF_R_TYPE
3549 #define ELF_R_TYPE  ELF64_R_TYPE
3550 #endif
3551 #ifndef ELF_R_SYM
3552 #define ELF_R_SYM   ELF64_R_SYM
3553 #endif
3554 #else
3555 #define ELFCLASS    ELFCLASS32
3556 #define Elf_Addr    Elf32_Addr
3557 #define Elf_Word    Elf32_Word
3558 #define Elf_Sword   Elf32_Sword
3559 #define Elf_Ehdr    Elf32_Ehdr
3560 #define Elf_Phdr    Elf32_Phdr
3561 #define Elf_Shdr    Elf32_Shdr
3562 #define Elf_Sym     Elf32_Sym
3563 #define Elf_Rel     Elf32_Rel
3564 #define Elf_Rela    Elf32_Rela
3565 #ifndef ELF_ST_TYPE
3566 #define ELF_ST_TYPE ELF32_ST_TYPE
3567 #endif
3568 #ifndef ELF_ST_BIND
3569 #define ELF_ST_BIND ELF32_ST_BIND
3570 #endif
3571 #ifndef ELF_R_TYPE
3572 #define ELF_R_TYPE  ELF32_R_TYPE
3573 #endif
3574 #ifndef ELF_R_SYM
3575 #define ELF_R_SYM   ELF32_R_SYM
3576 #endif
3577 #endif
3578
3579
3580 /*
3581  * Functions to allocate entries in dynamic sections.  Currently we simply
3582  * preallocate a large number, and we don't check if a entry for the given
3583  * target already exists (a linear search is too slow).  Ideally these
3584  * entries would be associated with symbols.
3585  */
3586
3587 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
3588 #define GOT_SIZE            0x20000
3589 #define FUNCTION_TABLE_SIZE 0x10000
3590 #define PLT_SIZE            0x08000
3591
3592 #ifdef ELF_NEED_GOT
3593 static Elf_Addr got[GOT_SIZE];
3594 static unsigned int gotIndex;
3595 static Elf_Addr gp_val = (Elf_Addr)got;
3596
3597 static Elf_Addr
3598 allocateGOTEntry(Elf_Addr target)
3599 {
3600    Elf_Addr *entry;
3601
3602    if (gotIndex >= GOT_SIZE)
3603       barf("Global offset table overflow");
3604
3605    entry = &got[gotIndex++];
3606    *entry = target;
3607    return (Elf_Addr)entry;
3608 }
3609 #endif
3610
3611 #ifdef ELF_FUNCTION_DESC
3612 typedef struct {
3613    Elf_Addr ip;
3614    Elf_Addr gp;
3615 } FunctionDesc;
3616
3617 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
3618 static unsigned int functionTableIndex;
3619
3620 static Elf_Addr
3621 allocateFunctionDesc(Elf_Addr target)
3622 {
3623    FunctionDesc *entry;
3624
3625    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
3626       barf("Function table overflow");
3627
3628    entry = &functionTable[functionTableIndex++];
3629    entry->ip = target;
3630    entry->gp = (Elf_Addr)gp_val;
3631    return (Elf_Addr)entry;
3632 }
3633
3634 static Elf_Addr
3635 copyFunctionDesc(Elf_Addr target)
3636 {
3637    FunctionDesc *olddesc = (FunctionDesc *)target;
3638    FunctionDesc *newdesc;
3639
3640    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
3641    newdesc->gp = olddesc->gp;
3642    return (Elf_Addr)newdesc;
3643 }
3644 #endif
3645
3646 #ifdef ELF_NEED_PLT
3647
3648 typedef struct {
3649    unsigned char code[sizeof(plt_code)];
3650 } PLTEntry;
3651
3652 static Elf_Addr
3653 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
3654 {
3655    PLTEntry *plt = (PLTEntry *)oc->plt;
3656    PLTEntry *entry;
3657
3658    if (oc->pltIndex >= PLT_SIZE)
3659       barf("Procedure table overflow");
3660
3661    entry = &plt[oc->pltIndex++];
3662    memcpy(entry->code, plt_code, sizeof(entry->code));
3663    PLT_RELOC(entry->code, target);
3664    return (Elf_Addr)entry;
3665 }
3666
3667 static unsigned int
3668 PLTSize(void)
3669 {
3670    return (PLT_SIZE * sizeof(PLTEntry));
3671 }
3672 #endif
3673
3674
3675 /*
3676  * Generic ELF functions
3677  */
3678
3679 static char *
3680 findElfSection ( void* objImage, Elf_Word sh_type )
3681 {
3682    char* ehdrC = (char*)objImage;
3683    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
3684    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
3685    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3686    char* ptr = NULL;
3687    int i;
3688
3689    for (i = 0; i < ehdr->e_shnum; i++) {
3690       if (shdr[i].sh_type == sh_type
3691           /* Ignore the section header's string table. */
3692           && i != ehdr->e_shstrndx
3693           /* Ignore string tables named .stabstr, as they contain
3694              debugging info. */
3695           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3696          ) {
3697          ptr = ehdrC + shdr[i].sh_offset;
3698          break;
3699       }
3700    }
3701    return ptr;
3702 }
3703
3704 static int
3705 ocVerifyImage_ELF ( ObjectCode* oc )
3706 {
3707    Elf_Shdr* shdr;
3708    Elf_Sym*  stab;
3709    int i, j, nent, nstrtab, nsymtabs;
3710    char* sh_strtab;
3711    char* strtab;
3712
3713    char*     ehdrC = (char*)(oc->image);
3714    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
3715
3716    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
3717        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
3718        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
3719        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
3720       errorBelch("%s: not an ELF object", oc->fileName);
3721       return 0;
3722    }
3723
3724    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
3725       errorBelch("%s: unsupported ELF format", oc->fileName);
3726       return 0;
3727    }
3728
3729    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
3730        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
3731    } else
3732    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
3733        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
3734    } else {
3735        errorBelch("%s: unknown endiannness", oc->fileName);
3736        return 0;
3737    }
3738
3739    if (ehdr->e_type != ET_REL) {
3740       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
3741       return 0;
3742    }
3743    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
3744
3745    IF_DEBUG(linker,debugBelch( "Architecture is " ));
3746    switch (ehdr->e_machine) {
3747       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
3748 #ifdef EM_SPARC32PLUS
3749       case EM_SPARC32PLUS:
3750 #endif
3751       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
3752 #ifdef EM_IA_64
3753       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
3754 #endif
3755       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
3756 #ifdef EM_X86_64
3757       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
3758 #elif defined(EM_AMD64)
3759       case EM_AMD64: IF_DEBUG(linker,debugBelch( "amd64" )); break;
3760 #endif
3761       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
3762                      errorBelch("%s: unknown architecture (e_machine == %d)"
3763                                 , oc->fileName, ehdr->e_machine);
3764                      return 0;
3765    }
3766
3767    IF_DEBUG(linker,debugBelch(
3768              "\nSection header table: start %ld, n_entries %d, ent_size %d\n",
3769              (long)ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
3770
3771    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
3772
3773    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3774
3775    if (ehdr->e_shstrndx == SHN_UNDEF) {
3776       errorBelch("%s: no section header string table", oc->fileName);
3777       return 0;
3778    } else {
3779       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
3780                           ehdr->e_shstrndx));
3781       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3782    }
3783
3784    for (i = 0; i < ehdr->e_shnum; i++) {
3785       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
3786       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
3787       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
3788       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
3789       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
3790                ehdrC + shdr[i].sh_offset,
3791                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
3792
3793       if (shdr[i].sh_type == SHT_REL) {
3794           IF_DEBUG(linker,debugBelch("Rel  " ));
3795       } else if (shdr[i].sh_type == SHT_RELA) {
3796           IF_DEBUG(linker,debugBelch("RelA " ));
3797       } else {
3798           IF_DEBUG(linker,debugBelch("     "));
3799       }
3800       if (sh_strtab) {
3801           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
3802       }
3803    }
3804
3805    IF_DEBUG(linker,debugBelch( "\nString tables" ));
3806    strtab = NULL;
3807    nstrtab = 0;
3808    for (i = 0; i < ehdr->e_shnum; i++) {
3809       if (shdr[i].sh_type == SHT_STRTAB
3810           /* Ignore the section header's string table. */
3811           && i != ehdr->e_shstrndx
3812           /* Ignore string tables named .stabstr, as they contain
3813              debugging info. */
3814           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3815          ) {
3816          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
3817          strtab = ehdrC + shdr[i].sh_offset;
3818          nstrtab++;
3819       }
3820    }
3821    if (nstrtab != 1) {
3822       errorBelch("%s: no string tables, or too many", oc->fileName);
3823       return 0;
3824    }
3825
3826    nsymtabs = 0;
3827    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
3828    for (i = 0; i < ehdr->e_shnum; i++) {
3829       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3830       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
3831       nsymtabs++;
3832       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3833       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3834       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%ld rem)\n",
3835                nent,
3836                (long)shdr[i].sh_size % sizeof(Elf_Sym)
3837              ));
3838       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
3839          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
3840          return 0;
3841       }
3842       for (j = 0; j < nent; j++) {
3843          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
3844          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
3845                              (int)stab[j].st_shndx,
3846                              (int)stab[j].st_size,
3847                              (char*)stab[j].st_value ));
3848
3849          IF_DEBUG(linker,debugBelch("type=" ));
3850          switch (ELF_ST_TYPE(stab[j].st_info)) {
3851             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
3852             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
3853             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
3854             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
3855             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
3856             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
3857          }
3858          IF_DEBUG(linker,debugBelch("  " ));
3859
3860          IF_DEBUG(linker,debugBelch("bind=" ));
3861          switch (ELF_ST_BIND(stab[j].st_info)) {
3862             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
3863             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
3864             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
3865             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
3866          }
3867          IF_DEBUG(linker,debugBelch("  " ));
3868
3869          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
3870       }
3871    }
3872
3873    if (nsymtabs == 0) {
3874       errorBelch("%s: didn't find any symbol tables", oc->fileName);
3875       return 0;
3876    }
3877
3878    return 1;
3879 }
3880
3881 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
3882 {
3883     *is_bss = FALSE;
3884
3885     if (hdr->sh_type == SHT_PROGBITS
3886         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
3887         /* .text-style section */
3888         return SECTIONKIND_CODE_OR_RODATA;
3889     }
3890
3891     if (hdr->sh_type == SHT_PROGBITS
3892             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3893             /* .data-style section */
3894             return SECTIONKIND_RWDATA;
3895     }
3896
3897     if (hdr->sh_type == SHT_PROGBITS
3898         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
3899         /* .rodata-style section */
3900         return SECTIONKIND_CODE_OR_RODATA;
3901     }
3902
3903     if (hdr->sh_type == SHT_NOBITS
3904         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3905         /* .bss-style section */
3906         *is_bss = TRUE;
3907         return SECTIONKIND_RWDATA;
3908     }
3909
3910     return SECTIONKIND_OTHER;
3911 }
3912
3913
3914 static int
3915 ocGetNames_ELF ( ObjectCode* oc )
3916 {
3917    int i, j, k, nent;
3918    Elf_Sym* stab;
3919
3920    char*     ehdrC    = (char*)(oc->image);
3921    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
3922    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
3923    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3924
3925    ASSERT(symhash != NULL);
3926
3927    if (!strtab) {
3928       errorBelch("%s: no strtab", oc->fileName);
3929       return 0;
3930    }
3931
3932    k = 0;
3933    for (i = 0; i < ehdr->e_shnum; i++) {
3934       /* Figure out what kind of section it is.  Logic derived from
3935          Figure 1.14 ("Special Sections") of the ELF document
3936          ("Portable Formats Specification, Version 1.1"). */
3937       int         is_bss = FALSE;
3938       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
3939
3940       if (is_bss && shdr[i].sh_size > 0) {
3941          /* This is a non-empty .bss section.  Allocate zeroed space for
3942             it, and set its .sh_offset field such that
3943             ehdrC + .sh_offset == addr_of_zeroed_space.  */
3944          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
3945                                        "ocGetNames_ELF(BSS)");
3946          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
3947          /*
3948          debugBelch("BSS section at 0x%x, size %d\n",
3949                          zspace, shdr[i].sh_size);
3950          */
3951       }
3952
3953       /* fill in the section info */
3954       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
3955          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
3956          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
3957                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
3958       }
3959
3960       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3961
3962       /* copy stuff into this module's object symbol table */
3963       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3964       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3965
3966       oc->n_symbols = nent;
3967       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3968                                    "ocGetNames_ELF(oc->symbols)");
3969
3970       for (j = 0; j < nent; j++) {
3971
3972          char  isLocal = FALSE; /* avoids uninit-var warning */
3973          char* ad      = NULL;
3974          char* nm      = strtab + stab[j].st_name;
3975          int   secno   = stab[j].st_shndx;
3976
3977          /* Figure out if we want to add it; if so, set ad to its
3978             address.  Otherwise leave ad == NULL. */
3979
3980          if (secno == SHN_COMMON) {
3981             isLocal = FALSE;
3982             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
3983             /*
3984             debugBelch("COMMON symbol, size %d name %s\n",
3985                             stab[j].st_size, nm);
3986             */
3987             /* Pointless to do addProddableBlock() for this area,
3988                since the linker should never poke around in it. */
3989          }
3990          else
3991          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
3992                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
3993               )
3994               /* and not an undefined symbol */
3995               && stab[j].st_shndx != SHN_UNDEF
3996               /* and not in a "special section" */
3997               && stab[j].st_shndx < SHN_LORESERVE
3998               &&
3999               /* and it's a not a section or string table or anything silly */
4000               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
4001                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
4002                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
4003               )
4004             ) {
4005             /* Section 0 is the undefined section, hence > and not >=. */
4006             ASSERT(secno > 0 && secno < ehdr->e_shnum);
4007             /*
4008             if (shdr[secno].sh_type == SHT_NOBITS) {
4009                debugBelch("   BSS symbol, size %d off %d name %s\n",
4010                                stab[j].st_size, stab[j].st_value, nm);
4011             }
4012             */
4013             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
4014             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
4015                isLocal = TRUE;
4016             } else {
4017 #ifdef ELF_FUNCTION_DESC
4018                /* dlsym() and the initialisation table both give us function
4019                 * descriptors, so to be consistent we store function descriptors
4020                 * in the symbol table */
4021                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
4022                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
4023 #endif
4024                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s\n",
4025                                       ad, oc->fileName, nm ));
4026                isLocal = FALSE;
4027             }
4028          }
4029
4030          /* And the decision is ... */
4031
4032          if (ad != NULL) {
4033             ASSERT(nm != NULL);
4034             oc->symbols[j] = nm;
4035             /* Acquire! */
4036             if (isLocal) {
4037                /* Ignore entirely. */
4038             } else {
4039                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
4040             }
4041          } else {
4042             /* Skip. */
4043             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
4044                                    strtab + stab[j].st_name ));
4045             /*
4046             debugBelch(
4047                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
4048                     (int)ELF_ST_BIND(stab[j].st_info),
4049                     (int)ELF_ST_TYPE(stab[j].st_info),
4050                     (int)stab[j].st_shndx,
4051                     strtab + stab[j].st_name
4052                    );
4053             */
4054             oc->symbols[j] = NULL;
4055          }
4056
4057       }
4058    }
4059
4060    return 1;
4061 }
4062
4063 /* Do ELF relocations which lack an explicit addend.  All x86-linux
4064    relocations appear to be of this form. */
4065 static int
4066 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
4067                          Elf_Shdr* shdr, int shnum,
4068                          Elf_Sym*  stab, char* strtab )
4069 {
4070    int j;
4071    char *symbol;
4072    Elf_Word* targ;
4073    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
4074    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
4075    int target_shndx = shdr[shnum].sh_info;
4076    int symtab_shndx = shdr[shnum].sh_link;
4077
4078    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
4079    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
4080    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
4081                           target_shndx, symtab_shndx ));
4082
4083    /* Skip sections that we're not interested in. */
4084    {
4085        int is_bss;
4086        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
4087        if (kind == SECTIONKIND_OTHER) {
4088            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
4089            return 1;
4090        }
4091    }
4092
4093    for (j = 0; j < nent; j++) {
4094       Elf_Addr offset = rtab[j].r_offset;
4095       Elf_Addr info   = rtab[j].r_info;
4096
4097       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
4098       Elf_Word* pP = (Elf_Word*)P;
4099       Elf_Addr  A  = *pP;
4100       Elf_Addr  S;
4101       void*     S_tmp;
4102       Elf_Addr  value;
4103       StgStablePtr stablePtr;
4104       StgPtr stableVal;
4105
4106       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
4107                              j, (void*)offset, (void*)info ));
4108       if (!info) {
4109          IF_DEBUG(linker,debugBelch( " ZERO" ));
4110          S = 0;
4111       } else {
4112          Elf_Sym sym = stab[ELF_R_SYM(info)];
4113          /* First see if it is a local symbol. */
4114          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
4115             /* Yes, so we can get the address directly from the ELF symbol
4116                table. */
4117             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
4118             S = (Elf_Addr)
4119                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
4120                        + stab[ELF_R_SYM(info)].st_value);
4121
4122          } else {
4123             symbol = strtab + sym.st_name;
4124             stablePtr = (StgStablePtr)lookupHashTable(stablehash, (StgWord)symbol);
4125             if (NULL == stablePtr) {
4126               /* No, so look up the name in our global table. */
4127               S_tmp = lookupSymbol( symbol );
4128               S = (Elf_Addr)S_tmp;
4129             } else {
4130               stableVal = deRefStablePtr( stablePtr );
4131               S_tmp = stableVal;
4132               S = (Elf_Addr)S_tmp;
4133             }
4134          }
4135          if (!S) {
4136             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
4137             return 0;
4138          }
4139          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
4140       }
4141
4142       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
4143                              (void*)P, (void*)S, (void*)A ));
4144       checkProddableBlock ( oc, pP );
4145
4146       value = S + A;
4147
4148       switch (ELF_R_TYPE(info)) {
4149 #        ifdef i386_HOST_ARCH
4150          case R_386_32:   *pP = value;     break;
4151          case R_386_PC32: *pP = value - P; break;
4152 #        endif
4153          default:
4154             errorBelch("%s: unhandled ELF relocation(Rel) type %lu\n",
4155                   oc->fileName, (lnat)ELF_R_TYPE(info));
4156             return 0;
4157       }
4158
4159    }
4160    return 1;
4161 }
4162
4163 /* Do ELF relocations for which explicit addends are supplied.
4164    sparc-solaris relocations appear to be of this form. */
4165 static int
4166 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
4167                           Elf_Shdr* shdr, int shnum,
4168                           Elf_Sym*  stab, char* strtab )
4169 {
4170    int j;
4171    char *symbol = NULL;
4172    Elf_Addr targ;
4173    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
4174    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
4175    int target_shndx = shdr[shnum].sh_info;
4176    int symtab_shndx = shdr[shnum].sh_link;
4177
4178    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
4179    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
4180    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
4181                           target_shndx, symtab_shndx ));
4182
4183    for (j = 0; j < nent; j++) {
4184 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
4185       /* This #ifdef only serves to avoid unused-var warnings. */
4186       Elf_Addr  offset = rtab[j].r_offset;
4187       Elf_Addr  P      = targ + offset;
4188 #endif
4189       Elf_Addr  info   = rtab[j].r_info;
4190       Elf_Addr  A      = rtab[j].r_addend;
4191       Elf_Addr  S;
4192       void*     S_tmp;
4193       Elf_Addr  value;
4194 #     if defined(sparc_HOST_ARCH)
4195       Elf_Word* pP = (Elf_Word*)P;
4196       Elf_Word  w1, w2;
4197 #     elif defined(powerpc_HOST_ARCH)
4198       Elf_Sword delta;
4199 #     endif
4200
4201       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
4202                              j, (void*)offset, (void*)info,
4203                                 (void*)A ));
4204       if (!info) {
4205          IF_DEBUG(linker,debugBelch( " ZERO" ));
4206          S = 0;
4207       } else {
4208          Elf_Sym sym = stab[ELF_R_SYM(info)];
4209          /* First see if it is a local symbol. */
4210          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
4211             /* Yes, so we can get the address directly from the ELF symbol
4212                table. */
4213             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
4214             S = (Elf_Addr)
4215                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
4216                        + stab[ELF_R_SYM(info)].st_value);
4217 #ifdef ELF_FUNCTION_DESC
4218             /* Make a function descriptor for this function */
4219             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
4220                S = allocateFunctionDesc(S + A);
4221                A = 0;
4222             }
4223 #endif
4224          } else {
4225             /* No, so look up the name in our global table. */
4226             symbol = strtab + sym.st_name;
4227             S_tmp = lookupSymbol( symbol );
4228             S = (Elf_Addr)S_tmp;
4229
4230 #ifdef ELF_FUNCTION_DESC
4231             /* If a function, already a function descriptor - we would
4232                have to copy it to add an offset. */
4233             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
4234                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
4235 #endif
4236          }
4237          if (!S) {
4238            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
4239            return 0;
4240          }
4241          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
4242       }
4243
4244       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
4245                                         (void*)P, (void*)S, (void*)A ));
4246       /* checkProddableBlock ( oc, (void*)P ); */
4247
4248       value = S + A;
4249
4250       switch (ELF_R_TYPE(info)) {
4251 #        if defined(sparc_HOST_ARCH)
4252          case R_SPARC_WDISP30:
4253             w1 = *pP & 0xC0000000;
4254             w2 = (Elf_Word)((value - P) >> 2);
4255             ASSERT((w2 & 0xC0000000) == 0);
4256             w1 |= w2;
4257             *pP = w1;
4258             break;
4259          case R_SPARC_HI22:
4260             w1 = *pP & 0xFFC00000;
4261             w2 = (Elf_Word)(value >> 10);
4262             ASSERT((w2 & 0xFFC00000) == 0);
4263             w1 |= w2;
4264             *pP = w1;
4265             break;
4266          case R_SPARC_LO10:
4267             w1 = *pP & ~0x3FF;
4268             w2 = (Elf_Word)(value & 0x3FF);
4269             ASSERT((w2 & ~0x3FF) == 0);
4270             w1 |= w2;
4271             *pP = w1;
4272             break;
4273
4274          /* According to the Sun documentation:
4275             R_SPARC_UA32
4276             This relocation type resembles R_SPARC_32, except it refers to an
4277             unaligned word. That is, the word to be relocated must be treated
4278             as four separate bytes with arbitrary alignment, not as a word
4279             aligned according to the architecture requirements.
4280          */
4281          case R_SPARC_UA32:
4282             w2  = (Elf_Word)value;
4283
4284             // SPARC doesn't do misaligned writes of 32 bit words,
4285             //       so we have to do this one byte-at-a-time.
4286             char *pPc   = (char*)pP;
4287             pPc[0]      = (char) ((Elf_Word)(w2 & 0xff000000) >> 24);
4288             pPc[1]      = (char) ((Elf_Word)(w2 & 0x00ff0000) >> 16);
4289             pPc[2]      = (char) ((Elf_Word)(w2 & 0x0000ff00) >> 8);
4290             pPc[3]      = (char) ((Elf_Word)(w2 & 0x000000ff));
4291             break;
4292
4293          case R_SPARC_32:
4294             w2 = (Elf_Word)value;
4295             *pP = w2;
4296             break;
4297 #        elif defined(powerpc_HOST_ARCH)
4298          case R_PPC_ADDR16_LO:
4299             *(Elf32_Half*) P = value;
4300             break;
4301
4302          case R_PPC_ADDR16_HI:
4303             *(Elf32_Half*) P = value >> 16;
4304             break;
4305
4306          case R_PPC_ADDR16_HA:
4307             *(Elf32_Half*) P = (value + 0x8000) >> 16;
4308             break;
4309
4310          case R_PPC_ADDR32:
4311             *(Elf32_Word *) P = value;
4312             break;
4313
4314          case R_PPC_REL32:
4315             *(Elf32_Word *) P = value - P;
4316             break;
4317
4318          case R_PPC_REL24:
4319             delta = value - P;
4320
4321             if( delta << 6 >> 6 != delta )
4322             {
4323                value = (Elf_Addr) (&makeSymbolExtra( oc, ELF_R_SYM(info), value )
4324                                         ->jumpIsland);
4325                delta = value - P;
4326
4327                if( value == 0 || delta << 6 >> 6 != delta )
4328                {
4329                   barf( "Unable to make SymbolExtra for #%d",
4330                         ELF_R_SYM(info) );
4331                   return 0;
4332                }
4333             }
4334
4335             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
4336                                           | (delta & 0x3fffffc);
4337             break;
4338 #        endif
4339
4340 #if x86_64_HOST_ARCH
4341       case R_X86_64_64:
4342           *(Elf64_Xword *)P = value;
4343           break;
4344
4345       case R_X86_64_PC32:
4346       {
4347 #if defined(ALWAYS_PIC)
4348           barf("R_X86_64_PC32 relocation, but ALWAYS_PIC.");
4349 #else
4350           StgInt64 off = value - P;
4351           if (off >= 0x7fffffffL || off < -0x80000000L) {
4352 #if X86_64_ELF_NONPIC_HACK
4353               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4354                                                 -> jumpIsland;
4355               off = pltAddress + A - P;
4356 #else
4357               barf("R_X86_64_PC32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4358                    symbol, off, oc->fileName );
4359 #endif
4360           }
4361           *(Elf64_Word *)P = (Elf64_Word)off;
4362 #endif
4363           break;
4364       }
4365
4366       case R_X86_64_PC64:
4367       {
4368           StgInt64 off = value - P;
4369           *(Elf64_Word *)P = (Elf64_Word)off;
4370           break;
4371       }
4372
4373       case R_X86_64_32:
4374 #if defined(ALWAYS_PIC)
4375           barf("R_X86_64_32 relocation, but ALWAYS_PIC.");
4376 #else
4377           if (value >= 0x7fffffffL) {
4378 #if X86_64_ELF_NONPIC_HACK
4379               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4380                                                 -> jumpIsland;
4381               value = pltAddress + A;
4382 #else
4383               barf("R_X86_64_32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4384                    symbol, value, oc->fileName );
4385 #endif
4386           }
4387           *(Elf64_Word *)P = (Elf64_Word)value;
4388 #endif
4389           break;
4390
4391       case R_X86_64_32S:
4392 #if defined(ALWAYS_PIC)
4393           barf("R_X86_64_32S relocation, but ALWAYS_PIC.");
4394 #else
4395           if ((StgInt64)value > 0x7fffffffL || (StgInt64)value < -0x80000000L) {
4396 #if X86_64_ELF_NONPIC_HACK
4397               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4398                                                 -> jumpIsland;
4399               value = pltAddress + A;
4400 #else
4401               barf("R_X86_64_32S relocation out of range: %s = %p\nRecompile %s with -fPIC.",
4402                    symbol, value, oc->fileName );
4403 #endif
4404           }
4405           *(Elf64_Sword *)P = (Elf64_Sword)value;
4406 #endif
4407           break;
4408
4409       case R_X86_64_GOTPCREL:
4410       {
4411           StgInt64 gotAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)->addr;
4412           StgInt64 off = gotAddress + A - P;
4413           *(Elf64_Word *)P = (Elf64_Word)off;
4414           break;
4415       }
4416
4417       case R_X86_64_PLT32:
4418       {
4419 #if defined(ALWAYS_PIC)
4420           barf("R_X86_64_PLT32 relocation, but ALWAYS_PIC.");
4421 #else
4422           StgInt64 off = value - P;
4423           if (off >= 0x7fffffffL || off < -0x80000000L) {
4424               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
4425                                                     -> jumpIsland;
4426               off = pltAddress + A - P;
4427           }
4428           *(Elf64_Word *)P = (Elf64_Word)off;
4429 #endif
4430           break;
4431       }
4432 #endif
4433
4434          default:
4435             errorBelch("%s: unhandled ELF relocation(RelA) type %lu\n",
4436                   oc->fileName, (lnat)ELF_R_TYPE(info));
4437             return 0;
4438       }
4439
4440    }
4441    return 1;
4442 }
4443
4444 static int
4445 ocResolve_ELF ( ObjectCode* oc )
4446 {
4447    char *strtab;
4448    int   shnum, ok;
4449    Elf_Sym*  stab  = NULL;
4450    char*     ehdrC = (char*)(oc->image);
4451    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
4452    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
4453
4454    /* first find "the" symbol table */
4455    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
4456
4457    /* also go find the string table */
4458    strtab = findElfSection ( ehdrC, SHT_STRTAB );
4459
4460    if (stab == NULL || strtab == NULL) {
4461       errorBelch("%s: can't find string or symbol table", oc->fileName);
4462       return 0;
4463    }
4464
4465    /* Process the relocation sections. */
4466    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
4467       if (shdr[shnum].sh_type == SHT_REL) {
4468          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
4469                                        shnum, stab, strtab );
4470          if (!ok) return ok;
4471       }
4472       else
4473       if (shdr[shnum].sh_type == SHT_RELA) {
4474          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
4475                                         shnum, stab, strtab );
4476          if (!ok) return ok;
4477       }
4478    }
4479
4480 #if defined(powerpc_HOST_ARCH)
4481    ocFlushInstructionCache( oc );
4482 #endif
4483
4484    return 1;
4485 }
4486
4487 /*
4488  * PowerPC & X86_64 ELF specifics
4489  */
4490
4491 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
4492
4493 static int ocAllocateSymbolExtras_ELF( ObjectCode *oc )
4494 {
4495   Elf_Ehdr *ehdr;
4496   Elf_Shdr* shdr;
4497   int i;
4498
4499   ehdr = (Elf_Ehdr *) oc->image;
4500   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
4501
4502   for( i = 0; i < ehdr->e_shnum; i++ )
4503     if( shdr[i].sh_type == SHT_SYMTAB )
4504       break;
4505
4506   if( i == ehdr->e_shnum )
4507   {
4508     errorBelch( "This ELF file contains no symtab" );
4509     return 0;
4510   }
4511
4512   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
4513   {
4514     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
4515       (int) shdr[i].sh_entsize, (int) sizeof( Elf_Sym ) );
4516
4517     return 0;
4518   }
4519
4520   return ocAllocateSymbolExtras( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
4521 }
4522
4523 #endif /* powerpc */
4524
4525 #endif /* ELF */
4526
4527 /* --------------------------------------------------------------------------
4528  * Mach-O specifics
4529  * ------------------------------------------------------------------------*/
4530
4531 #if defined(OBJFORMAT_MACHO)
4532
4533 /*
4534   Support for MachO linking on Darwin/MacOS X
4535   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
4536
4537   I hereby formally apologize for the hackish nature of this code.
4538   Things that need to be done:
4539   *) implement ocVerifyImage_MachO
4540   *) add still more sanity checks.
4541 */
4542
4543 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4544 #define mach_header mach_header_64
4545 #define segment_command segment_command_64
4546 #define section section_64
4547 #define nlist nlist_64
4548 #endif
4549
4550 #ifdef powerpc_HOST_ARCH
4551 static int
4552 ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4553 {
4554     struct mach_header *header = (struct mach_header *) oc->image;
4555     struct load_command *lc = (struct load_command *) (header + 1);
4556     unsigned i;
4557
4558     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: start\n"));
4559
4560     for (i = 0; i < header->ncmds; i++) {
4561         if (lc->cmd == LC_SYMTAB) {
4562
4563                 // Find out the first and last undefined external
4564                 // symbol, so we don't have to allocate too many
4565             // jump islands/GOT entries.
4566
4567             struct symtab_command *symLC = (struct symtab_command *) lc;
4568             unsigned min = symLC->nsyms, max = 0;
4569             struct nlist *nlist =
4570                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
4571                       : NULL;
4572
4573             for (i = 0; i < symLC->nsyms; i++) {
4574
4575                 if (nlist[i].n_type & N_STAB) {
4576                     ;
4577                 } else if (nlist[i].n_type & N_EXT) {
4578
4579                     if((nlist[i].n_type & N_TYPE) == N_UNDF
4580                         && (nlist[i].n_value == 0)) {
4581
4582                         if (i < min) {
4583                             min = i;
4584                         }
4585
4586                         if (i > max) {
4587                             max = i;
4588                     }
4589                 }
4590             }
4591             }
4592
4593             if (max >= min) {
4594                 return ocAllocateSymbolExtras(oc, max - min + 1, min);
4595             }
4596
4597             break;
4598         }
4599
4600         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4601     }
4602
4603     return ocAllocateSymbolExtras(oc,0,0);
4604 }
4605
4606 #endif
4607 #ifdef x86_64_HOST_ARCH
4608 static int
4609 ocAllocateSymbolExtras_MachO(ObjectCode* oc)
4610 {
4611     struct mach_header *header = (struct mach_header *) oc->image;
4612     struct load_command *lc = (struct load_command *) (header + 1);
4613     unsigned i;
4614
4615     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: start\n"));
4616
4617     for (i = 0; i < header->ncmds; i++) {
4618         if (lc->cmd == LC_SYMTAB) {
4619
4620                 // Just allocate one entry for every symbol
4621             struct symtab_command *symLC = (struct symtab_command *) lc;
4622
4623             IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: allocate %d symbols\n", symLC->nsyms));
4624             IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: done\n"));
4625             return ocAllocateSymbolExtras(oc, symLC->nsyms, 0);
4626         }
4627
4628         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
4629     }
4630
4631     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: allocated no symbols\n"));
4632     IF_DEBUG(linker, debugBelch("ocAllocateSymbolExtras_MachO: done\n"));
4633     return ocAllocateSymbolExtras(oc,0,0);
4634 }
4635 #endif
4636
4637 static int
4638 ocVerifyImage_MachO(ObjectCode * oc)
4639 {
4640     char *image = (char*) oc->image;
4641     struct mach_header *header = (struct mach_header*) image;
4642
4643     IF_DEBUG(linker, debugBelch("ocVerifyImage_MachO: start\n"));
4644
4645 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
4646     if(header->magic != MH_MAGIC_64) {
4647         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4648                    oc->fileName, MH_MAGIC_64, header->magic);
4649         return 0;
4650     }
4651 #else
4652     if(header->magic != MH_MAGIC) {
4653         errorBelch("%s: Bad magic. Expected: %08x, got: %08x.\n",
4654                    oc->fileName, MH_MAGIC, header->magic);
4655         return 0;
4656     }
4657 #endif
4658
4659     // FIXME: do some more verifying here
4660     IF_DEBUG(linker, debugBelch("ocVerifyImage_MachO: done\n"));
4661     return 1;
4662 }
4663
4664 static int
4665 resolveImports(
4666     ObjectCode* oc,
4667     char *image,
4668     struct symtab_command *symLC,
4669     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
4670     unsigned long *indirectSyms,
4671     struct nlist *nlist)
4672 {
4673     unsigned i;
4674     size_t itemSize = 4;
4675
4676     IF_DEBUG(linker, debugBelch("resolveImports: start\n"));
4677
4678 #if i386_HOST_ARCH
4679     int isJumpTable = 0;
4680
4681     if (strcmp(sect->sectname,"__jump_table") == 0) {
4682         isJumpTable = 1;
4683         itemSize = 5;
4684         ASSERT(sect->reserved2 == itemSize);
4685     }
4686
4687 #endif
4688
4689     for(i=0; i*itemSize < sect->size;i++)
4690     {
4691         // according to otool, reserved1 contains the first index into the indirect symbol table
4692         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
4693         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4694         void *addr = NULL;
4695
4696         IF_DEBUG(linker, debugBelch("resolveImports: resolving %s\n", nm));
4697
4698         if ((symbol->n_type & N_TYPE) == N_UNDF
4699             && (symbol->n_type & N_EXT) && (symbol->n_value != 0)) {
4700             addr = (void*) (symbol->n_value);
4701             IF_DEBUG(linker, debugBelch("resolveImports: undefined external %s has value %p\n", nm, addr));
4702         } else {
4703             addr = lookupSymbol(nm);
4704             IF_DEBUG(linker, debugBelch("resolveImports: looking up %s, %p\n", nm, addr));
4705         }
4706         if (!addr)
4707         {
4708             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
4709             return 0;
4710         }
4711         ASSERT(addr);
4712
4713 #if i386_HOST_ARCH
4714         if (isJumpTable) {
4715             checkProddableBlock(oc,image + sect->offset + i*itemSize);
4716
4717             *(image + sect->offset + i * itemSize) = 0xe9; // jmp opcode
4718             *(unsigned*)(image + sect->offset + i*itemSize + 1)
4719                 = (char*)addr - (image + sect->offset + i*itemSize + 5);
4720         }
4721         else
4722 #endif
4723         {
4724             checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
4725             ((void**)(image + sect->offset))[i] = addr;
4726         }
4727     }
4728
4729     IF_DEBUG(linker, debugBelch("resolveImports: done\n"));
4730     return 1;
4731 }
4732
4733 static unsigned long relocateAddress(
4734     ObjectCode* oc,
4735     int nSections,
4736     struct section* sections,
4737     unsigned long address)
4738 {
4739     int i;
4740     IF_DEBUG(linker, debugBelch("relocateAddress: start\n"));
4741     for (i = 0; i < nSections; i++)
4742     {
4743             IF_DEBUG(linker, debugBelch("    relocating address in section %d\n", i));
4744         if (sections[i].addr <= address
4745             && address < sections[i].addr + sections[i].size)
4746         {
4747             return (unsigned long)oc->image
4748                     + sections[i].offset + address - sections[i].addr;
4749         }
4750     }
4751     barf("Invalid Mach-O file:"
4752          "Address out of bounds while relocating object file");
4753     return 0;
4754 }
4755
4756 static int relocateSection(
4757     ObjectCode* oc,
4758     char *image,
4759     struct symtab_command *symLC, struct nlist *nlist,
4760     int nSections, struct section* sections, struct section *sect)
4761 {
4762     struct relocation_info *relocs;
4763     int i, n;
4764
4765     IF_DEBUG(linker, debugBelch("relocateSection: start\n"));
4766
4767     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
4768         return 1;
4769     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
4770         return 1;
4771     else if(!strcmp(sect->sectname,"__la_sym_ptr2"))
4772         return 1;
4773     else if(!strcmp(sect->sectname,"__la_sym_ptr3"))
4774         return 1;
4775
4776     n = sect->nreloc;
4777     IF_DEBUG(linker, debugBelch("relocateSection: number of relocations: %d\n", n));
4778
4779     relocs = (struct relocation_info*) (image + sect->reloff);
4780
4781     for(i=0;i<n;i++)
4782     {
4783 #ifdef x86_64_HOST_ARCH
4784         struct relocation_info *reloc = &relocs[i];
4785
4786         char    *thingPtr = image + sect->offset + reloc->r_address;
4787         uint64_t thing;
4788         /* We shouldn't need to initialise this, but gcc on OS X 64 bit
4789            complains that it may be used uninitialized if we don't */
4790         uint64_t value = 0;
4791         uint64_t baseValue;
4792         int type = reloc->r_type;
4793
4794         checkProddableBlock(oc,thingPtr);
4795         switch(reloc->r_length)
4796         {
4797             case 0:
4798                 thing = *(uint8_t*)thingPtr;
4799                 baseValue = (uint64_t)thingPtr + 1;
4800                 break;
4801             case 1:
4802                 thing = *(uint16_t*)thingPtr;
4803                 baseValue = (uint64_t)thingPtr + 2;
4804                 break;
4805             case 2:
4806                 thing = *(uint32_t*)thingPtr;
4807                 baseValue = (uint64_t)thingPtr + 4;
4808                 break;
4809             case 3:
4810                 thing = *(uint64_t*)thingPtr;
4811                 baseValue = (uint64_t)thingPtr + 8;
4812                 break;
4813             default:
4814                 barf("Unknown size.");
4815         }
4816
4817         IF_DEBUG(linker,
4818                  debugBelch("relocateSection: length = %d, thing = %" PRId64 ", baseValue = %p\n",
4819                             reloc->r_length, thing, (char *)baseValue));
4820
4821         if (type == X86_64_RELOC_GOT
4822            || type == X86_64_RELOC_GOT_LOAD)
4823         {
4824             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4825             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4826
4827             IF_DEBUG(linker, debugBelch("relocateSection: making jump island for %s, extern = %d, X86_64_RELOC_GOT\n", nm, reloc->r_extern));
4828             ASSERT(reloc->r_extern);
4829             value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, (unsigned long)lookupSymbol(nm))->addr;
4830
4831             type = X86_64_RELOC_SIGNED;
4832         }
4833         else if(reloc->r_extern)
4834         {
4835             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4836             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4837
4838             IF_DEBUG(linker, debugBelch("relocateSection: looking up external symbol %s\n", nm));
4839             IF_DEBUG(linker, debugBelch("               : type  = %d\n", symbol->n_type));
4840             IF_DEBUG(linker, debugBelch("               : sect  = %d\n", symbol->n_sect));
4841             IF_DEBUG(linker, debugBelch("               : desc  = %d\n", symbol->n_desc));
4842             IF_DEBUG(linker, debugBelch("               : value = %p\n", (void *)symbol->n_value));
4843             if ((symbol->n_type & N_TYPE) == N_SECT) {
4844                 value = relocateAddress(oc, nSections, sections,
4845                                         symbol->n_value);
4846                 IF_DEBUG(linker, debugBelch("relocateSection, defined external symbol %s, relocated address %p\n", nm, (void *)value));
4847             }
4848             else {
4849                 value = (uint64_t) lookupSymbol(nm);
4850                 IF_DEBUG(linker, debugBelch("relocateSection: external symbol %s, address %p\n", nm, (void *)value));
4851             }
4852         }
4853         else
4854         {
4855             // If the relocation is not through the global offset table
4856             // or external, then set the value to the baseValue.  This
4857             // will leave displacements into the __const section
4858             // unchanged (as they ought to be).
4859
4860             value = baseValue;
4861         }
4862
4863         IF_DEBUG(linker, debugBelch("relocateSection: value = %p\n", (void *)value));
4864
4865         if (type == X86_64_RELOC_BRANCH)
4866         {
4867             if((int32_t)(value - baseValue) != (int64_t)(value - baseValue))
4868             {
4869                 ASSERT(reloc->r_extern);
4870                 value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)
4871                                         -> jumpIsland;
4872             }
4873             ASSERT((int32_t)(value - baseValue) == (int64_t)(value - baseValue));
4874             type = X86_64_RELOC_SIGNED;
4875         }
4876
4877         switch(type)
4878         {
4879             case X86_64_RELOC_UNSIGNED:
4880                 ASSERT(!reloc->r_pcrel);
4881                 thing += value;
4882                 break;
4883             case X86_64_RELOC_SIGNED:
4884             case X86_64_RELOC_SIGNED_1:
4885             case X86_64_RELOC_SIGNED_2:
4886             case X86_64_RELOC_SIGNED_4:
4887                 ASSERT(reloc->r_pcrel);
4888                 thing += value - baseValue;
4889                 break;
4890             case X86_64_RELOC_SUBTRACTOR:
4891                 ASSERT(!reloc->r_pcrel);
4892                 thing -= value;
4893                 break;
4894             default:
4895                 barf("unkown relocation");
4896         }
4897
4898         switch(reloc->r_length)
4899         {
4900             case 0:
4901                 *(uint8_t*)thingPtr = thing;
4902                 break;
4903             case 1:
4904                 *(uint16_t*)thingPtr = thing;
4905                 break;
4906             case 2:
4907                 *(uint32_t*)thingPtr = thing;
4908                 break;
4909             case 3:
4910                 *(uint64_t*)thingPtr = thing;
4911                 break;
4912         }
4913 #else
4914         if(relocs[i].r_address & R_SCATTERED)
4915         {
4916             struct scattered_relocation_info *scat =
4917                 (struct scattered_relocation_info*) &relocs[i];
4918
4919             if(!scat->r_pcrel)
4920             {
4921                 if(scat->r_length == 2)
4922                 {
4923                     unsigned long word = 0;
4924                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
4925                     checkProddableBlock(oc,wordPtr);
4926
4927                     // Note on relocation types:
4928                     // i386 uses the GENERIC_RELOC_* types,
4929                     // while ppc uses special PPC_RELOC_* types.
4930                     // *_RELOC_VANILLA and *_RELOC_PAIR have the same value
4931                     // in both cases, all others are different.
4932                     // Therefore, we use GENERIC_RELOC_VANILLA
4933                     // and GENERIC_RELOC_PAIR instead of the PPC variants,
4934                     // and use #ifdefs for the other types.
4935
4936                     // Step 1: Figure out what the relocated value should be
4937                     if (scat->r_type == GENERIC_RELOC_VANILLA) {
4938                         word = *wordPtr
4939                              + (unsigned long) relocateAddress(oc,
4940                                                                 nSections,
4941                                                                 sections,
4942                                                                 scat->r_value)
4943                                         - scat->r_value;
4944                     }
4945 #ifdef powerpc_HOST_ARCH
4946                     else if(scat->r_type == PPC_RELOC_SECTDIFF
4947                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
4948                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
4949                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF
4950                         || scat->r_type == PPC_RELOC_LOCAL_SECTDIFF)
4951 #else
4952                     else if(scat->r_type == GENERIC_RELOC_SECTDIFF
4953                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
4954 #endif
4955                     {
4956                         struct scattered_relocation_info *pair =
4957                                 (struct scattered_relocation_info*) &relocs[i+1];
4958
4959                         if (!pair->r_scattered || pair->r_type != GENERIC_RELOC_PAIR) {
4960                             barf("Invalid Mach-O file: "
4961                                  "RELOC_*_SECTDIFF not followed by RELOC_PAIR");
4962                         }
4963
4964                         word = (unsigned long)
4965                                (relocateAddress(oc, nSections, sections, scat->r_value)
4966                               - relocateAddress(oc, nSections, sections, pair->r_value));
4967                         i++;
4968                     }
4969 #ifdef powerpc_HOST_ARCH
4970                     else if(scat->r_type == PPC_RELOC_HI16
4971                          || scat->r_type == PPC_RELOC_LO16
4972                          || scat->r_type == PPC_RELOC_HA16
4973                          || scat->r_type == PPC_RELOC_LO14)
4974                     {   // these are generated by label+offset things
4975                         struct relocation_info *pair = &relocs[i+1];
4976
4977                         if ((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR) {
4978                             barf("Invalid Mach-O file: "
4979                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
4980                         }
4981
4982                         if(scat->r_type == PPC_RELOC_LO16)
4983                         {
4984                             word = ((unsigned short*) wordPtr)[1];
4985                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4986                         }
4987                         else if(scat->r_type == PPC_RELOC_LO14)
4988                         {
4989                             barf("Unsupported Relocation: PPC_RELOC_LO14");
4990                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
4991                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4992                         }
4993                         else if(scat->r_type == PPC_RELOC_HI16)
4994                         {
4995                             word = ((unsigned short*) wordPtr)[1] << 16;
4996                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4997                         }
4998                         else if(scat->r_type == PPC_RELOC_HA16)
4999                         {
5000                             word = ((unsigned short*) wordPtr)[1] << 16;
5001                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
5002                         }
5003
5004
5005                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
5006                                                 - scat->r_value;
5007
5008                         i++;
5009                     }
5010  #endif
5011                     else {
5012                         barf ("Don't know how to handle this Mach-O "
5013                               "scattered relocation entry: "
5014                               "object file %s; entry type %ld; "
5015                               "address %#lx\n",
5016                               OC_INFORMATIVE_FILENAME(oc),
5017                               scat->r_type,
5018                               scat->r_address);
5019                         return 0;
5020                      }
5021
5022 #ifdef powerpc_HOST_ARCH
5023                     if(scat->r_type == GENERIC_RELOC_VANILLA
5024                         || scat->r_type == PPC_RELOC_SECTDIFF)
5025 #else
5026                     if(scat->r_type == GENERIC_RELOC_VANILLA
5027                         || scat->r_type == GENERIC_RELOC_SECTDIFF
5028                         || scat->r_type == GENERIC_RELOC_LOCAL_SECTDIFF)
5029 #endif
5030                     {
5031                         *wordPtr = word;
5032                     }
5033 #ifdef powerpc_HOST_ARCH
5034                     else if (scat->r_type == PPC_RELOC_LO16_SECTDIFF
5035                           || scat->r_type == PPC_RELOC_LO16)
5036                     {
5037                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
5038                     }
5039                     else if (scat->r_type == PPC_RELOC_HI16_SECTDIFF
5040                           || scat->r_type == PPC_RELOC_HI16)
5041                     {
5042                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
5043                     }
5044                     else if (scat->r_type == PPC_RELOC_HA16_SECTDIFF
5045                           || scat->r_type == PPC_RELOC_HA16)
5046                     {
5047                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
5048                             + ((word & (1<<15)) ? 1 : 0);
5049                     }
5050 #endif
5051                 }
5052                 else
5053                 {
5054                     barf("Can't handle Mach-O scattered relocation entry "
5055                          "with this r_length tag: "
5056                          "object file %s; entry type %ld; "
5057                          "r_length tag %ld; address %#lx\n",
5058                          OC_INFORMATIVE_FILENAME(oc),
5059                          scat->r_type,
5060                          scat->r_length,
5061                          scat->r_address);
5062                     return 0;
5063                 }
5064             }
5065             else /* scat->r_pcrel */
5066             {
5067                 barf("Don't know how to handle *PC-relative* Mach-O "
5068                      "scattered relocation entry: "
5069                      "object file %s; entry type %ld; address %#lx\n",
5070                      OC_INFORMATIVE_FILENAME(oc),
5071                      scat->r_type,
5072                      scat->r_address);
5073                return 0;
5074             }
5075
5076         }
5077         else /* !(relocs[i].r_address & R_SCATTERED) */
5078         {
5079             struct relocation_info *reloc = &relocs[i];
5080             if (reloc->r_pcrel && !reloc->r_extern) {
5081                 IF_DEBUG(linker, debugBelch("relocateSection: pc relative but not external, skipping\n"));
5082                 continue;
5083             }
5084
5085             if (reloc->r_length == 2) {
5086                 unsigned long word = 0;
5087 #ifdef powerpc_HOST_ARCH
5088                 unsigned long jumpIsland = 0;
5089                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
5090                                                       // to avoid warning and to catch
5091                                                       // bugs.
5092 #endif
5093
5094                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
5095                 checkProddableBlock(oc,wordPtr);
5096
5097                 if (reloc->r_type == GENERIC_RELOC_VANILLA) {
5098                     word = *wordPtr;
5099                 }
5100 #ifdef powerpc_HOST_ARCH
5101                 else if (reloc->r_type == PPC_RELOC_LO16) {
5102                     word = ((unsigned short*) wordPtr)[1];
5103                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
5104                 }
5105                 else if (reloc->r_type == PPC_RELOC_HI16) {
5106                     word = ((unsigned short*) wordPtr)[1] << 16;
5107                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
5108                 }
5109                 else if (reloc->r_type == PPC_RELOC_HA16) {
5110                     word = ((unsigned short*) wordPtr)[1] << 16;
5111                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
5112                 }
5113                 else if (reloc->r_type == PPC_RELOC_BR24) {
5114                     word = *wordPtr;
5115                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
5116                 }
5117 #endif
5118                 else {
5119                     barf("Can't handle this Mach-O relocation entry "
5120                          "(not scattered): "
5121                          "object file %s; entry type %ld; address %#lx\n",
5122                          OC_INFORMATIVE_FILENAME(oc),
5123                          reloc->r_type,
5124                          reloc->r_address);
5125                     return 0;
5126                 }
5127
5128                 if (!reloc->r_extern) {
5129                     long delta = sections[reloc->r_symbolnum-1].offset
5130                         - sections[reloc->r_symbolnum-1].addr
5131                         + ((long) image);
5132
5133                     word += delta;
5134                 }
5135                 else {
5136                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
5137                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
5138                     void *symbolAddress = lookupSymbol(nm);
5139
5140                     if (!symbolAddress) {
5141                         errorBelch("\nunknown symbol `%s'", nm);
5142                         return 0;
5143                     }
5144
5145                     if (reloc->r_pcrel) {
5146 #ifdef powerpc_HOST_ARCH
5147                             // In the .o file, this should be a relative jump to NULL
5148                             // and we'll change it to a relative jump to the symbol
5149                         ASSERT(word + reloc->r_address == 0);
5150                         jumpIsland = (unsigned long)
5151                                         &makeSymbolExtra(oc,
5152                                                          reloc->r_symbolnum,
5153                                                          (unsigned long) symbolAddress)
5154                                          -> jumpIsland;
5155                         if (jumpIsland != 0) {
5156                             offsetToJumpIsland = word + jumpIsland
5157                                 - (((long)image) + sect->offset - sect->addr);
5158                         }
5159 #endif
5160                         word += (unsigned long) symbolAddress
5161                                 - (((long)image) + sect->offset - sect->addr);
5162                     }
5163                     else {
5164                         word += (unsigned long) symbolAddress;
5165                     }
5166                 }
5167
5168                 if (reloc->r_type == GENERIC_RELOC_VANILLA) {
5169                     *wordPtr = word;
5170                     continue;
5171                 }
5172 #ifdef powerpc_HOST_ARCH
5173                 else if(reloc->r_type == PPC_RELOC_LO16)
5174                 {
5175                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
5176                     i++;
5177                     continue;
5178                 }
5179                 else if(reloc->r_type == PPC_RELOC_HI16)
5180                 {
5181                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
5182                     i++;
5183                     continue;
5184                 }
5185                 else if(reloc->r_type == PPC_RELOC_HA16)
5186                 {
5187                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
5188                         + ((word & (1<<15)) ? 1 : 0);
5189                     i++;
5190                     continue;
5191                 }
5192                 else if(reloc->r_type == PPC_RELOC_BR24)
5193                 {
5194                     if ((word & 0x03) != 0) {
5195                         barf("%s: unconditional relative branch with a displacement "
5196                              "which isn't a multiple of 4 bytes: %#lx",
5197                              OC_INFORMATIVE_FILENAME(oc),
5198                              word);
5199                     }
5200
5201                     if((word & 0xFE000000) != 0xFE000000 &&
5202                         (word & 0xFE000000) != 0x00000000) {
5203                         // The branch offset is too large.
5204                         // Therefore, we try to use a jump island.
5205                         if (jumpIsland == 0) {
5206                             barf("%s: unconditional relative branch out of range: "
5207                                  "no jump island available: %#lx",
5208                                  OC_INFORMATIVE_FILENAME(oc),
5209                                  word);
5210                         }
5211
5212                         word = offsetToJumpIsland;
5213
5214                         if((word & 0xFE000000) != 0xFE000000 &&
5215                             (word & 0xFE000000) != 0x00000000) {
5216                             barf("%s: unconditional relative branch out of range: "
5217                                  "jump island out of range: %#lx",
5218                                  OC_INFORMATIVE_FILENAME(oc),
5219                                  word);
5220                     }
5221                     }
5222                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
5223                     continue;
5224                 }
5225 #endif
5226             }
5227             else
5228             {
5229                  barf("Can't handle Mach-O relocation entry (not scattered) "
5230                       "with this r_length tag: "
5231                       "object file %s; entry type %ld; "
5232                       "r_length tag %ld; address %#lx\n",
5233                       OC_INFORMATIVE_FILENAME(oc),
5234                       reloc->r_type,
5235                       reloc->r_length,
5236                       reloc->r_address);
5237                  return 0;
5238             }
5239         }
5240 #endif
5241     }
5242
5243     IF_DEBUG(linker, debugBelch("relocateSection: done\n"));
5244     return 1;
5245 }
5246
5247 static int
5248 ocGetNames_MachO(ObjectCode* oc)
5249 {
5250     char *image = (char*) oc->image;
5251     struct mach_header *header = (struct mach_header*) image;
5252     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
5253     unsigned i,curSymbol = 0;
5254     struct segment_command *segLC = NULL;
5255     struct section *sections;
5256     struct symtab_command *symLC = NULL;
5257     struct nlist *nlist;
5258     unsigned long commonSize = 0;
5259     char    *commonStorage = NULL;
5260     unsigned long commonCounter;
5261
5262     IF_DEBUG(linker,debugBelch("ocGetNames_MachO: start\n"));
5263
5264     for(i=0;i<header->ncmds;i++)
5265     {
5266         if (lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64) {
5267             segLC = (struct segment_command*) lc;
5268         }
5269         else if (lc->cmd == LC_SYMTAB) {
5270             symLC = (struct symtab_command*) lc;
5271         }
5272
5273         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
5274     }
5275
5276     sections = (struct section*) (segLC+1);
5277     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
5278                   : NULL;
5279
5280     if (!segLC) {
5281         barf("ocGetNames_MachO: no segment load command");
5282     }
5283
5284     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: will load %d sections\n", segLC->nsects));
5285     for(i=0;i<segLC->nsects;i++)
5286     {
5287         IF_DEBUG(linker, debugBelch("ocGetNames_MachO: section %d\n", i));
5288
5289         if (sections[i].size == 0) {
5290             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: found a zero length section, skipping\n"));
5291             continue;
5292         }
5293
5294         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
5295         {
5296             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
5297                                       "ocGetNames_MachO(common symbols)");
5298             sections[i].offset = zeroFillArea - image;
5299         }
5300
5301         if (!strcmp(sections[i].sectname,"__text")) {
5302
5303             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __text section\n"));
5304             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
5305                 (void*) (image + sections[i].offset),
5306                 (void*) (image + sections[i].offset + sections[i].size));
5307         }
5308         else if (!strcmp(sections[i].sectname,"__const")) {
5309
5310             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __const section\n"));
5311             addSection(oc, SECTIONKIND_RWDATA,
5312                 (void*) (image + sections[i].offset),
5313                 (void*) (image + sections[i].offset + sections[i].size));
5314         }
5315         else if (!strcmp(sections[i].sectname,"__data")) {
5316
5317             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __data section\n"));
5318             addSection(oc, SECTIONKIND_RWDATA,
5319                 (void*) (image + sections[i].offset),
5320                 (void*) (image + sections[i].offset + sections[i].size));
5321         }
5322         else if(!strcmp(sections[i].sectname,"__bss")
5323                 || !strcmp(sections[i].sectname,"__common")) {
5324
5325             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: adding __bss section\n"));
5326             addSection(oc, SECTIONKIND_RWDATA,
5327                 (void*) (image + sections[i].offset),
5328                 (void*) (image + sections[i].offset + sections[i].size));
5329         }
5330         addProddableBlock(oc,
5331                           (void *) (image + sections[i].offset),
5332                                         sections[i].size);
5333     }
5334
5335         // count external symbols defined here
5336     oc->n_symbols = 0;
5337     if (symLC) {
5338         for (i = 0; i < symLC->nsyms; i++) {
5339             if (nlist[i].n_type & N_STAB) {
5340                 ;
5341             }
5342             else if(nlist[i].n_type & N_EXT)
5343             {
5344                 if((nlist[i].n_type & N_TYPE) == N_UNDF
5345                     && (nlist[i].n_value != 0))
5346                 {
5347                     commonSize += nlist[i].n_value;
5348                     oc->n_symbols++;
5349                 }
5350                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
5351                     oc->n_symbols++;
5352             }
5353         }
5354     }
5355     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: %d external symbols\n", oc->n_symbols));
5356     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
5357                                    "ocGetNames_MachO(oc->symbols)");
5358
5359     if(symLC)
5360     {
5361         for(i=0;i<symLC->nsyms;i++)
5362         {
5363             if(nlist[i].n_type & N_STAB)
5364                 ;
5365             else if((nlist[i].n_type & N_TYPE) == N_SECT)
5366             {
5367                 if(nlist[i].n_type & N_EXT)
5368                 {
5369                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
5370                     if ((nlist[i].n_desc & N_WEAK_DEF) && lookupSymbol(nm)) {
5371                         // weak definition, and we already have a definition
5372                         IF_DEBUG(linker, debugBelch("    weak: %s\n", nm));
5373                     }
5374                     else
5375                     {
5376                             IF_DEBUG(linker, debugBelch("ocGetNames_MachO: inserting %s\n", nm));
5377                             ghciInsertStrHashTable(oc->fileName, symhash, nm,
5378                                                     image
5379                                                     + sections[nlist[i].n_sect-1].offset
5380                                                     - sections[nlist[i].n_sect-1].addr
5381                                                     + nlist[i].n_value);
5382                             oc->symbols[curSymbol++] = nm;
5383                     }
5384                 }
5385                 else
5386                 {
5387                     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: \t...not external, skipping\n"));
5388                 }
5389             }
5390             else
5391             {
5392                 IF_DEBUG(linker, debugBelch("ocGetNames_MachO: \t...not defined in this section, skipping\n"));
5393             }
5394         }
5395     }
5396
5397     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
5398     commonCounter = (unsigned long)commonStorage;
5399
5400     if (symLC) {
5401         for (i = 0; i < symLC->nsyms; i++) {
5402             if((nlist[i].n_type & N_TYPE) == N_UNDF
5403              && (nlist[i].n_type & N_EXT)
5404              && (nlist[i].n_value != 0)) {
5405
5406                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
5407                 unsigned long sz = nlist[i].n_value;
5408
5409                 nlist[i].n_value = commonCounter;
5410
5411                 IF_DEBUG(linker, debugBelch("ocGetNames_MachO: inserting common symbol: %s\n", nm));
5412                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
5413                                        (void*)commonCounter);
5414                 oc->symbols[curSymbol++] = nm;
5415
5416                 commonCounter += sz;
5417             }
5418         }
5419     }
5420
5421     IF_DEBUG(linker, debugBelch("ocGetNames_MachO: done\n"));
5422     return 1;
5423 }
5424
5425 static int
5426 ocResolve_MachO(ObjectCode* oc)
5427 {
5428     char *image = (char*) oc->image;
5429     struct mach_header *header = (struct mach_header*) image;
5430     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
5431     unsigned i;
5432     struct segment_command *segLC = NULL;
5433     struct section *sections;
5434     struct symtab_command *symLC = NULL;
5435     struct dysymtab_command *dsymLC = NULL;
5436     struct nlist *nlist;
5437
5438     IF_DEBUG(linker, debugBelch("ocResolve_MachO: start\n"));
5439     for (i = 0; i < header->ncmds; i++)
5440     {
5441         if (lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64) {
5442             segLC = (struct segment_command*) lc;
5443             IF_DEBUG(linker, debugBelch("ocResolve_MachO: found a 32 or 64 bit segment load command\n"));
5444         }
5445         else if (lc->cmd == LC_SYMTAB) {
5446             symLC = (struct symtab_command*) lc;
5447             IF_DEBUG(linker, debugBelch("ocResolve_MachO: found a symbol table load command\n"));
5448         }
5449         else if (lc->cmd == LC_DYSYMTAB) {
5450             dsymLC = (struct dysymtab_command*) lc;
5451             IF_DEBUG(linker, debugBelch("ocResolve_MachO: found a dynamic symbol table load command\n"));
5452         }
5453
5454         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
5455     }
5456
5457     sections = (struct section*) (segLC+1);
5458     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
5459                   : NULL;
5460
5461     if(dsymLC)
5462     {
5463         unsigned long *indirectSyms
5464             = (unsigned long*) (image + dsymLC->indirectsymoff);
5465
5466         IF_DEBUG(linker, debugBelch("ocResolve_MachO: resolving dsymLC\n"));
5467         for (i = 0; i < segLC->nsects; i++)
5468         {
5469             if(    !strcmp(sections[i].sectname,"__la_symbol_ptr")
5470                 || !strcmp(sections[i].sectname,"__la_sym_ptr2")
5471                 || !strcmp(sections[i].sectname,"__la_sym_ptr3"))
5472             {
5473                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5474                     return 0;
5475             }
5476             else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr")
5477                 ||  !strcmp(sections[i].sectname,"__pointers"))
5478             {
5479                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5480                     return 0;
5481             }
5482             else if(!strcmp(sections[i].sectname,"__jump_table"))
5483             {
5484                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
5485                     return 0;
5486             }
5487             else
5488             {
5489                 IF_DEBUG(linker, debugBelch("ocResolve_MachO: unknown section\n"));
5490             }
5491         }
5492     }
5493
5494     for(i=0;i<segLC->nsects;i++)
5495     {
5496             IF_DEBUG(linker, debugBelch("ocResolve_MachO: relocating section %d\n", i));
5497
5498         if (!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
5499             return 0;
5500     }
5501
5502 #if defined (powerpc_HOST_ARCH)
5503     ocFlushInstructionCache( oc );
5504 #endif
5505
5506     return 1;
5507 }
5508
5509 #ifdef powerpc_HOST_ARCH
5510 /*
5511  * The Mach-O object format uses leading underscores. But not everywhere.
5512  * There is a small number of runtime support functions defined in
5513  * libcc_dynamic.a whose name does not have a leading underscore.
5514  * As a consequence, we can't get their address from C code.
5515  * We have to use inline assembler just to take the address of a function.
5516  * Yuck.
5517  */
5518
5519 extern void* symbolsWithoutUnderscore[];
5520
5521 static void
5522 machoInitSymbolsWithoutUnderscore(void)
5523 {
5524     void **p = symbolsWithoutUnderscore;
5525     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
5526
5527 #undef SymI_NeedsProto
5528 #define SymI_NeedsProto(x)  \
5529     __asm__ volatile(".long " # x);
5530
5531     RTS_MACHO_NOUNDERLINE_SYMBOLS
5532
5533     __asm__ volatile(".text");
5534
5535 #undef SymI_NeedsProto
5536 #define SymI_NeedsProto(x)  \
5537     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
5538
5539     RTS_MACHO_NOUNDERLINE_SYMBOLS
5540
5541 #undef SymI_NeedsProto
5542 }
5543 #endif
5544
5545 #ifndef USE_MMAP
5546 /*
5547  * Figure out by how much to shift the entire Mach-O file in memory
5548  * when loading so that its single segment ends up 16-byte-aligned
5549  */
5550 static int
5551 machoGetMisalignment( FILE * f )
5552 {
5553     struct mach_header header;
5554     int misalignment;
5555
5556     {
5557         int n = fread(&header, sizeof(header), 1, f);
5558         if (n != 1) {
5559             barf("machoGetMisalignment: can't read the Mach-O header");
5560         }
5561     }
5562     fseek(f, -sizeof(header), SEEK_CUR);
5563
5564 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
5565     if(header.magic != MH_MAGIC_64) {
5566         barf("Bad magic. Expected: %08x, got: %08x.",
5567              MH_MAGIC_64, header.magic);
5568     }
5569 #else
5570     if(header.magic != MH_MAGIC) {
5571         barf("Bad magic. Expected: %08x, got: %08x.",
5572              MH_MAGIC, header.magic);
5573     }
5574 #endif
5575
5576     misalignment = (header.sizeofcmds + sizeof(header))
5577                     & 0xF;
5578
5579     return misalignment ? (16 - misalignment) : 0;
5580 }
5581 #endif
5582
5583 #endif