rts/Linker.c

   1 /* -----------------------------------------------------------------------------
   2  *
   3  * (c) The GHC Team, 2000-2004
   4  *
   5  * RTS Object Linker
   6  *
   7  * ---------------------------------------------------------------------------*/
   8
   9 #if 0
  10 #include "PosixSource.h"
  11 #endif
  12
  13 /* Linux needs _GNU_SOURCE to get RTLD_DEFAULT from <dlfcn.h> and
  14    MREMAP_MAYMOVE from <sys/mman.h>.
  15  */
  16 #ifdef __linux__
  17 #define _GNU_SOURCE
  18 #endif
  19
  20 #include "Rts.h"
  21 #include "RtsFlags.h"
  22 #include "HsFFI.h"
  23 #include "Hash.h"
  24 #include "Linker.h"
  25 #include "LinkerInternals.h"
  26 #include "RtsUtils.h"
  27 #include "Schedule.h"
  28 #include "Sparks.h"
  29 #include "RtsGlobals.h"
  30 #include "Timer.h"
  31 #include "Trace.h"
  32
  33 #ifdef HAVE_SYS_TYPES_H
  34 #include <sys/types.h>
  35 #endif
  36
  37 #include <stdlib.h>
  38 #include <string.h>
  39
  40 #ifdef HAVE_SYS_STAT_H
  41 #include <sys/stat.h>
  42 #endif
  43
  44 #if defined(HAVE_DLFCN_H)
  45 #include <dlfcn.h>
  46 #endif
  47
  48 #if defined(cygwin32_HOST_OS)
  49 #ifdef HAVE_DIRENT_H
  50 #include <dirent.h>
  51 #endif
  52
  53 #ifdef HAVE_SYS_TIME_H
  54 #include <sys/time.h>
  55 #endif
  56 #include <regex.h>
  57 #include <sys/fcntl.h>
  58 #include <sys/termios.h>
  59 #include <sys/utime.h>
  60 #include <sys/utsname.h>
  61 #include <sys/wait.h>
  62 #endif
  63
  64 #if defined(ia64_HOST_ARCH) || defined(linux_HOST_OS) || defined(freebsd_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  65 #define USE_MMAP
  66 #include <fcntl.h>
  67 #include <sys/mman.h>
  68
  69 #if defined(linux_HOST_OS) || defined(freebsd_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  70 #ifdef HAVE_UNISTD_H
  71 #include <unistd.h>
  72 #endif
  73 #endif
  74
  75 #endif
  76
  77 #if defined(linux_HOST_OS) || defined(solaris2_HOST_OS) || defined(freebsd_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS)
  78 #  define OBJFORMAT_ELF
  79 #elif defined(cygwin32_HOST_OS) || defined (mingw32_HOST_OS)
  80 #  define OBJFORMAT_PEi386
  81 #  include <windows.h>
  82 #  include <math.h>
  83 #elif defined(darwin_HOST_OS)
  84 #  define OBJFORMAT_MACHO
  85 #  include <mach-o/loader.h>
  86 #  include <mach-o/nlist.h>
  87 #  include <mach-o/reloc.h>
  88 #if !defined(HAVE_DLFCN_H)
  89 #  include <mach-o/dyld.h>
  90 #endif
  91 #if defined(powerpc_HOST_ARCH)
  92 #  include <mach-o/ppc/reloc.h>
  93 #endif
  94 #if defined(x86_64_HOST_ARCH)
  95 #  include <mach-o/x86_64/reloc.h>
  96 #endif
  97 #endif
  98
  99 /* Hash table mapping symbol names to Symbol */
 100 static /*Str*/HashTable *symhash;
 101
 102 /* Hash table mapping symbol names to StgStablePtr */
 103 static /*Str*/HashTable *stablehash;
 104
 105 /* List of currently loaded objects */
 106 ObjectCode *objects = NULL;     /* initially empty */
 107
 108 #if defined(OBJFORMAT_ELF)
 109 static int ocVerifyImage_ELF    ( ObjectCode* oc );
 110 static int ocGetNames_ELF       ( ObjectCode* oc );
 111 static int ocResolve_ELF        ( ObjectCode* oc );
 112 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 113 static int ocAllocateSymbolExtras_ELF ( ObjectCode* oc );
 114 #endif
 115 #elif defined(OBJFORMAT_PEi386)
 116 static int ocVerifyImage_PEi386 ( ObjectCode* oc );
 117 static int ocGetNames_PEi386    ( ObjectCode* oc );
 118 static int ocResolve_PEi386     ( ObjectCode* oc );
 119 static void *lookupSymbolInDLLs ( unsigned char *lbl );
 120 static void zapTrailingAtSign   ( unsigned char *sym );
 121 #elif defined(OBJFORMAT_MACHO)
 122 static int ocVerifyImage_MachO    ( ObjectCode* oc );
 123 static int ocGetNames_MachO       ( ObjectCode* oc );
 124 static int ocResolve_MachO        ( ObjectCode* oc );
 125
 126 static int machoGetMisalignment( FILE * );
 127 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
 128 static int ocAllocateSymbolExtras_MachO ( ObjectCode* oc );
 129 #endif
 130 #ifdef powerpc_HOST_ARCH
 131 static void machoInitSymbolsWithoutUnderscore( void );
 132 #endif
 133 #endif
 134
 135 /* on x86_64 we have a problem with relocating symbol references in
 136  * code that was compiled without -fPIC.  By default, the small memory
 137  * model is used, which assumes that symbol references can fit in a
 138  * 32-bit slot.  The system dynamic linker makes this work for
 139  * references to shared libraries by either (a) allocating a jump
 140  * table slot for code references, or (b) moving the symbol at load
 141  * time (and copying its contents, if necessary) for data references.
 142  *
 143  * We unfortunately can't tell whether symbol references are to code
 144  * or data.  So for now we assume they are code (the vast majority
 145  * are), and allocate jump-table slots.  Unfortunately this will
 146  * SILENTLY generate crashing code for data references.  This hack is
 147  * enabled by X86_64_ELF_NONPIC_HACK.
 148  *
 149  * One workaround is to use shared Haskell libraries.  This is
 150  * coming.  Another workaround is to keep the static libraries but
 151  * compile them with -fPIC, because that will generate PIC references
 152  * to data which can be relocated.  The PIC code is still too green to
 153  * do this systematically, though.
 154  *
 155  * See bug #781
 156  * See thread http://www.haskell.org/pipermail/cvs-ghc/2007-September/038458.html
 157  *
 158  * Naming Scheme for Symbol Macros
 159  *
 160  * SymI_*: symbol is internal to the RTS. It resides in an object
 161  *         file/library that is statically.
 162  * SymE_*: symbol is external to the RTS library. It might be linked
 163  *         dynamically.
 164  *
 165  * Sym*_HasProto  : the symbol prototype is imported in an include file
 166  *                  or defined explicitly
 167  * Sym*_NeedsProto: the symbol is undefined and we add a dummy
 168  *                  default proto extern void sym(void);
 169  */
 170 #define X86_64_ELF_NONPIC_HACK 1
 171
 172 /* Link objects into the lower 2Gb on x86_64.  GHC assumes the
 173  * small memory model on this architecture (see gcc docs,
 174  * -mcmodel=small).
 175  *
 176  * MAP_32BIT not available on OpenBSD/amd64
 177  */
 178 #if defined(x86_64_HOST_ARCH) && defined(MAP_32BIT)
 179 #define TRY_MAP_32BIT MAP_32BIT
 180 #else
 181 #define TRY_MAP_32BIT 0
 182 #endif
 183
 184 /*
 185  * Due to the small memory model (see above), on x86_64 we have to map
 186  * all our non-PIC object files into the low 2Gb of the address space
 187  * (why 2Gb and not 4Gb?  Because all addresses must be reachable
 188  * using a 32-bit signed PC-relative offset). On Linux we can do this
 189  * using the MAP_32BIT flag to mmap(), however on other OSs
 190  * (e.g. *BSD, see #2063, and also on Linux inside Xen, see #2512), we
 191  * can't do this.  So on these systems, we have to pick a base address
 192  * in the low 2Gb of the address space and try to allocate memory from
 193  * there.
 194  *
 195  * We pick a default address based on the OS, but also make this
 196  * configurable via an RTS flag (+RTS -xm)
 197  */
 198 #if defined(x86_64_HOST_ARCH)
 199
 200 #if defined(MAP_32BIT)
 201 // Try to use MAP_32BIT
 202 #define MMAP_32BIT_BASE_DEFAULT 0
 203 #else
 204 // A guess: 1Gb.
 205 #define MMAP_32BIT_BASE_DEFAULT 0x40000000
 206 #endif
 207
 208 static void *mmap_32bit_base = MMAP_32BIT_BASE_DEFAULT;
 209 #endif
 210
 211 /* MAP_ANONYMOUS is MAP_ANON on some systems, e.g. OpenBSD */
 212 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
 213 #define MAP_ANONYMOUS MAP_ANON
 214 #endif
 215
 216 /* -----------------------------------------------------------------------------
 217  * Built-in symbols from the RTS
 218  */
 219
 220 typedef struct _RtsSymbolVal {
 221     char   *lbl;
 222     void   *addr;
 223 } RtsSymbolVal;
 224
 225 #if !defined(PAR)
 226 #define Maybe_Stable_Names      SymI_HasProto(mkWeakzh_fast)                    \
 227                                 SymI_HasProto(mkWeakForeignEnvzh_fast)          \
 228                                 SymI_HasProto(makeStableNamezh_fast)            \
 229                                 SymI_HasProto(finalizzeWeakzh_fast)
 230 #else
 231 /* These are not available in GUM!!! -- HWL */
 232 #define Maybe_Stable_Names
 233 #endif
 234
 235 #if !defined (mingw32_HOST_OS)
 236 #define RTS_POSIX_ONLY_SYMBOLS                  \
 237       SymI_HasProto(shutdownHaskellAndSignal)   \
 238       SymI_NeedsProto(lockFile)                 \
 239       SymI_NeedsProto(unlockFile)               \
 240       SymI_HasProto(signal_handlers)            \
 241       SymI_HasProto(stg_sig_install)            \
 242       SymI_NeedsProto(nocldstop)
 243 #endif
 244
 245 #if defined (cygwin32_HOST_OS)
 246 #define RTS_MINGW_ONLY_SYMBOLS /**/
 247 /* Don't have the ability to read import libs / archives, so
 248  * we have to stupidly list a lot of what libcygwin.a
 249  * exports; sigh.
 250  */
 251 #define RTS_CYGWIN_ONLY_SYMBOLS                          \
 252       SymI_HasProto(regfree)                             \
 253       SymI_HasProto(regexec)                             \
 254       SymI_HasProto(regerror)                            \
 255       SymI_HasProto(regcomp)                             \
 256       SymI_HasProto(__errno)                             \
 257       SymI_HasProto(access)                              \
 258       SymI_HasProto(chmod)                               \
 259       SymI_HasProto(chdir)                               \
 260       SymI_HasProto(close)                               \
 261       SymI_HasProto(creat)                               \
 262       SymI_HasProto(dup)                                 \
 263       SymI_HasProto(dup2)                                \
 264       SymI_HasProto(fstat)                               \
 265       SymI_HasProto(fcntl)                               \
 266       SymI_HasProto(getcwd)                              \
 267       SymI_HasProto(getenv)                              \
 268       SymI_HasProto(lseek)                               \
 269       SymI_HasProto(open)                                \
 270       SymI_HasProto(fpathconf)                           \
 271       SymI_HasProto(pathconf)                            \
 272       SymI_HasProto(stat)                                \
 273       SymI_HasProto(pow)                                 \
 274       SymI_HasProto(tanh)                                \
 275       SymI_HasProto(cosh)                                \
 276       SymI_HasProto(sinh)                                \
 277       SymI_HasProto(atan)                                \
 278       SymI_HasProto(acos)                                \
 279       SymI_HasProto(asin)                                \
 280       SymI_HasProto(tan)                                 \
 281       SymI_HasProto(cos)                                 \
 282       SymI_HasProto(sin)                                 \
 283       SymI_HasProto(exp)                                 \
 284       SymI_HasProto(log)                                 \
 285       SymI_HasProto(sqrt)                                \
 286       SymI_HasProto(localtime_r)                         \
 287       SymI_HasProto(gmtime_r)                            \
 288       SymI_HasProto(mktime)                              \
 289       SymI_NeedsProto(_imp___tzname)                     \
 290       SymI_HasProto(gettimeofday)                        \
 291       SymI_HasProto(timezone)                            \
 292       SymI_HasProto(tcgetattr)                           \
 293       SymI_HasProto(tcsetattr)                           \
 294       SymI_HasProto(memcpy)                              \
 295       SymI_HasProto(memmove)                             \
 296       SymI_HasProto(realloc)                             \
 297       SymI_HasProto(malloc)                              \
 298       SymI_HasProto(free)                                \
 299       SymI_HasProto(fork)                                \
 300       SymI_HasProto(lstat)                               \
 301       SymI_HasProto(isatty)                              \
 302       SymI_HasProto(mkdir)                               \
 303       SymI_HasProto(opendir)                             \
 304       SymI_HasProto(readdir)                             \
 305       SymI_HasProto(rewinddir)                           \
 306       SymI_HasProto(closedir)                            \
 307       SymI_HasProto(link)                                \
 308       SymI_HasProto(mkfifo)                              \
 309       SymI_HasProto(pipe)                                \
 310       SymI_HasProto(read)                                \
 311       SymI_HasProto(rename)                              \
 312       SymI_HasProto(rmdir)                               \
 313       SymI_HasProto(select)                              \
 314       SymI_HasProto(system)                              \
 315       SymI_HasProto(write)                               \
 316       SymI_HasProto(strcmp)                              \
 317       SymI_HasProto(strcpy)                              \
 318       SymI_HasProto(strncpy)                             \
 319       SymI_HasProto(strerror)                            \
 320       SymI_HasProto(sigaddset)                           \
 321       SymI_HasProto(sigemptyset)                         \
 322       SymI_HasProto(sigprocmask)                         \
 323       SymI_HasProto(umask)                               \
 324       SymI_HasProto(uname)                               \
 325       SymI_HasProto(unlink)                              \
 326       SymI_HasProto(utime)                               \
 327       SymI_HasProto(waitpid)
 328
 329 #elif !defined(mingw32_HOST_OS)
 330 #define RTS_MINGW_ONLY_SYMBOLS /**/
 331 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 332 #else /* defined(mingw32_HOST_OS) */
 333 #define RTS_POSIX_ONLY_SYMBOLS  /**/
 334 #define RTS_CYGWIN_ONLY_SYMBOLS /**/
 335
 336 /* Extra syms gen'ed by mingw-2's gcc-3.2: */
 337 #if __GNUC__>=3
 338 #define RTS_MINGW_EXTRA_SYMS                    \
 339       SymI_NeedsProto(_imp____mb_cur_max)       \
 340       SymI_NeedsProto(_imp___pctype)
 341 #else
 342 #define RTS_MINGW_EXTRA_SYMS
 343 #endif
 344
 345 #if HAVE_GETTIMEOFDAY
 346 #define RTS_MINGW_GETTIMEOFDAY_SYM SymI_NeedsProto(gettimeofday)
 347 #else
 348 #define RTS_MINGW_GETTIMEOFDAY_SYM /**/
 349 #endif
 350
 351 /* These are statically linked from the mingw libraries into the ghc
 352    executable, so we have to employ this hack. */
 353 #define RTS_MINGW_ONLY_SYMBOLS                           \
 354       SymI_HasProto(asyncReadzh_fast)                    \
 355       SymI_HasProto(asyncWritezh_fast)                   \
 356       SymI_HasProto(asyncDoProczh_fast)                  \
 357       SymI_HasProto(memset)                              \
 358       SymI_HasProto(inet_ntoa)                           \
 359       SymI_HasProto(inet_addr)                           \
 360       SymI_HasProto(htonl)                               \
 361       SymI_HasProto(recvfrom)                            \
 362       SymI_HasProto(listen)                              \
 363       SymI_HasProto(bind)                                \
 364       SymI_HasProto(shutdown)                            \
 365       SymI_HasProto(connect)                             \
 366       SymI_HasProto(htons)                               \
 367       SymI_HasProto(ntohs)                               \
 368       SymI_HasProto(getservbyname)                       \
 369       SymI_HasProto(getservbyport)                       \
 370       SymI_HasProto(getprotobynumber)                    \
 371       SymI_HasProto(getprotobyname)                      \
 372       SymI_HasProto(gethostbyname)                       \
 373       SymI_HasProto(gethostbyaddr)                       \
 374       SymI_HasProto(gethostname)                         \
 375       SymI_HasProto(strcpy)                              \
 376       SymI_HasProto(strncpy)                             \
 377       SymI_HasProto(abort)                               \
 378       SymI_NeedsProto(_alloca)                           \
 379       SymI_NeedsProto(isxdigit)                          \
 380       SymI_NeedsProto(isupper)                           \
 381       SymI_NeedsProto(ispunct)                           \
 382       SymI_NeedsProto(islower)                           \
 383       SymI_NeedsProto(isspace)                           \
 384       SymI_NeedsProto(isprint)                           \
 385       SymI_NeedsProto(isdigit)                           \
 386       SymI_NeedsProto(iscntrl)                           \
 387       SymI_NeedsProto(isalpha)                           \
 388       SymI_NeedsProto(isalnum)                           \
 389       SymI_HasProto(strcmp)                              \
 390       SymI_HasProto(memmove)                             \
 391       SymI_HasProto(realloc)                             \
 392       SymI_HasProto(malloc)                              \
 393       SymI_HasProto(pow)                                 \
 394       SymI_HasProto(tanh)                                \
 395       SymI_HasProto(cosh)                                \
 396       SymI_HasProto(sinh)                                \
 397       SymI_HasProto(atan)                                \
 398       SymI_HasProto(acos)                                \
 399       SymI_HasProto(asin)                                \
 400       SymI_HasProto(tan)                                 \
 401       SymI_HasProto(cos)                                 \
 402       SymI_HasProto(sin)                                 \
 403       SymI_HasProto(exp)                                 \
 404       SymI_HasProto(log)                                 \
 405       SymI_HasProto(sqrt)                                \
 406       SymI_HasProto(powf)                                \
 407       SymI_HasProto(tanhf)                               \
 408       SymI_HasProto(coshf)                               \
 409       SymI_HasProto(sinhf)                               \
 410       SymI_HasProto(atanf)                               \
 411       SymI_HasProto(acosf)                               \
 412       SymI_HasProto(asinf)                               \
 413       SymI_HasProto(tanf)                                \
 414       SymI_HasProto(cosf)                                \
 415       SymI_HasProto(sinf)                                \
 416       SymI_HasProto(expf)                                \
 417       SymI_HasProto(logf)                                \
 418       SymI_HasProto(sqrtf)                               \
 419       SymI_HasProto(memcpy)                              \
 420       SymI_HasProto(rts_InstallConsoleEvent)             \
 421       SymI_HasProto(rts_ConsoleHandlerDone)              \
 422       SymI_NeedsProto(mktime)                            \
 423       SymI_NeedsProto(_imp___timezone)                   \
 424       SymI_NeedsProto(_imp___tzname)                     \
 425       SymI_NeedsProto(_imp__tzname)                      \
 426       SymI_NeedsProto(_imp___iob)                        \
 427       SymI_NeedsProto(_imp___osver)                      \
 428       SymI_NeedsProto(localtime)                         \
 429       SymI_NeedsProto(gmtime)                            \
 430       SymI_NeedsProto(opendir)                           \
 431       SymI_NeedsProto(readdir)                           \
 432       SymI_NeedsProto(rewinddir)                         \
 433       RTS_MINGW_EXTRA_SYMS                               \
 434       RTS_MINGW_GETTIMEOFDAY_SYM                         \
 435       SymI_NeedsProto(closedir)
 436 #endif
 437
 438 #if defined(darwin_TARGET_OS) && HAVE_PRINTF_LDBLSTUB
 439 #define RTS_DARWIN_ONLY_SYMBOLS                             \
 440      SymI_NeedsProto(asprintf$LDBLStub)                     \
 441      SymI_NeedsProto(err$LDBLStub)                          \
 442      SymI_NeedsProto(errc$LDBLStub)                         \
 443      SymI_NeedsProto(errx$LDBLStub)                         \
 444      SymI_NeedsProto(fprintf$LDBLStub)                      \
 445      SymI_NeedsProto(fscanf$LDBLStub)                       \
 446      SymI_NeedsProto(fwprintf$LDBLStub)                     \
 447      SymI_NeedsProto(fwscanf$LDBLStub)                      \
 448      SymI_NeedsProto(printf$LDBLStub)                       \
 449      SymI_NeedsProto(scanf$LDBLStub)                        \
 450      SymI_NeedsProto(snprintf$LDBLStub)                     \
 451      SymI_NeedsProto(sprintf$LDBLStub)                      \
 452      SymI_NeedsProto(sscanf$LDBLStub)                       \
 453      SymI_NeedsProto(strtold$LDBLStub)                      \
 454      SymI_NeedsProto(swprintf$LDBLStub)                     \
 455      SymI_NeedsProto(swscanf$LDBLStub)                      \
 456      SymI_NeedsProto(syslog$LDBLStub)                       \
 457      SymI_NeedsProto(vasprintf$LDBLStub)                    \
 458      SymI_NeedsProto(verr$LDBLStub)                         \
 459      SymI_NeedsProto(verrc$LDBLStub)                        \
 460      SymI_NeedsProto(verrx$LDBLStub)                        \
 461      SymI_NeedsProto(vfprintf$LDBLStub)                     \
 462      SymI_NeedsProto(vfscanf$LDBLStub)                      \
 463      SymI_NeedsProto(vfwprintf$LDBLStub)                    \
 464      SymI_NeedsProto(vfwscanf$LDBLStub)                     \
 465      SymI_NeedsProto(vprintf$LDBLStub)                      \
 466      SymI_NeedsProto(vscanf$LDBLStub)                       \
 467      SymI_NeedsProto(vsnprintf$LDBLStub)                    \
 468      SymI_NeedsProto(vsprintf$LDBLStub)                     \
 469      SymI_NeedsProto(vsscanf$LDBLStub)                      \
 470      SymI_NeedsProto(vswprintf$LDBLStub)                    \
 471      SymI_NeedsProto(vswscanf$LDBLStub)                     \
 472      SymI_NeedsProto(vsyslog$LDBLStub)                      \
 473      SymI_NeedsProto(vwarn$LDBLStub)                        \
 474      SymI_NeedsProto(vwarnc$LDBLStub)                       \
 475      SymI_NeedsProto(vwarnx$LDBLStub)                       \
 476      SymI_NeedsProto(vwprintf$LDBLStub)                     \
 477      SymI_NeedsProto(vwscanf$LDBLStub)                      \
 478      SymI_NeedsProto(warn$LDBLStub)                         \
 479      SymI_NeedsProto(warnc$LDBLStub)                        \
 480      SymI_NeedsProto(warnx$LDBLStub)                        \
 481      SymI_NeedsProto(wcstold$LDBLStub)                      \
 482      SymI_NeedsProto(wprintf$LDBLStub)                      \
 483      SymI_NeedsProto(wscanf$LDBLStub)
 484 #else
 485 #define RTS_DARWIN_ONLY_SYMBOLS
 486 #endif
 487
 488 #ifndef SMP
 489 # define MAIN_CAP_SYM SymI_HasProto(MainCapability)
 490 #else
 491 # define MAIN_CAP_SYM
 492 #endif
 493
 494 #if !defined(mingw32_HOST_OS)
 495 #define RTS_USER_SIGNALS_SYMBOLS \
 496    SymI_HasProto(setIOManagerPipe) \
 497    SymI_NeedsProto(blockUserSignals) \
 498    SymI_NeedsProto(unblockUserSignals)
 499 #else
 500 #define RTS_USER_SIGNALS_SYMBOLS     \
 501    SymI_HasProto(sendIOManagerEvent) \
 502    SymI_HasProto(readIOManagerEvent) \
 503    SymI_HasProto(getIOManagerEvent)  \
 504    SymI_HasProto(console_handler)
 505 #endif
 506
 507 #define RTS_LIBFFI_SYMBOLS                                  \
 508      SymE_NeedsProto(ffi_prep_cif)                          \
 509      SymE_NeedsProto(ffi_call)                              \
 510      SymE_NeedsProto(ffi_type_void)                         \
 511      SymE_NeedsProto(ffi_type_float)                        \
 512      SymE_NeedsProto(ffi_type_double)                       \
 513      SymE_NeedsProto(ffi_type_sint64)                       \
 514      SymE_NeedsProto(ffi_type_uint64)                       \
 515      SymE_NeedsProto(ffi_type_sint32)                       \
 516      SymE_NeedsProto(ffi_type_uint32)                       \
 517      SymE_NeedsProto(ffi_type_sint16)                       \
 518      SymE_NeedsProto(ffi_type_uint16)                       \
 519      SymE_NeedsProto(ffi_type_sint8)                        \
 520      SymE_NeedsProto(ffi_type_uint8)                        \
 521      SymE_NeedsProto(ffi_type_pointer)
 522
 523 #ifdef TABLES_NEXT_TO_CODE
 524 #define RTS_RET_SYMBOLS /* nothing */
 525 #else
 526 #define RTS_RET_SYMBOLS                                 \
 527       SymI_HasProto(stg_enter_ret)                      \
 528       SymI_HasProto(stg_gc_fun_ret)                     \
 529       SymI_HasProto(stg_ap_v_ret)                       \
 530       SymI_HasProto(stg_ap_f_ret)                       \
 531       SymI_HasProto(stg_ap_d_ret)                       \
 532       SymI_HasProto(stg_ap_l_ret)                       \
 533       SymI_HasProto(stg_ap_n_ret)                       \
 534       SymI_HasProto(stg_ap_p_ret)                       \
 535       SymI_HasProto(stg_ap_pv_ret)                      \
 536       SymI_HasProto(stg_ap_pp_ret)                      \
 537       SymI_HasProto(stg_ap_ppv_ret)                     \
 538       SymI_HasProto(stg_ap_ppp_ret)                     \
 539       SymI_HasProto(stg_ap_pppv_ret)                    \
 540       SymI_HasProto(stg_ap_pppp_ret)                    \
 541       SymI_HasProto(stg_ap_ppppp_ret)                   \
 542       SymI_HasProto(stg_ap_pppppp_ret)
 543 #endif
 544
 545 /* On Windows, we link libgmp.a statically into libHSrts.dll */
 546 #ifdef mingw32_HOST_OS
 547 #define GMP_SYMS                                        \
 548       SymI_HasProto(__gmpz_cmp)                         \
 549       SymI_HasProto(__gmpz_cmp_si)                      \
 550       SymI_HasProto(__gmpz_cmp_ui)                      \
 551       SymI_HasProto(__gmpz_get_si)                      \
 552       SymI_HasProto(__gmpz_get_ui)
 553 #else
 554 #define GMP_SYMS                                        \
 555       SymE_HasProto(__gmpz_cmp)                         \
 556       SymE_HasProto(__gmpz_cmp_si)                      \
 557       SymE_HasProto(__gmpz_cmp_ui)                      \
 558       SymE_HasProto(__gmpz_get_si)                      \
 559       SymE_HasProto(__gmpz_get_ui)
 560 #endif
 561
 562 #define RTS_SYMBOLS                                     \
 563       Maybe_Stable_Names                                \
 564       SymI_HasProto(StgReturn)                          \
 565       SymI_HasProto(stg_enter_info)                     \
 566       SymI_HasProto(stg_gc_void_info)                   \
 567       SymI_HasProto(__stg_gc_enter_1)                   \
 568       SymI_HasProto(stg_gc_noregs)                      \
 569       SymI_HasProto(stg_gc_unpt_r1_info)                \
 570       SymI_HasProto(stg_gc_unpt_r1)                     \
 571       SymI_HasProto(stg_gc_unbx_r1_info)                \
 572       SymI_HasProto(stg_gc_unbx_r1)                     \
 573       SymI_HasProto(stg_gc_f1_info)                     \
 574       SymI_HasProto(stg_gc_f1)                          \
 575       SymI_HasProto(stg_gc_d1_info)                     \
 576       SymI_HasProto(stg_gc_d1)                          \
 577       SymI_HasProto(stg_gc_l1_info)                     \
 578       SymI_HasProto(stg_gc_l1)                          \
 579       SymI_HasProto(__stg_gc_fun)                       \
 580       SymI_HasProto(stg_gc_fun_info)                    \
 581       SymI_HasProto(stg_gc_gen)                         \
 582       SymI_HasProto(stg_gc_gen_info)                    \
 583       SymI_HasProto(stg_gc_gen_hp)                      \
 584       SymI_HasProto(stg_gc_ut)                          \
 585       SymI_HasProto(stg_gen_yield)                      \
 586       SymI_HasProto(stg_yield_noregs)                   \
 587       SymI_HasProto(stg_yield_to_interpreter)           \
 588       SymI_HasProto(stg_gen_block)                      \
 589       SymI_HasProto(stg_block_noregs)                   \
 590       SymI_HasProto(stg_block_1)                        \
 591       SymI_HasProto(stg_block_takemvar)                 \
 592       SymI_HasProto(stg_block_putmvar)                  \
 593       MAIN_CAP_SYM                                      \
 594       SymI_HasProto(MallocFailHook)                     \
 595       SymI_HasProto(OnExitHook)                         \
 596       SymI_HasProto(OutOfHeapHook)                      \
 597       SymI_HasProto(StackOverflowHook)                  \
 598       SymI_HasProto(__encodeDouble)                     \
 599       SymI_HasProto(__encodeFloat)                      \
 600       SymI_HasProto(addDLL)                             \
 601       GMP_SYMS                                          \
 602       SymI_HasProto(__int_encodeDouble)                 \
 603       SymI_HasProto(__word_encodeDouble)                \
 604       SymI_HasProto(__2Int_encodeDouble)                \
 605       SymI_HasProto(__int_encodeFloat)                  \
 606       SymI_HasProto(__word_encodeFloat)                 \
 607       SymI_HasProto(andIntegerzh_fast)                  \
 608       SymI_HasProto(atomicallyzh_fast)                  \
 609       SymI_HasProto(barf)                               \
 610       SymI_HasProto(debugBelch)                         \
 611       SymI_HasProto(errorBelch)                         \
 612       SymI_HasProto(asyncExceptionsBlockedzh_fast)      \
 613       SymI_HasProto(blockAsyncExceptionszh_fast)        \
 614       SymI_HasProto(catchzh_fast)                       \
 615       SymI_HasProto(catchRetryzh_fast)                  \
 616       SymI_HasProto(catchSTMzh_fast)                    \
 617       SymI_HasProto(checkzh_fast)                       \
 618       SymI_HasProto(closure_flags)                      \
 619       SymI_HasProto(cmp_thread)                         \
 620       SymI_HasProto(cmpIntegerzh_fast)                  \
 621       SymI_HasProto(cmpIntegerIntzh_fast)               \
 622       SymI_HasProto(complementIntegerzh_fast)           \
 623       SymI_HasProto(createAdjustor)                     \
 624       SymI_HasProto(decodeDoublezh_fast)                \
 625       SymI_HasProto(decodeFloatzh_fast)                 \
 626       SymI_HasProto(decodeDoublezu2Intzh_fast)          \
 627       SymI_HasProto(decodeFloatzuIntzh_fast)            \
 628       SymI_HasProto(defaultsHook)                       \
 629       SymI_HasProto(delayzh_fast)                       \
 630       SymI_HasProto(deRefWeakzh_fast)                   \
 631       SymI_HasProto(deRefStablePtrzh_fast)              \
 632       SymI_HasProto(dirty_MUT_VAR)                      \
 633       SymI_HasProto(divExactIntegerzh_fast)             \
 634       SymI_HasProto(divModIntegerzh_fast)               \
 635       SymI_HasProto(forkzh_fast)                        \
 636       SymI_HasProto(forkOnzh_fast)                      \
 637       SymI_HasProto(forkProcess)                        \
 638       SymI_HasProto(forkOS_createThread)                \
 639       SymI_HasProto(freeHaskellFunctionPtr)             \
 640       SymI_HasProto(freeStablePtr)                      \
 641       SymI_HasProto(getOrSetTypeableStore)              \
 642       SymI_HasProto(getOrSetSignalHandlerStore)         \
 643       SymI_HasProto(gcdIntegerzh_fast)                  \
 644       SymI_HasProto(gcdIntegerIntzh_fast)               \
 645       SymI_HasProto(gcdIntzh_fast)                      \
 646       SymI_HasProto(genSymZh)                           \
 647       SymI_HasProto(genericRaise)                       \
 648       SymI_HasProto(getProgArgv)                        \
 649       SymI_HasProto(getFullProgArgv)                    \
 650       SymI_HasProto(getStablePtr)                       \
 651       SymI_HasProto(hs_init)                            \
 652       SymI_HasProto(hs_exit)                            \
 653       SymI_HasProto(hs_set_argv)                        \
 654       SymI_HasProto(hs_add_root)                        \
 655       SymI_HasProto(hs_perform_gc)                      \
 656       SymI_HasProto(hs_free_stable_ptr)                 \
 657       SymI_HasProto(hs_free_fun_ptr)                    \
 658       SymI_HasProto(hs_hpc_rootModule)                  \
 659       SymI_HasProto(hs_hpc_module)                      \
 660       SymI_HasProto(initLinker)                         \
 661       SymI_HasProto(unpackClosurezh_fast)               \
 662       SymI_HasProto(getApStackValzh_fast)               \
 663       SymI_HasProto(getSparkzh_fast)                    \
 664       SymI_HasProto(int2Integerzh_fast)                 \
 665       SymI_HasProto(integer2Intzh_fast)                 \
 666       SymI_HasProto(integer2Wordzh_fast)                \
 667       SymI_HasProto(isCurrentThreadBoundzh_fast)        \
 668       SymI_HasProto(isDoubleDenormalized)               \
 669       SymI_HasProto(isDoubleInfinite)                   \
 670       SymI_HasProto(isDoubleNaN)                        \
 671       SymI_HasProto(isDoubleNegativeZero)               \
 672       SymI_HasProto(isEmptyMVarzh_fast)                 \
 673       SymI_HasProto(isFloatDenormalized)                \
 674       SymI_HasProto(isFloatInfinite)                    \
 675       SymI_HasProto(isFloatNaN)                         \
 676       SymI_HasProto(isFloatNegativeZero)                \
 677       SymI_HasProto(killThreadzh_fast)                  \
 678       SymI_HasProto(loadObj)                            \
 679       SymI_HasProto(insertStableSymbol)                 \
 680       SymI_HasProto(insertSymbol)                       \
 681       SymI_HasProto(lookupSymbol)                       \
 682       SymI_HasProto(makeStablePtrzh_fast)               \
 683       SymI_HasProto(minusIntegerzh_fast)                \
 684       SymI_HasProto(mkApUpd0zh_fast)                    \
 685       SymI_HasProto(myThreadIdzh_fast)                  \
 686       SymI_HasProto(labelThreadzh_fast)                 \
 687       SymI_HasProto(newArrayzh_fast)                    \
 688       SymI_HasProto(newBCOzh_fast)                      \
 689       SymI_HasProto(newByteArrayzh_fast)                \
 690       SymI_HasProto_redirect(newCAF, newDynCAF)         \
 691       SymI_HasProto(newMVarzh_fast)                     \
 692       SymI_HasProto(newMutVarzh_fast)                   \
 693       SymI_HasProto(newTVarzh_fast)                     \
 694       SymI_HasProto(noDuplicatezh_fast)                 \
 695       SymI_HasProto(atomicModifyMutVarzh_fast)          \
 696       SymI_HasProto(newPinnedByteArrayzh_fast)          \
 697       SymI_HasProto(newAlignedPinnedByteArrayzh_fast)   \
 698       SymI_HasProto(newSpark)                           \
 699       SymI_HasProto(orIntegerzh_fast)                   \
 700       SymI_HasProto(performGC)                          \
 701       SymI_HasProto(performMajorGC)                     \
 702       SymI_HasProto(plusIntegerzh_fast)                 \
 703       SymI_HasProto(prog_argc)                          \
 704       SymI_HasProto(prog_argv)                          \
 705       SymI_HasProto(putMVarzh_fast)                     \
 706       SymI_HasProto(quotIntegerzh_fast)                 \
 707       SymI_HasProto(quotRemIntegerzh_fast)              \
 708       SymI_HasProto(raisezh_fast)                       \
 709       SymI_HasProto(raiseIOzh_fast)                     \
 710       SymI_HasProto(readTVarzh_fast)                    \
 711       SymI_HasProto(readTVarIOzh_fast)                  \
 712       SymI_HasProto(remIntegerzh_fast)                  \
 713       SymI_HasProto(resetNonBlockingFd)                 \
 714       SymI_HasProto(resumeThread)                       \
 715       SymI_HasProto(resolveObjs)                        \
 716       SymI_HasProto(retryzh_fast)                       \
 717       SymI_HasProto(rts_apply)                          \
 718       SymI_HasProto(rts_checkSchedStatus)               \
 719       SymI_HasProto(rts_eval)                           \
 720       SymI_HasProto(rts_evalIO)                         \
 721       SymI_HasProto(rts_evalLazyIO)                     \
 722       SymI_HasProto(rts_evalStableIO)                   \
 723       SymI_HasProto(rts_eval_)                          \
 724       SymI_HasProto(rts_getBool)                        \
 725       SymI_HasProto(rts_getChar)                        \
 726       SymI_HasProto(rts_getDouble)                      \
 727       SymI_HasProto(rts_getFloat)                       \
 728       SymI_HasProto(rts_getInt)                         \
 729       SymI_HasProto(rts_getInt8)                        \
 730       SymI_HasProto(rts_getInt16)                       \
 731       SymI_HasProto(rts_getInt32)                       \
 732       SymI_HasProto(rts_getInt64)                       \
 733       SymI_HasProto(rts_getPtr)                         \
 734       SymI_HasProto(rts_getFunPtr)                      \
 735       SymI_HasProto(rts_getStablePtr)                   \
 736       SymI_HasProto(rts_getThreadId)                    \
 737       SymI_HasProto(rts_getWord)                        \
 738       SymI_HasProto(rts_getWord8)                       \
 739       SymI_HasProto(rts_getWord16)                      \
 740       SymI_HasProto(rts_getWord32)                      \
 741       SymI_HasProto(rts_getWord64)                      \
 742       SymI_HasProto(rts_lock)                           \
 743       SymI_HasProto(rts_mkBool)                         \
 744       SymI_HasProto(rts_mkChar)                         \
 745       SymI_HasProto(rts_mkDouble)                       \
 746       SymI_HasProto(rts_mkFloat)                        \
 747       SymI_HasProto(rts_mkInt)                          \
 748       SymI_HasProto(rts_mkInt8)                         \
 749       SymI_HasProto(rts_mkInt16)                        \
 750       SymI_HasProto(rts_mkInt32)                        \
 751       SymI_HasProto(rts_mkInt64)                        \
 752       SymI_HasProto(rts_mkPtr)                          \
 753       SymI_HasProto(rts_mkFunPtr)                       \
 754       SymI_HasProto(rts_mkStablePtr)                    \
 755       SymI_HasProto(rts_mkString)                       \
 756       SymI_HasProto(rts_mkWord)                         \
 757       SymI_HasProto(rts_mkWord8)                        \
 758       SymI_HasProto(rts_mkWord16)                       \
 759       SymI_HasProto(rts_mkWord32)                       \
 760       SymI_HasProto(rts_mkWord64)                       \
 761       SymI_HasProto(rts_unlock)                         \
 762       SymI_HasProto(rtsSupportsBoundThreads)            \
 763       SymI_HasProto(__hscore_get_saved_termios)         \
 764       SymI_HasProto(__hscore_set_saved_termios)         \
 765       SymI_HasProto(setProgArgv)                        \
 766       SymI_HasProto(startupHaskell)                     \
 767       SymI_HasProto(shutdownHaskell)                    \
 768       SymI_HasProto(shutdownHaskellAndExit)             \
 769       SymI_HasProto(stable_ptr_table)                   \
 770       SymI_HasProto(stackOverflow)                      \
 771       SymI_HasProto(stg_CAF_BLACKHOLE_info)             \
 772       SymI_HasProto(__stg_EAGER_BLACKHOLE_info)         \
 773       SymI_HasProto(awakenBlockedQueue)                 \
 774       SymI_HasProto(startTimer)                         \
 775       SymI_HasProto(stg_CHARLIKE_closure)               \
 776       SymI_HasProto(stg_MVAR_CLEAN_info)                \
 777       SymI_HasProto(stg_MVAR_DIRTY_info)                \
 778       SymI_HasProto(stg_IND_STATIC_info)                \
 779       SymI_HasProto(stg_INTLIKE_closure)                \
 780       SymI_HasProto(stg_MUT_ARR_PTRS_DIRTY_info)        \
 781       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN_info)       \
 782       SymI_HasProto(stg_MUT_ARR_PTRS_FROZEN0_info)      \
 783       SymI_HasProto(stg_WEAK_info)                      \
 784       SymI_HasProto(stg_ap_v_info)                      \
 785       SymI_HasProto(stg_ap_f_info)                      \
 786       SymI_HasProto(stg_ap_d_info)                      \
 787       SymI_HasProto(stg_ap_l_info)                      \
 788       SymI_HasProto(stg_ap_n_info)                      \
 789       SymI_HasProto(stg_ap_p_info)                      \
 790       SymI_HasProto(stg_ap_pv_info)                     \
 791       SymI_HasProto(stg_ap_pp_info)                     \
 792       SymI_HasProto(stg_ap_ppv_info)                    \
 793       SymI_HasProto(stg_ap_ppp_info)                    \
 794       SymI_HasProto(stg_ap_pppv_info)                   \
 795       SymI_HasProto(stg_ap_pppp_info)                   \
 796       SymI_HasProto(stg_ap_ppppp_info)                  \
 797       SymI_HasProto(stg_ap_pppppp_info)                 \
 798       SymI_HasProto(stg_ap_0_fast)                      \
 799       SymI_HasProto(stg_ap_v_fast)                      \
 800       SymI_HasProto(stg_ap_f_fast)                      \
 801       SymI_HasProto(stg_ap_d_fast)                      \
 802       SymI_HasProto(stg_ap_l_fast)                      \
 803       SymI_HasProto(stg_ap_n_fast)                      \
 804       SymI_HasProto(stg_ap_p_fast)                      \
 805       SymI_HasProto(stg_ap_pv_fast)                     \
 806       SymI_HasProto(stg_ap_pp_fast)                     \
 807       SymI_HasProto(stg_ap_ppv_fast)                    \
 808       SymI_HasProto(stg_ap_ppp_fast)                    \
 809       SymI_HasProto(stg_ap_pppv_fast)                   \
 810       SymI_HasProto(stg_ap_pppp_fast)                   \
 811       SymI_HasProto(stg_ap_ppppp_fast)                  \
 812       SymI_HasProto(stg_ap_pppppp_fast)                 \
 813       SymI_HasProto(stg_ap_1_upd_info)                  \
 814       SymI_HasProto(stg_ap_2_upd_info)                  \
 815       SymI_HasProto(stg_ap_3_upd_info)                  \
 816       SymI_HasProto(stg_ap_4_upd_info)                  \
 817       SymI_HasProto(stg_ap_5_upd_info)                  \
 818       SymI_HasProto(stg_ap_6_upd_info)                  \
 819       SymI_HasProto(stg_ap_7_upd_info)                  \
 820       SymI_HasProto(stg_exit)                           \
 821       SymI_HasProto(stg_sel_0_upd_info)                 \
 822       SymI_HasProto(stg_sel_10_upd_info)                \
 823       SymI_HasProto(stg_sel_11_upd_info)                \
 824       SymI_HasProto(stg_sel_12_upd_info)                \
 825       SymI_HasProto(stg_sel_13_upd_info)                \
 826       SymI_HasProto(stg_sel_14_upd_info)                \
 827       SymI_HasProto(stg_sel_15_upd_info)                \
 828       SymI_HasProto(stg_sel_1_upd_info)                 \
 829       SymI_HasProto(stg_sel_2_upd_info)                 \
 830       SymI_HasProto(stg_sel_3_upd_info)                 \
 831       SymI_HasProto(stg_sel_4_upd_info)                 \
 832       SymI_HasProto(stg_sel_5_upd_info)                 \
 833       SymI_HasProto(stg_sel_6_upd_info)                 \
 834       SymI_HasProto(stg_sel_7_upd_info)                 \
 835       SymI_HasProto(stg_sel_8_upd_info)                 \
 836       SymI_HasProto(stg_sel_9_upd_info)                 \
 837       SymI_HasProto(stg_upd_frame_info)                 \
 838       SymI_HasProto(suspendThread)                      \
 839       SymI_HasProto(takeMVarzh_fast)                    \
 840       SymI_HasProto(threadStatuszh_fast)                \
 841       SymI_HasProto(timesIntegerzh_fast)                \
 842       SymI_HasProto(tryPutMVarzh_fast)                  \
 843       SymI_HasProto(tryTakeMVarzh_fast)                 \
 844       SymI_HasProto(unblockAsyncExceptionszh_fast)      \
 845       SymI_HasProto(unloadObj)                          \
 846       SymI_HasProto(unsafeThawArrayzh_fast)             \
 847       SymI_HasProto(waitReadzh_fast)                    \
 848       SymI_HasProto(waitWritezh_fast)                   \
 849       SymI_HasProto(word2Integerzh_fast)                \
 850       SymI_HasProto(writeTVarzh_fast)                   \
 851       SymI_HasProto(xorIntegerzh_fast)                  \
 852       SymI_HasProto(yieldzh_fast)                       \
 853       SymI_NeedsProto(stg_interp_constr_entry)          \
 854       SymI_HasProto(allocateExec)                       \
 855       SymI_HasProto(freeExec)                           \
 856       SymI_HasProto(getAllocations)                     \
 857       SymI_HasProto(revertCAFs)                         \
 858       SymI_HasProto(RtsFlags)                           \
 859       SymI_NeedsProto(rts_breakpoint_io_action)         \
 860       SymI_NeedsProto(rts_stop_next_breakpoint)         \
 861       SymI_NeedsProto(rts_stop_on_exception)            \
 862       SymI_HasProto(stopTimer)                          \
 863       SymI_HasProto(n_capabilities)                     \
 864       SymI_HasProto(traceCcszh_fast)                    \
 865       RTS_USER_SIGNALS_SYMBOLS
 866
 867 #ifdef SUPPORT_LONG_LONGS
 868 #define RTS_LONG_LONG_SYMS                              \
 869       SymI_HasProto(int64ToIntegerzh_fast)              \
 870       SymI_HasProto(word64ToIntegerzh_fast)
 871 #else
 872 #define RTS_LONG_LONG_SYMS /* nothing */
 873 #endif
 874
 875 // 64-bit support functions in libgcc.a
 876 #if defined(__GNUC__) && SIZEOF_VOID_P <= 4
 877 #define RTS_LIBGCC_SYMBOLS                             \
 878       SymI_NeedsProto(__divdi3)                        \
 879       SymI_NeedsProto(__udivdi3)                       \
 880       SymI_NeedsProto(__moddi3)                        \
 881       SymI_NeedsProto(__umoddi3)                       \
 882       SymI_NeedsProto(__muldi3)                        \
 883       SymI_NeedsProto(__ashldi3)                       \
 884       SymI_NeedsProto(__ashrdi3)                       \
 885       SymI_NeedsProto(__lshrdi3)                       \
 886       SymI_NeedsProto(__eprintf)
 887 #elif defined(ia64_HOST_ARCH)
 888 #define RTS_LIBGCC_SYMBOLS                              \
 889       SymI_NeedsProto(__divdi3)                         \
 890       SymI_NeedsProto(__udivdi3)                        \
 891       SymI_NeedsProto(__moddi3)                         \
 892       SymI_NeedsProto(__umoddi3)                        \
 893       SymI_NeedsProto(__divsf3)                         \
 894       SymI_NeedsProto(__divdf3)
 895 #else
 896 #define RTS_LIBGCC_SYMBOLS
 897 #endif
 898
 899 #if defined(darwin_HOST_OS) && defined(powerpc_HOST_ARCH)
 900       // Symbols that don't have a leading underscore
 901       // on Mac OS X. They have to receive special treatment,
 902       // see machoInitSymbolsWithoutUnderscore()
 903 #define RTS_MACHO_NOUNDERLINE_SYMBOLS           \
 904       SymI_NeedsProto(saveFP)                           \
 905       SymI_NeedsProto(restFP)
 906 #endif
 907
 908 /* entirely bogus claims about types of these symbols */
 909 #define SymI_NeedsProto(vvv)  extern void vvv(void);
 910 #if defined(__PIC__) && defined(mingw32_TARGET_OS)
 911 #define SymE_HasProto(vvv)    SymE_HasProto(vvv);
 912 #define SymE_NeedsProto(vvv)    extern void _imp__ ## vvv (void);
 913 #else
 914 #define SymE_NeedsProto(vvv)  SymI_NeedsProto(vvv);
 915 #define SymE_HasProto(vvv)    SymI_HasProto(vvv)
 916 #endif
 917 #define SymI_HasProto(vvv) /**/
 918 #define SymI_HasProto_redirect(vvv,xxx) /**/
 919 RTS_SYMBOLS
 920 RTS_RET_SYMBOLS
 921 RTS_LONG_LONG_SYMS
 922 RTS_POSIX_ONLY_SYMBOLS
 923 RTS_MINGW_ONLY_SYMBOLS
 924 RTS_CYGWIN_ONLY_SYMBOLS
 925 RTS_DARWIN_ONLY_SYMBOLS
 926 RTS_LIBGCC_SYMBOLS
 927 RTS_LIBFFI_SYMBOLS
 928 #undef SymI_NeedsProto
 929 #undef SymI_HasProto
 930 #undef SymI_HasProto_redirect
 931 #undef SymE_HasProto
 932 #undef SymE_NeedsProto
 933
 934 #ifdef LEADING_UNDERSCORE
 935 #define MAYBE_LEADING_UNDERSCORE_STR(s) ("_" s)
 936 #else
 937 #define MAYBE_LEADING_UNDERSCORE_STR(s) (s)
 938 #endif
 939
 940 #define SymI_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
 941                     (void*)(&(vvv)) },
 942 #define SymE_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
 943             (void*)DLL_IMPORT_DATA_REF(vvv) },
 944
 945 #define SymI_NeedsProto(vvv) SymI_HasProto(vvv)
 946 #define SymE_NeedsProto(vvv) SymE_HasProto(vvv)
 947
 948 // SymI_HasProto_redirect allows us to redirect references to one symbol to
 949 // another symbol.  See newCAF/newDynCAF for an example.
 950 #define SymI_HasProto_redirect(vvv,xxx) \
 951     { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
 952       (void*)(&(xxx)) },
 953
 954 static RtsSymbolVal rtsSyms[] = {
 955       RTS_SYMBOLS
 956       RTS_RET_SYMBOLS
 957       RTS_LONG_LONG_SYMS
 958       RTS_POSIX_ONLY_SYMBOLS
 959       RTS_MINGW_ONLY_SYMBOLS
 960       RTS_CYGWIN_ONLY_SYMBOLS
 961       RTS_DARWIN_ONLY_SYMBOLS
 962       RTS_LIBGCC_SYMBOLS
 963       RTS_LIBFFI_SYMBOLS
 964 #if defined(darwin_HOST_OS) && defined(i386_HOST_ARCH)
 965       // dyld stub code contains references to this,
 966       // but it should never be called because we treat
 967       // lazy pointers as nonlazy.
 968       { "dyld_stub_binding_helper", (void*)0xDEADBEEF },
 969 #endif
 970       { 0, 0 } /* sentinel */
 971 };
 972
 973
 974
 975 /* -----------------------------------------------------------------------------
 976  * Insert symbols into hash tables, checking for duplicates.
 977  */
 978
 979 static void ghciInsertStrHashTable ( char* obj_name,
 980                                      HashTable *table,
 981                                      char* key,
 982                                      void *data
 983                                    )
 984 {
 985    if (lookupHashTable(table, (StgWord)key) == NULL)
 986    {
 987       insertStrHashTable(table, (StgWord)key, data);
 988       return;
 989    }
 990    debugBelch(
 991       "\n\n"
 992       "GHCi runtime linker: fatal error: I found a duplicate definition for symbol\n"
 993       "   %s\n"
 994       "whilst processing object file\n"
 995       "   %s\n"
 996       "This could be caused by:\n"
 997       "   * Loading two different object files which export the same symbol\n"
 998       "   * Specifying the same object file twice on the GHCi command line\n"
 999       "   * An incorrect `package.conf' entry, causing some object to be\n"
1000       "     loaded twice.\n"
1001       "GHCi cannot safely continue in this situation.  Exiting now.  Sorry.\n"
1002       "\n",
1003       (char*)key,
1004       obj_name
1005    );
1006    exit(1);
1007 }
1008 /* -----------------------------------------------------------------------------
1009  * initialize the object linker
1010  */
1011
1012
1013 static int linker_init_done = 0 ;
1014
1015 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1016 static void *dl_prog_handle;
1017 #endif
1018
1019 void
1020 initLinker( void )
1021 {
1022     RtsSymbolVal *sym;
1023
1024     /* Make initLinker idempotent, so we can call it
1025        before evey relevant operation; that means we
1026        don't need to initialise the linker separately */
1027     if (linker_init_done == 1) { return; } else {
1028       linker_init_done = 1;
1029     }
1030
1031     stablehash = allocStrHashTable();
1032     symhash = allocStrHashTable();
1033
1034     /* populate the symbol table with stuff from the RTS */
1035     for (sym = rtsSyms; sym->lbl != NULL; sym++) {
1036         ghciInsertStrHashTable("(GHCi built-in symbols)",
1037                                symhash, sym->lbl, sym->addr);
1038     }
1039 #   if defined(OBJFORMAT_MACHO) && defined(powerpc_HOST_ARCH)
1040     machoInitSymbolsWithoutUnderscore();
1041 #   endif
1042
1043 #   if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1044 #   if defined(RTLD_DEFAULT)
1045     dl_prog_handle = RTLD_DEFAULT;
1046 #   else
1047     dl_prog_handle = dlopen(NULL, RTLD_LAZY);
1048 #   endif /* RTLD_DEFAULT */
1049 #   endif
1050
1051 #if defined(x86_64_HOST_ARCH)
1052     if (RtsFlags.MiscFlags.linkerMemBase != 0) {
1053         // User-override for mmap_32bit_base
1054         mmap_32bit_base = (void*)RtsFlags.MiscFlags.linkerMemBase;
1055     }
1056 #endif
1057 }
1058
1059 /* -----------------------------------------------------------------------------
1060  *                  Loading DLL or .so dynamic libraries
1061  * -----------------------------------------------------------------------------
1062  *
1063  * Add a DLL from which symbols may be found.  In the ELF case, just
1064  * do RTLD_GLOBAL-style add, so no further messing around needs to
1065  * happen in order that symbols in the loaded .so are findable --
1066  * lookupSymbol() will subsequently see them by dlsym on the program's
1067  * dl-handle.  Returns NULL if success, otherwise ptr to an err msg.
1068  *
1069  * In the PEi386 case, open the DLLs and put handles to them in a
1070  * linked list.  When looking for a symbol, try all handles in the
1071  * list.  This means that we need to load even DLLs that are guaranteed
1072  * to be in the ghc.exe image already, just so we can get a handle
1073  * to give to loadSymbol, so that we can find the symbols.  For such
1074  * libraries, the LoadLibrary call should be a no-op except for returning
1075  * the handle.
1076  *
1077  */
1078
1079 #if defined(OBJFORMAT_PEi386)
1080 /* A record for storing handles into DLLs. */
1081
1082 typedef
1083    struct _OpenedDLL {
1084       char*              name;
1085       struct _OpenedDLL* next;
1086       HINSTANCE instance;
1087    }
1088    OpenedDLL;
1089
1090 /* A list thereof. */
1091 static OpenedDLL* opened_dlls = NULL;
1092 #endif
1093
1094 const char *
1095 addDLL( char *dll_name )
1096 {
1097 #  if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
1098    /* ------------------- ELF DLL loader ------------------- */
1099    void *hdl;
1100    const char *errmsg;
1101
1102    initLinker();
1103
1104    // omitted: RTLD_NOW
1105    // see http://www.haskell.org/pipermail/cvs-ghc/2007-September/038570.html
1106    hdl= dlopen(dll_name, RTLD_LAZY | RTLD_GLOBAL);
1107
1108    if (hdl == NULL) {
1109       /* dlopen failed; return a ptr to the error msg. */
1110       errmsg = dlerror();
1111       if (errmsg == NULL) errmsg = "addDLL: unknown error";
1112       return errmsg;
1113    } else {
1114       return NULL;
1115    }
1116    /*NOTREACHED*/
1117
1118 #  elif defined(OBJFORMAT_PEi386)
1119    /* ------------------- Win32 DLL loader ------------------- */
1120
1121    char*      buf;
1122    OpenedDLL* o_dll;
1123    HINSTANCE  instance;
1124
1125    initLinker();
1126
1127    /* debugBelch("\naddDLL; dll_name = `%s'\n", dll_name); */
1128
1129    /* See if we've already got it, and ignore if so. */
1130    for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
1131       if (0 == strcmp(o_dll->name, dll_name))
1132          return NULL;
1133    }
1134
1135    /* The file name has no suffix (yet) so that we can try
1136       both foo.dll and foo.drv
1137
1138       The documentation for LoadLibrary says:
1139         If no file name extension is specified in the lpFileName
1140         parameter, the default library extension .dll is
1141         appended. However, the file name string can include a trailing
1142         point character (.) to indicate that the module name has no
1143         extension. */
1144
1145    buf = stgMallocBytes(strlen(dll_name) + 10, "addDLL");
1146    sprintf(buf, "%s.DLL", dll_name);
1147    instance = LoadLibrary(buf);
1148    if (instance == NULL) {
1149        if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1150        // KAA: allow loading of drivers (like winspool.drv)
1151        sprintf(buf, "%s.DRV", dll_name);
1152        instance = LoadLibrary(buf);
1153        if (instance == NULL) {
1154            if (GetLastError() != ERROR_MOD_NOT_FOUND) goto error;
1155            // #1883: allow loading of unix-style libfoo.dll DLLs
1156            sprintf(buf, "lib%s.DLL", dll_name);
1157            instance = LoadLibrary(buf);
1158            if (instance == NULL) {
1159                goto error;
1160            }
1161        }
1162    }
1163    stgFree(buf);
1164
1165    /* Add this DLL to the list of DLLs in which to search for symbols. */
1166    o_dll = stgMallocBytes( sizeof(OpenedDLL), "addDLL" );
1167    o_dll->name     = stgMallocBytes(1+strlen(dll_name), "addDLL");
1168    strcpy(o_dll->name, dll_name);
1169    o_dll->instance = instance;
1170    o_dll->next     = opened_dlls;
1171    opened_dlls     = o_dll;
1172
1173    return NULL;
1174
1175 error:
1176    stgFree(buf);
1177    sysErrorBelch(dll_name);
1178
1179    /* LoadLibrary failed; return a ptr to the error msg. */
1180    return "addDLL: could not load DLL";
1181
1182 #  else
1183    barf("addDLL: not implemented on this platform");
1184 #  endif
1185 }
1186
1187 /* -----------------------------------------------------------------------------
1188  * insert a stable symbol in the hash table
1189  */
1190
1191 void
1192 insertStableSymbol(char* obj_name, char* key, StgPtr p)
1193 {
1194   ghciInsertStrHashTable(obj_name, stablehash, key, getStablePtr(p));
1195 }
1196
1197
1198 /* -----------------------------------------------------------------------------
1199  * insert a symbol in the hash table
1200  */
1201 void
1202 insertSymbol(char* obj_name, char* key, void* data)
1203 {
1204   ghciInsertStrHashTable(obj_name, symhash, key, data);
1205 }
1206
1207 /* -----------------------------------------------------------------------------
1208  * lookup a symbol in the hash table
1209  */
1210 void *
1211 lookupSymbol( char *lbl )
1212 {
1213     void *val;
1214     initLinker() ;
1215     ASSERT(symhash != NULL);
1216     val = lookupStrHashTable(symhash, lbl);
1217
1218     if (val == NULL) {
1219 #       if defined(OBJFORMAT_ELF)
1220         return dlsym(dl_prog_handle, lbl);
1221 #       elif defined(OBJFORMAT_MACHO)
1222 #       if HAVE_DLFCN_H
1223         /* On OS X 10.3 and later, we use dlsym instead of the old legacy
1224            interface.
1225
1226            HACK: On OS X, global symbols are prefixed with an underscore.
1227                  However, dlsym wants us to omit the leading underscore from the
1228                  symbol name. For now, we simply strip it off here (and ONLY
1229                  here).
1230         */
1231         ASSERT(lbl[0] == '_');
1232         return dlsym(dl_prog_handle, lbl+1);
1233 #       else
1234         if(NSIsSymbolNameDefined(lbl)) {
1235             NSSymbol symbol = NSLookupAndBindSymbol(lbl);
1236             return NSAddressOfSymbol(symbol);
1237         } else {
1238             return NULL;
1239         }
1240 #       endif /* HAVE_DLFCN_H */
1241 #       elif defined(OBJFORMAT_PEi386)
1242         void* sym;
1243
1244         sym = lookupSymbolInDLLs(lbl);
1245         if (sym != NULL) { return sym; };
1246
1247         // Also try looking up the symbol without the @N suffix.  Some
1248         // DLLs have the suffixes on their symbols, some don't.
1249         zapTrailingAtSign ( lbl );
1250         sym = lookupSymbolInDLLs(lbl);
1251         if (sym != NULL) { return sym; };
1252         return NULL;
1253
1254 #       else
1255         ASSERT(2+2 == 5);
1256         return NULL;
1257 #       endif
1258     } else {
1259         return val;
1260     }
1261 }
1262
1263 /* -----------------------------------------------------------------------------
1264  * Debugging aid: look in GHCi's object symbol tables for symbols
1265  * within DELTA bytes of the specified address, and show their names.
1266  */
1267 #ifdef DEBUG
1268 void ghci_enquire ( char* addr );
1269
1270 void ghci_enquire ( char* addr )
1271 {
1272    int   i;
1273    char* sym;
1274    char* a;
1275    const int DELTA = 64;
1276    ObjectCode* oc;
1277
1278    initLinker();
1279
1280    for (oc = objects; oc; oc = oc->next) {
1281       for (i = 0; i < oc->n_symbols; i++) {
1282          sym = oc->symbols[i];
1283          if (sym == NULL) continue;
1284          a = NULL;
1285          if (a == NULL) {
1286             a = lookupStrHashTable(symhash, sym);
1287          }
1288          if (a == NULL) {
1289              // debugBelch("ghci_enquire: can't find %s\n", sym);
1290          }
1291          else if (addr-DELTA <= a && a <= addr+DELTA) {
1292             debugBelch("%p + %3d  ==  `%s'\n", addr, (int)(a - addr), sym);
1293          }
1294       }
1295    }
1296 }
1297 #endif
1298
1299 #ifdef ia64_HOST_ARCH
1300 static unsigned int PLTSize(void);
1301 #endif
1302
1303 #ifdef USE_MMAP
1304 #define ROUND_UP(x,size) ((x + size - 1) & ~(size - 1))
1305
1306 static void *
1307 mmapForLinker (size_t bytes, nat flags, int fd)
1308 {
1309    void *map_addr = NULL;
1310    void *result;
1311    int pagesize, size;
1312    static nat fixed = 0;
1313
1314    pagesize = getpagesize();
1315    size = ROUND_UP(bytes, pagesize);
1316
1317 #if defined(x86_64_HOST_ARCH)
1318 mmap_again:
1319
1320    if (mmap_32bit_base != 0) {
1321        map_addr = mmap_32bit_base;
1322    }
1323 #endif
1324
1325    result = mmap(map_addr, size, PROT_EXEC|PROT_READ|PROT_WRITE,
1326                     MAP_PRIVATE|TRY_MAP_32BIT|fixed|flags, fd, 0);
1327
1328    if (result == MAP_FAILED) {
1329        sysErrorBelch("mmap %lu bytes at %p",(lnat)size,map_addr);
1330        errorBelch("Try specifying an address with +RTS -xm<addr> -RTS");
1331        stg_exit(EXIT_FAILURE);
1332    }
1333
1334 #if defined(x86_64_HOST_ARCH)
1335    if (mmap_32bit_base != 0) {
1336        if (result == map_addr) {
1337            mmap_32bit_base = map_addr + size;
1338        } else {
1339            if ((W_)result > 0x80000000) {
1340                // oops, we were given memory over 2Gb
1341 #if defined(freebsd_HOST_OS)
1342                // Some platforms require MAP_FIXED.  This is normally
1343                // a bad idea, because MAP_FIXED will overwrite
1344                // existing mappings.
1345                munmap(result,size);
1346                fixed = MAP_FIXED;
1347                goto mmap_again;
1348 #else
1349                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);
1350 #endif
1351            } else {
1352                // hmm, we were given memory somewhere else, but it's
1353                // still under 2Gb so we can use it.  Next time, ask
1354                // for memory right after the place we just got some
1355                mmap_32bit_base = (void*)result + size;
1356            }
1357        }
1358    } else {
1359        if ((W_)result > 0x80000000) {
1360            // oops, we were given memory over 2Gb
1361            // ... try allocating memory somewhere else?;
1362            debugTrace(DEBUG_linker,"MAP_32BIT didn't work; gave us %lu bytes at 0x%p", bytes, result);
1363            munmap(result, size);
1364
1365            // Set a base address and try again... (guess: 1Gb)
1366            mmap_32bit_base = (void*)0x40000000;
1367            goto mmap_again;
1368        }
1369    }
1370 #endif
1371
1372    return result;
1373 }
1374 #endif // USE_MMAP
1375
1376 /* -----------------------------------------------------------------------------
1377  * Load an obj (populate the global symbol table, but don't resolve yet)
1378  *
1379  * Returns: 1 if ok, 0 on error.
1380  */
1381 HsInt
1382 loadObj( char *path )
1383 {
1384    ObjectCode* oc;
1385    struct stat st;
1386    int r;
1387 #ifdef USE_MMAP
1388    int fd;
1389 #else
1390    FILE *f;
1391 #endif
1392    initLinker();
1393
1394    /* debugBelch("loadObj %s\n", path ); */
1395
1396    /* Check that we haven't already loaded this object.
1397       Ignore requests to load multiple times */
1398    {
1399        ObjectCode *o;
1400        int is_dup = 0;
1401        for (o = objects; o; o = o->next) {
1402           if (0 == strcmp(o->fileName, path)) {
1403              is_dup = 1;
1404              break; /* don't need to search further */
1405           }
1406        }
1407        if (is_dup) {
1408           IF_DEBUG(linker, debugBelch(
1409             "GHCi runtime linker: warning: looks like you're trying to load the\n"
1410             "same object file twice:\n"
1411             "   %s\n"
1412             "GHCi will ignore this, but be warned.\n"
1413             , path));
1414           return 1; /* success */
1415        }
1416    }
1417
1418    oc = stgMallocBytes(sizeof(ObjectCode), "loadObj(oc)");
1419
1420 #  if defined(OBJFORMAT_ELF)
1421    oc->formatName = "ELF";
1422 #  elif defined(OBJFORMAT_PEi386)
1423    oc->formatName = "PEi386";
1424 #  elif defined(OBJFORMAT_MACHO)
1425    oc->formatName = "Mach-O";
1426 #  else
1427    stgFree(oc);
1428    barf("loadObj: not implemented on this platform");
1429 #  endif
1430
1431    r = stat(path, &st);
1432    if (r == -1) { return 0; }
1433
1434    /* sigh, strdup() isn't a POSIX function, so do it the long way */
1435    oc->fileName = stgMallocBytes( strlen(path)+1, "loadObj" );
1436    strcpy(oc->fileName, path);
1437
1438    oc->fileSize          = st.st_size;
1439    oc->symbols           = NULL;
1440    oc->sections          = NULL;
1441    oc->proddables        = NULL;
1442
1443    /* chain it onto the list of objects */
1444    oc->next              = objects;
1445    objects               = oc;
1446
1447 #ifdef USE_MMAP
1448    /* On many architectures malloc'd memory isn't executable, so we need to use mmap. */
1449
1450 #if defined(openbsd_HOST_OS)
1451    fd = open(path, O_RDONLY, S_IRUSR);
1452 #else
1453    fd = open(path, O_RDONLY);
1454 #endif
1455    if (fd == -1)
1456       barf("loadObj: can't open `%s'", path);
1457
1458 #ifdef ia64_HOST_ARCH
1459    /* The PLT needs to be right before the object */
1460    {
1461    int pagesize, n;
1462    pagesize = getpagesize();
1463    n = ROUND_UP(PLTSize(), pagesize);
1464    oc->plt = mmap(NULL, n, PROT_EXEC|PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
1465    if (oc->plt == MAP_FAILED)
1466       barf("loadObj: can't allocate PLT");
1467
1468    oc->pltIndex = 0;
1469    map_addr = oc->plt + n;
1470
1471    n = ROUND_UP(oc->fileSize, pagesize);
1472    oc->image = mmap(map_addr, n, PROT_EXEC|PROT_READ|PROT_WRITE,
1473                     MAP_PRIVATE|TRY_MAP_32BIT, fd, 0);
1474    if (oc->image == MAP_FAILED)
1475       barf("loadObj: can't map `%s'", path);
1476    }
1477 #else
1478    oc->image = mmapForLinker(oc->fileSize, 0, fd);
1479 #endif
1480
1481    close(fd);
1482
1483 #else /* !USE_MMAP */
1484    /* load the image into memory */
1485    f = fopen(path, "rb");
1486    if (!f)
1487        barf("loadObj: can't read `%s'", path);
1488
1489 #   if defined(mingw32_HOST_OS)
1490         // TODO: We would like to use allocateExec here, but allocateExec
1491         //       cannot currently allocate blocks large enough.
1492     oc->image = VirtualAlloc(NULL, oc->fileSize, MEM_RESERVE | MEM_COMMIT,
1493                              PAGE_EXECUTE_READWRITE);
1494 #   elif defined(darwin_HOST_OS)
1495     // In a Mach-O .o file, all sections can and will be misaligned
1496     // if the total size of the headers is not a multiple of the
1497     // desired alignment. This is fine for .o files that only serve
1498     // as input for the static linker, but it's not fine for us,
1499     // as SSE (used by gcc for floating point) and Altivec require
1500     // 16-byte alignment.
1501     // We calculate the correct alignment from the header before
1502     // reading the file, and then we misalign oc->image on purpose so
1503     // that the actual sections end up aligned again.
1504    oc->misalignment = machoGetMisalignment(f);
1505    oc->image = stgMallocBytes(oc->fileSize + oc->misalignment, "loadObj(image)");
1506    oc->image += oc->misalignment;
1507 #  else
1508    oc->image = stgMallocBytes(oc->fileSize, "loadObj(image)");
1509 #  endif
1510
1511    {
1512        int n;
1513        n = fread ( oc->image, 1, oc->fileSize, f );
1514        if (n != oc->fileSize)
1515            barf("loadObj: error whilst reading `%s'", path);
1516    }
1517    fclose(f);
1518 #endif /* USE_MMAP */
1519
1520 #  if defined(OBJFORMAT_MACHO) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
1521    r = ocAllocateSymbolExtras_MachO ( oc );
1522    if (!r) { return r; }
1523 #  elif defined(OBJFORMAT_ELF) && (defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH))
1524    r = ocAllocateSymbolExtras_ELF ( oc );
1525    if (!r) { return r; }
1526 #endif
1527
1528    /* verify the in-memory image */
1529 #  if defined(OBJFORMAT_ELF)
1530    r = ocVerifyImage_ELF ( oc );
1531 #  elif defined(OBJFORMAT_PEi386)
1532    r = ocVerifyImage_PEi386 ( oc );
1533 #  elif defined(OBJFORMAT_MACHO)
1534    r = ocVerifyImage_MachO ( oc );
1535 #  else
1536    barf("loadObj: no verify method");
1537 #  endif
1538    if (!r) { return r; }
1539
1540    /* build the symbol list for this image */
1541 #  if defined(OBJFORMAT_ELF)
1542    r = ocGetNames_ELF ( oc );
1543 #  elif defined(OBJFORMAT_PEi386)
1544    r = ocGetNames_PEi386 ( oc );
1545 #  elif defined(OBJFORMAT_MACHO)
1546    r = ocGetNames_MachO ( oc );
1547 #  else
1548    barf("loadObj: no getNames method");
1549 #  endif
1550    if (!r) { return r; }
1551
1552    /* loaded, but not resolved yet */
1553    oc->status = OBJECT_LOADED;
1554
1555    return 1;
1556 }
1557
1558 /* -----------------------------------------------------------------------------
1559  * resolve all the currently unlinked objects in memory
1560  *
1561  * Returns: 1 if ok, 0 on error.
1562  */
1563 HsInt
1564 resolveObjs( void )
1565 {
1566     ObjectCode *oc;
1567     int r;
1568
1569     initLinker();
1570
1571     for (oc = objects; oc; oc = oc->next) {
1572         if (oc->status != OBJECT_RESOLVED) {
1573 #           if defined(OBJFORMAT_ELF)
1574             r = ocResolve_ELF ( oc );
1575 #           elif defined(OBJFORMAT_PEi386)
1576             r = ocResolve_PEi386 ( oc );
1577 #           elif defined(OBJFORMAT_MACHO)
1578             r = ocResolve_MachO ( oc );
1579 #           else
1580             barf("resolveObjs: not implemented on this platform");
1581 #           endif
1582             if (!r) { return r; }
1583             oc->status = OBJECT_RESOLVED;
1584         }
1585     }
1586     return 1;
1587 }
1588
1589 /* -----------------------------------------------------------------------------
1590  * delete an object from the pool
1591  */
1592 HsInt
1593 unloadObj( char *path )
1594 {
1595     ObjectCode *oc, *prev;
1596
1597     ASSERT(symhash != NULL);
1598     ASSERT(objects != NULL);
1599
1600     initLinker();
1601
1602     prev = NULL;
1603     for (oc = objects; oc; prev = oc, oc = oc->next) {
1604         if (!strcmp(oc->fileName,path)) {
1605
1606             /* Remove all the mappings for the symbols within this
1607              * object..
1608              */
1609             {
1610                 int i;
1611                 for (i = 0; i < oc->n_symbols; i++) {
1612                    if (oc->symbols[i] != NULL) {
1613                        removeStrHashTable(symhash, oc->symbols[i], NULL);
1614                    }
1615                 }
1616             }
1617
1618             if (prev == NULL) {
1619                 objects = oc->next;
1620             } else {
1621                 prev->next = oc->next;
1622             }
1623
1624             // We're going to leave this in place, in case there are
1625             // any pointers from the heap into it:
1626                 // #ifdef mingw32_HOST_OS
1627                 //  VirtualFree(oc->image);
1628                 // #else
1629             //  stgFree(oc->image);
1630             // #endif
1631             stgFree(oc->fileName);
1632             stgFree(oc->symbols);
1633             stgFree(oc->sections);
1634             stgFree(oc);
1635             return 1;
1636         }
1637     }
1638
1639     errorBelch("unloadObj: can't find `%s' to unload", path);
1640     return 0;
1641 }
1642
1643 /* -----------------------------------------------------------------------------
1644  * Sanity checking.  For each ObjectCode, maintain a list of address ranges
1645  * which may be prodded during relocation, and abort if we try and write
1646  * outside any of these.
1647  */
1648 static void addProddableBlock ( ObjectCode* oc, void* start, int size )
1649 {
1650    ProddableBlock* pb
1651       = stgMallocBytes(sizeof(ProddableBlock), "addProddableBlock");
1652    /* debugBelch("aPB %p %p %d\n", oc, start, size); */
1653    ASSERT(size > 0);
1654    pb->start      = start;
1655    pb->size       = size;
1656    pb->next       = oc->proddables;
1657    oc->proddables = pb;
1658 }
1659
1660 static void checkProddableBlock ( ObjectCode* oc, void* addr )
1661 {
1662    ProddableBlock* pb;
1663    for (pb = oc->proddables; pb != NULL; pb = pb->next) {
1664       char* s = (char*)(pb->start);
1665       char* e = s + pb->size - 1;
1666       char* a = (char*)addr;
1667       /* Assumes that the biggest fixup involves a 4-byte write.  This
1668          probably needs to be changed to 8 (ie, +7) on 64-bit
1669          plats. */
1670       if (a >= s && (a+3) <= e) return;
1671    }
1672    barf("checkProddableBlock: invalid fixup in runtime linker");
1673 }
1674
1675 /* -----------------------------------------------------------------------------
1676  * Section management.
1677  */
1678 static void addSection ( ObjectCode* oc, SectionKind kind,
1679                          void* start, void* end )
1680 {
1681    Section* s   = stgMallocBytes(sizeof(Section), "addSection");
1682    s->start     = start;
1683    s->end       = end;
1684    s->kind      = kind;
1685    s->next      = oc->sections;
1686    oc->sections = s;
1687    /*
1688    debugBelch("addSection: %p-%p (size %d), kind %d\n",
1689                    start, ((char*)end)-1, end - start + 1, kind );
1690    */
1691 }
1692
1693
1694 /* --------------------------------------------------------------------------
1695  * Symbol Extras.
1696  * This is about allocating a small chunk of memory for every symbol in the
1697  * object file. We make sure that the SymboLExtras are always "in range" of
1698  * limited-range PC-relative instructions on various platforms by allocating
1699  * them right next to the object code itself.
1700  */
1701
1702 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
1703
1704 /*
1705   ocAllocateSymbolExtras
1706
1707   Allocate additional space at the end of the object file image to make room
1708   for jump islands (powerpc, x86_64) and GOT entries (x86_64).
1709
1710   PowerPC relative branch instructions have a 24 bit displacement field.
1711   As PPC code is always 4-byte-aligned, this yields a +-32MB range.
1712   If a particular imported symbol is outside this range, we have to redirect
1713   the jump to a short piece of new code that just loads the 32bit absolute
1714   address and jumps there.
1715   On x86_64, PC-relative jumps and PC-relative accesses to the GOT are limited
1716   to 32 bits (+-2GB).
1717
1718   This function just allocates space for one SymbolExtra for every
1719   undefined symbol in the object file. The code for the jump islands is
1720   filled in by makeSymbolExtra below.
1721 */
1722
1723 static int ocAllocateSymbolExtras( ObjectCode* oc, int count, int first )
1724 {
1725 #ifdef USE_MMAP
1726   int pagesize, n, m;
1727 #endif
1728   int aligned;
1729 #ifndef USE_MMAP
1730   int misalignment = 0;
1731 #ifdef darwin_HOST_OS
1732   misalignment = oc->misalignment;
1733 #endif
1734 #endif
1735
1736   if( count > 0 )
1737   {
1738     // round up to the nearest 4
1739     aligned = (oc->fileSize + 3) & ~3;
1740
1741 #ifdef USE_MMAP
1742     pagesize = getpagesize();
1743     n = ROUND_UP( oc->fileSize, pagesize );
1744     m = ROUND_UP( aligned + sizeof (SymbolExtra) * count, pagesize );
1745
1746     /* we try to use spare space at the end of the last page of the
1747      * image for the jump islands, but if there isn't enough space
1748      * then we have to map some (anonymously, remembering MAP_32BIT).
1749      */
1750     if( m > n ) // we need to allocate more pages
1751     {
1752         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
1753                                           MAP_ANONYMOUS, -1);
1754     }
1755     else
1756     {
1757         oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
1758     }
1759 #else
1760     oc->image -= misalignment;
1761     oc->image = stgReallocBytes( oc->image,
1762                                  misalignment +
1763                                  aligned + sizeof (SymbolExtra) * count,
1764                                  "ocAllocateSymbolExtras" );
1765     oc->image += misalignment;
1766
1767     oc->symbol_extras = (SymbolExtra *) (oc->image + aligned);
1768 #endif /* USE_MMAP */
1769
1770     memset( oc->symbol_extras, 0, sizeof (SymbolExtra) * count );
1771   }
1772   else
1773     oc->symbol_extras = NULL;
1774
1775   oc->first_symbol_extra = first;
1776   oc->n_symbol_extras = count;
1777
1778   return 1;
1779 }
1780
1781 static SymbolExtra* makeSymbolExtra( ObjectCode* oc,
1782                                      unsigned long symbolNumber,
1783                                      unsigned long target )
1784 {
1785   SymbolExtra *extra;
1786
1787   ASSERT( symbolNumber >= oc->first_symbol_extra
1788         && symbolNumber - oc->first_symbol_extra < oc->n_symbol_extras);
1789
1790   extra = &oc->symbol_extras[symbolNumber - oc->first_symbol_extra];
1791
1792 #ifdef powerpc_HOST_ARCH
1793   // lis r12, hi16(target)
1794   extra->jumpIsland.lis_r12     = 0x3d80;
1795   extra->jumpIsland.hi_addr     = target >> 16;
1796
1797   // ori r12, r12, lo16(target)
1798   extra->jumpIsland.ori_r12_r12 = 0x618c;
1799   extra->jumpIsland.lo_addr     = target & 0xffff;
1800
1801   // mtctr r12
1802   extra->jumpIsland.mtctr_r12   = 0x7d8903a6;
1803
1804   // bctr
1805   extra->jumpIsland.bctr        = 0x4e800420;
1806 #endif
1807 #ifdef x86_64_HOST_ARCH
1808         // jmp *-14(%rip)
1809   static uint8_t jmp[] = { 0xFF, 0x25, 0xF2, 0xFF, 0xFF, 0xFF };
1810   extra->addr = target;
1811   memcpy(extra->jumpIsland, jmp, 6);
1812 #endif
1813
1814   return extra;
1815 }
1816
1817 #endif
1818
1819 /* --------------------------------------------------------------------------
1820  * PowerPC specifics (instruction cache flushing)
1821  * ------------------------------------------------------------------------*/
1822
1823 #ifdef powerpc_TARGET_ARCH
1824 /*
1825    ocFlushInstructionCache
1826
1827    Flush the data & instruction caches.
1828    Because the PPC has split data/instruction caches, we have to
1829    do that whenever we modify code at runtime.
1830  */
1831
1832 static void ocFlushInstructionCache( ObjectCode *oc )
1833 {
1834     int n = (oc->fileSize + sizeof( SymbolExtra ) * oc->n_symbol_extras + 3) / 4;
1835     unsigned long *p = (unsigned long *) oc->image;
1836
1837     while( n-- )
1838     {
1839         __asm__ volatile ( "dcbf 0,%0\n\t"
1840                            "sync\n\t"
1841                            "icbi 0,%0"
1842                            :
1843                            : "r" (p)
1844                          );
1845         p++;
1846     }
1847     __asm__ volatile ( "sync\n\t"
1848                        "isync"
1849                      );
1850 }
1851 #endif
1852
1853 /* --------------------------------------------------------------------------
1854  * PEi386 specifics (Win32 targets)
1855  * ------------------------------------------------------------------------*/
1856
1857 /* The information for this linker comes from
1858       Microsoft Portable Executable
1859       and Common Object File Format Specification
1860       revision 5.1 January 1998
1861    which SimonM says comes from the MS Developer Network CDs.
1862
1863    It can be found there (on older CDs), but can also be found
1864    online at:
1865
1866       http://www.microsoft.com/hwdev/hardware/PECOFF.asp
1867
1868    (this is Rev 6.0 from February 1999).
1869
1870    Things move, so if that fails, try searching for it via
1871
1872       http://www.google.com/search?q=PE+COFF+specification
1873
1874    The ultimate reference for the PE format is the Winnt.h
1875    header file that comes with the Platform SDKs; as always,
1876    implementations will drift wrt their documentation.
1877
1878    A good background article on the PE format is Matt Pietrek's
1879    March 1994 article in Microsoft System Journal (MSJ)
1880    (Vol.9, No. 3): "Peering Inside the PE: A Tour of the
1881    Win32 Portable Executable File Format." The info in there
1882    has recently been updated in a two part article in
1883    MSDN magazine, issues Feb and March 2002,
1884    "Inside Windows: An In-Depth Look into the Win32 Portable
1885    Executable File Format"
1886
1887    John Levine's book "Linkers and Loaders" contains useful
1888    info on PE too.
1889 */
1890
1891
1892 #if defined(OBJFORMAT_PEi386)
1893
1894
1895
1896 typedef unsigned char  UChar;
1897 typedef unsigned short UInt16;
1898 typedef unsigned int   UInt32;
1899 typedef          int   Int32;
1900
1901
1902 typedef
1903    struct {
1904       UInt16 Machine;
1905       UInt16 NumberOfSections;
1906       UInt32 TimeDateStamp;
1907       UInt32 PointerToSymbolTable;
1908       UInt32 NumberOfSymbols;
1909       UInt16 SizeOfOptionalHeader;
1910       UInt16 Characteristics;
1911    }
1912    COFF_header;
1913
1914 #define sizeof_COFF_header 20
1915
1916
1917 typedef
1918    struct {
1919       UChar  Name[8];
1920       UInt32 VirtualSize;
1921       UInt32 VirtualAddress;
1922       UInt32 SizeOfRawData;
1923       UInt32 PointerToRawData;
1924       UInt32 PointerToRelocations;
1925       UInt32 PointerToLinenumbers;
1926       UInt16 NumberOfRelocations;
1927       UInt16 NumberOfLineNumbers;
1928       UInt32 Characteristics;
1929    }
1930    COFF_section;
1931
1932 #define sizeof_COFF_section 40
1933
1934
1935 typedef
1936    struct {
1937       UChar  Name[8];
1938       UInt32 Value;
1939       UInt16 SectionNumber;
1940       UInt16 Type;
1941       UChar  StorageClass;
1942       UChar  NumberOfAuxSymbols;
1943    }
1944    COFF_symbol;
1945
1946 #define sizeof_COFF_symbol 18
1947
1948
1949 typedef
1950    struct {
1951       UInt32 VirtualAddress;
1952       UInt32 SymbolTableIndex;
1953       UInt16 Type;
1954    }
1955    COFF_reloc;
1956
1957 #define sizeof_COFF_reloc 10
1958
1959
1960 /* From PE spec doc, section 3.3.2 */
1961 /* Note use of MYIMAGE_* since IMAGE_* are already defined in
1962    windows.h -- for the same purpose, but I want to know what I'm
1963    getting, here. */
1964 #define MYIMAGE_FILE_RELOCS_STRIPPED     0x0001
1965 #define MYIMAGE_FILE_EXECUTABLE_IMAGE    0x0002
1966 #define MYIMAGE_FILE_DLL                 0x2000
1967 #define MYIMAGE_FILE_SYSTEM              0x1000
1968 #define MYIMAGE_FILE_BYTES_REVERSED_HI   0x8000
1969 #define MYIMAGE_FILE_BYTES_REVERSED_LO   0x0080
1970 #define MYIMAGE_FILE_32BIT_MACHINE       0x0100
1971
1972 /* From PE spec doc, section 5.4.2 and 5.4.4 */
1973 #define MYIMAGE_SYM_CLASS_EXTERNAL       2
1974 #define MYIMAGE_SYM_CLASS_STATIC         3
1975 #define MYIMAGE_SYM_UNDEFINED            0
1976
1977 /* From PE spec doc, section 4.1 */
1978 #define MYIMAGE_SCN_CNT_CODE             0x00000020
1979 #define MYIMAGE_SCN_CNT_INITIALIZED_DATA 0x00000040
1980 #define MYIMAGE_SCN_LNK_NRELOC_OVFL      0x01000000
1981
1982 /* From PE spec doc, section 5.2.1 */
1983 #define MYIMAGE_REL_I386_DIR32           0x0006
1984 #define MYIMAGE_REL_I386_REL32           0x0014
1985
1986
1987 /* We use myindex to calculate array addresses, rather than
1988    simply doing the normal subscript thing.  That's because
1989    some of the above structs have sizes which are not
1990    a whole number of words.  GCC rounds their sizes up to a
1991    whole number of words, which means that the address calcs
1992    arising from using normal C indexing or pointer arithmetic
1993    are just plain wrong.  Sigh.
1994 */
1995 static UChar *
1996 myindex ( int scale, void* base, int index )
1997 {
1998    return
1999       ((UChar*)base) + scale * index;
2000 }
2001
2002
2003 static void
2004 printName ( UChar* name, UChar* strtab )
2005 {
2006    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2007       UInt32 strtab_offset = * (UInt32*)(name+4);
2008       debugBelch("%s", strtab + strtab_offset );
2009    } else {
2010       int i;
2011       for (i = 0; i < 8; i++) {
2012          if (name[i] == 0) break;
2013          debugBelch("%c", name[i] );
2014       }
2015    }
2016 }
2017
2018
2019 static void
2020 copyName ( UChar* name, UChar* strtab, UChar* dst, int dstSize )
2021 {
2022    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2023       UInt32 strtab_offset = * (UInt32*)(name+4);
2024       strncpy ( dst, strtab+strtab_offset, dstSize );
2025       dst[dstSize-1] = 0;
2026    } else {
2027       int i = 0;
2028       while (1) {
2029          if (i >= 8) break;
2030          if (name[i] == 0) break;
2031          dst[i] = name[i];
2032          i++;
2033       }
2034       dst[i] = 0;
2035    }
2036 }
2037
2038
2039 static UChar *
2040 cstring_from_COFF_symbol_name ( UChar* name, UChar* strtab )
2041 {
2042    UChar* newstr;
2043    /* If the string is longer than 8 bytes, look in the
2044       string table for it -- this will be correctly zero terminated.
2045    */
2046    if (name[0]==0 && name[1]==0 && name[2]==0 && name[3]==0) {
2047       UInt32 strtab_offset = * (UInt32*)(name+4);
2048       return ((UChar*)strtab) + strtab_offset;
2049    }
2050    /* Otherwise, if shorter than 8 bytes, return the original,
2051       which by defn is correctly terminated.
2052    */
2053    if (name[7]==0) return name;
2054    /* The annoying case: 8 bytes.  Copy into a temporary
2055       (which is never freed ...)
2056    */
2057    newstr = stgMallocBytes(9, "cstring_from_COFF_symbol_name");
2058    ASSERT(newstr);
2059    strncpy(newstr,name,8);
2060    newstr[8] = 0;
2061    return newstr;
2062 }
2063
2064
2065 /* Just compares the short names (first 8 chars) */
2066 static COFF_section *
2067 findPEi386SectionCalled ( ObjectCode* oc,  char* name )
2068 {
2069    int i;
2070    COFF_header* hdr
2071       = (COFF_header*)(oc->image);
2072    COFF_section* sectab
2073       = (COFF_section*) (
2074            ((UChar*)(oc->image))
2075            + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2076         );
2077    for (i = 0; i < hdr->NumberOfSections; i++) {
2078       UChar* n1;
2079       UChar* n2;
2080       COFF_section* section_i
2081          = (COFF_section*)
2082            myindex ( sizeof_COFF_section, sectab, i );
2083       n1 = (UChar*) &(section_i->Name);
2084       n2 = name;
2085       if (n1[0]==n2[0] && n1[1]==n2[1] && n1[2]==n2[2] &&
2086           n1[3]==n2[3] && n1[4]==n2[4] && n1[5]==n2[5] &&
2087           n1[6]==n2[6] && n1[7]==n2[7])
2088          return section_i;
2089    }
2090
2091    return NULL;
2092 }
2093
2094
2095 static void
2096 zapTrailingAtSign ( UChar* sym )
2097 {
2098 #  define my_isdigit(c) ((c) >= '0' && (c) <= '9')
2099    int i, j;
2100    if (sym[0] == 0) return;
2101    i = 0;
2102    while (sym[i] != 0) i++;
2103    i--;
2104    j = i;
2105    while (j > 0 && my_isdigit(sym[j])) j--;
2106    if (j > 0 && sym[j] == '@' && j != i) sym[j] = 0;
2107 #  undef my_isdigit
2108 }
2109
2110 static void *
2111 lookupSymbolInDLLs ( UChar *lbl )
2112 {
2113     OpenedDLL* o_dll;
2114     void *sym;
2115
2116     for (o_dll = opened_dlls; o_dll != NULL; o_dll = o_dll->next) {
2117         /* debugBelch("look in %s for %s\n", o_dll->name, lbl); */
2118
2119         if (lbl[0] == '_') {
2120             /* HACK: if the name has an initial underscore, try stripping
2121                it off & look that up first. I've yet to verify whether there's
2122                a Rule that governs whether an initial '_' *should always* be
2123                stripped off when mapping from import lib name to the DLL name.
2124             */
2125             sym = GetProcAddress(o_dll->instance, (lbl+1));
2126             if (sym != NULL) {
2127                 /*debugBelch("found %s in %s\n", lbl+1,o_dll->name);*/
2128                 return sym;
2129             }
2130         }
2131         sym = GetProcAddress(o_dll->instance, lbl);
2132         if (sym != NULL) {
2133             /*debugBelch("found %s in %s\n", lbl,o_dll->name);*/
2134             return sym;
2135            }
2136     }
2137     return NULL;
2138 }
2139
2140
2141 static int
2142 ocVerifyImage_PEi386 ( ObjectCode* oc )
2143 {
2144    int i;
2145    UInt32 j, noRelocs;
2146    COFF_header*  hdr;
2147    COFF_section* sectab;
2148    COFF_symbol*  symtab;
2149    UChar*        strtab;
2150    /* debugBelch("\nLOADING %s\n", oc->fileName); */
2151    hdr = (COFF_header*)(oc->image);
2152    sectab = (COFF_section*) (
2153                ((UChar*)(oc->image))
2154                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2155             );
2156    symtab = (COFF_symbol*) (
2157                ((UChar*)(oc->image))
2158                + hdr->PointerToSymbolTable
2159             );
2160    strtab = ((UChar*)symtab)
2161             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2162
2163    if (hdr->Machine != 0x14c) {
2164       errorBelch("%s: Not x86 PEi386", oc->fileName);
2165       return 0;
2166    }
2167    if (hdr->SizeOfOptionalHeader != 0) {
2168       errorBelch("%s: PEi386 with nonempty optional header", oc->fileName);
2169       return 0;
2170    }
2171    if ( /* (hdr->Characteristics & MYIMAGE_FILE_RELOCS_STRIPPED) || */
2172         (hdr->Characteristics & MYIMAGE_FILE_EXECUTABLE_IMAGE) ||
2173         (hdr->Characteristics & MYIMAGE_FILE_DLL) ||
2174         (hdr->Characteristics & MYIMAGE_FILE_SYSTEM) ) {
2175       errorBelch("%s: Not a PEi386 object file", oc->fileName);
2176       return 0;
2177    }
2178    if ( (hdr->Characteristics & MYIMAGE_FILE_BYTES_REVERSED_HI)
2179         /* || !(hdr->Characteristics & MYIMAGE_FILE_32BIT_MACHINE) */ ) {
2180       errorBelch("%s: Invalid PEi386 word size or endiannness: %d",
2181                  oc->fileName,
2182                  (int)(hdr->Characteristics));
2183       return 0;
2184    }
2185    /* If the string table size is way crazy, this might indicate that
2186       there are more than 64k relocations, despite claims to the
2187       contrary.  Hence this test. */
2188    /* debugBelch("strtab size %d\n", * (UInt32*)strtab); */
2189 #if 0
2190    if ( (*(UInt32*)strtab) > 600000 ) {
2191       /* Note that 600k has no special significance other than being
2192          big enough to handle the almost-2MB-sized lumps that
2193          constitute HSwin32*.o. */
2194       debugBelch("PEi386 object has suspiciously large string table; > 64k relocs?");
2195       return 0;
2196    }
2197 #endif
2198
2199    /* No further verification after this point; only debug printing. */
2200    i = 0;
2201    IF_DEBUG(linker, i=1);
2202    if (i == 0) return 1;
2203
2204    debugBelch( "sectab offset = %d\n", ((UChar*)sectab) - ((UChar*)hdr) );
2205    debugBelch( "symtab offset = %d\n", ((UChar*)symtab) - ((UChar*)hdr) );
2206    debugBelch( "strtab offset = %d\n", ((UChar*)strtab) - ((UChar*)hdr) );
2207
2208    debugBelch("\n" );
2209    debugBelch( "Machine:           0x%x\n", (UInt32)(hdr->Machine) );
2210    debugBelch( "# sections:        %d\n",   (UInt32)(hdr->NumberOfSections) );
2211    debugBelch( "time/date:         0x%x\n", (UInt32)(hdr->TimeDateStamp) );
2212    debugBelch( "symtab offset:     %d\n",   (UInt32)(hdr->PointerToSymbolTable) );
2213    debugBelch( "# symbols:         %d\n",   (UInt32)(hdr->NumberOfSymbols) );
2214    debugBelch( "sz of opt hdr:     %d\n",   (UInt32)(hdr->SizeOfOptionalHeader) );
2215    debugBelch( "characteristics:   0x%x\n", (UInt32)(hdr->Characteristics) );
2216
2217    /* Print the section table. */
2218    debugBelch("\n" );
2219    for (i = 0; i < hdr->NumberOfSections; i++) {
2220       COFF_reloc* reltab;
2221       COFF_section* sectab_i
2222          = (COFF_section*)
2223            myindex ( sizeof_COFF_section, sectab, i );
2224       debugBelch(
2225                 "\n"
2226                 "section %d\n"
2227                 "     name `",
2228                 i
2229               );
2230       printName ( sectab_i->Name, strtab );
2231       debugBelch(
2232                 "'\n"
2233                 "    vsize %d\n"
2234                 "    vaddr %d\n"
2235                 "  data sz %d\n"
2236                 " data off %d\n"
2237                 "  num rel %d\n"
2238                 "  off rel %d\n"
2239                 "  ptr raw 0x%x\n",
2240                 sectab_i->VirtualSize,
2241                 sectab_i->VirtualAddress,
2242                 sectab_i->SizeOfRawData,
2243                 sectab_i->PointerToRawData,
2244                 sectab_i->NumberOfRelocations,
2245                 sectab_i->PointerToRelocations,
2246                 sectab_i->PointerToRawData
2247               );
2248       reltab = (COFF_reloc*) (
2249                   ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2250                );
2251
2252       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2253         /* If the relocation field (a short) has overflowed, the
2254          * real count can be found in the first reloc entry.
2255          *
2256          * See Section 4.1 (last para) of the PE spec (rev6.0).
2257          */
2258         COFF_reloc* rel = (COFF_reloc*)
2259                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2260         noRelocs = rel->VirtualAddress;
2261         j = 1;
2262       } else {
2263         noRelocs = sectab_i->NumberOfRelocations;
2264         j = 0;
2265       }
2266
2267       for (; j < noRelocs; j++) {
2268          COFF_symbol* sym;
2269          COFF_reloc* rel = (COFF_reloc*)
2270                            myindex ( sizeof_COFF_reloc, reltab, j );
2271          debugBelch(
2272                    "        type 0x%-4x   vaddr 0x%-8x   name `",
2273                    (UInt32)rel->Type,
2274                    rel->VirtualAddress );
2275          sym = (COFF_symbol*)
2276                myindex ( sizeof_COFF_symbol, symtab, rel->SymbolTableIndex );
2277          /* Hmm..mysterious looking offset - what's it for? SOF */
2278          printName ( sym->Name, strtab -10 );
2279          debugBelch("'\n" );
2280       }
2281
2282       debugBelch("\n" );
2283    }
2284    debugBelch("\n" );
2285    debugBelch("string table has size 0x%x\n", * (UInt32*)strtab );
2286    debugBelch("---START of string table---\n");
2287    for (i = 4; i < *(Int32*)strtab; i++) {
2288       if (strtab[i] == 0)
2289          debugBelch("\n"); else
2290          debugBelch("%c", strtab[i] );
2291    }
2292    debugBelch("--- END  of string table---\n");
2293
2294    debugBelch("\n" );
2295    i = 0;
2296    while (1) {
2297       COFF_symbol* symtab_i;
2298       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2299       symtab_i = (COFF_symbol*)
2300                  myindex ( sizeof_COFF_symbol, symtab, i );
2301       debugBelch(
2302                 "symbol %d\n"
2303                 "     name `",
2304                 i
2305               );
2306       printName ( symtab_i->Name, strtab );
2307       debugBelch(
2308                 "'\n"
2309                 "    value 0x%x\n"
2310                 "   1+sec# %d\n"
2311                 "     type 0x%x\n"
2312                 "   sclass 0x%x\n"
2313                 "     nAux %d\n",
2314                 symtab_i->Value,
2315                 (Int32)(symtab_i->SectionNumber),
2316                 (UInt32)symtab_i->Type,
2317                 (UInt32)symtab_i->StorageClass,
2318                 (UInt32)symtab_i->NumberOfAuxSymbols
2319               );
2320       i += symtab_i->NumberOfAuxSymbols;
2321       i++;
2322    }
2323
2324    debugBelch("\n" );
2325    return 1;
2326 }
2327
2328
2329 static int
2330 ocGetNames_PEi386 ( ObjectCode* oc )
2331 {
2332    COFF_header*  hdr;
2333    COFF_section* sectab;
2334    COFF_symbol*  symtab;
2335    UChar*        strtab;
2336
2337    UChar* sname;
2338    void*  addr;
2339    int    i;
2340
2341    hdr = (COFF_header*)(oc->image);
2342    sectab = (COFF_section*) (
2343                ((UChar*)(oc->image))
2344                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2345             );
2346    symtab = (COFF_symbol*) (
2347                ((UChar*)(oc->image))
2348                + hdr->PointerToSymbolTable
2349             );
2350    strtab = ((UChar*)(oc->image))
2351             + hdr->PointerToSymbolTable
2352             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2353
2354    /* Allocate space for any (local, anonymous) .bss sections. */
2355
2356    for (i = 0; i < hdr->NumberOfSections; i++) {
2357       UInt32 bss_sz;
2358       UChar* zspace;
2359       COFF_section* sectab_i
2360          = (COFF_section*)
2361            myindex ( sizeof_COFF_section, sectab, i );
2362       if (0 != strcmp(sectab_i->Name, ".bss")) continue;
2363       /* sof 10/05: the PE spec text isn't too clear regarding what
2364        * the SizeOfRawData field is supposed to hold for object
2365        * file sections containing just uninitialized data -- for executables,
2366        * it is supposed to be zero; unclear what it's supposed to be
2367        * for object files. However, VirtualSize is guaranteed to be
2368        * zero for object files, which definitely suggests that SizeOfRawData
2369        * will be non-zero (where else would the size of this .bss section be
2370        * stored?) Looking at the COFF_section info for incoming object files,
2371        * this certainly appears to be the case.
2372        *
2373        * => I suspect we've been incorrectly handling .bss sections in (relocatable)
2374        * object files up until now. This turned out to bite us with ghc-6.4.1's use
2375        * of gcc-3.4.x, which has started to emit initially-zeroed-out local 'static'
2376        * variable decls into to the .bss section. (The specific function in Q which
2377        * triggered this is libraries/base/cbits/dirUtils.c:__hscore_getFolderPath())
2378        */
2379       if (sectab_i->VirtualSize == 0 && sectab_i->SizeOfRawData == 0) continue;
2380       /* This is a non-empty .bss section.  Allocate zeroed space for
2381          it, and set its PointerToRawData field such that oc->image +
2382          PointerToRawData == addr_of_zeroed_space.  */
2383       bss_sz = sectab_i->VirtualSize;
2384       if ( bss_sz < sectab_i->SizeOfRawData) { bss_sz = sectab_i->SizeOfRawData; }
2385       zspace = stgCallocBytes(1, bss_sz, "ocGetNames_PEi386(anonymous bss)");
2386       sectab_i->PointerToRawData = ((UChar*)zspace) - ((UChar*)(oc->image));
2387       addProddableBlock(oc, zspace, bss_sz);
2388       /* debugBelch("BSS anon section at 0x%x\n", zspace); */
2389    }
2390
2391    /* Copy section information into the ObjectCode. */
2392
2393    for (i = 0; i < hdr->NumberOfSections; i++) {
2394       UChar* start;
2395       UChar* end;
2396       UInt32 sz;
2397
2398       SectionKind kind
2399          = SECTIONKIND_OTHER;
2400       COFF_section* sectab_i
2401          = (COFF_section*)
2402            myindex ( sizeof_COFF_section, sectab, i );
2403       IF_DEBUG(linker, debugBelch("section name = %s\n", sectab_i->Name ));
2404
2405 #     if 0
2406       /* I'm sure this is the Right Way to do it.  However, the
2407          alternative of testing the sectab_i->Name field seems to
2408          work ok with Cygwin.
2409       */
2410       if (sectab_i->Characteristics & MYIMAGE_SCN_CNT_CODE ||
2411           sectab_i->Characteristics & MYIMAGE_SCN_CNT_INITIALIZED_DATA)
2412          kind = SECTIONKIND_CODE_OR_RODATA;
2413 #     endif
2414
2415       if (0==strcmp(".text",sectab_i->Name) ||
2416           0==strcmp(".rdata",sectab_i->Name)||
2417           0==strcmp(".rodata",sectab_i->Name))
2418          kind = SECTIONKIND_CODE_OR_RODATA;
2419       if (0==strcmp(".data",sectab_i->Name) ||
2420           0==strcmp(".bss",sectab_i->Name))
2421          kind = SECTIONKIND_RWDATA;
2422
2423       ASSERT(sectab_i->SizeOfRawData == 0 || sectab_i->VirtualSize == 0);
2424       sz = sectab_i->SizeOfRawData;
2425       if (sz < sectab_i->VirtualSize) sz = sectab_i->VirtualSize;
2426
2427       start = ((UChar*)(oc->image)) + sectab_i->PointerToRawData;
2428       end   = start + sz - 1;
2429
2430       if (kind == SECTIONKIND_OTHER
2431           /* Ignore sections called which contain stabs debugging
2432              information. */
2433           && 0 != strcmp(".stab", sectab_i->Name)
2434           && 0 != strcmp(".stabstr", sectab_i->Name)
2435           /* ignore constructor section for now */
2436           && 0 != strcmp(".ctors", sectab_i->Name)
2437           /* ignore section generated from .ident */
2438           && 0!= strcmp("/4", sectab_i->Name)
2439           /* ignore unknown section that appeared in gcc 3.4.5(?) */
2440           && 0!= strcmp(".reloc", sectab_i->Name)
2441          ) {
2442          errorBelch("Unknown PEi386 section name `%s' (while processing: %s)", sectab_i->Name, oc->fileName);
2443          return 0;
2444       }
2445
2446       if (kind != SECTIONKIND_OTHER && end >= start) {
2447          addSection(oc, kind, start, end);
2448          addProddableBlock(oc, start, end - start + 1);
2449       }
2450    }
2451
2452    /* Copy exported symbols into the ObjectCode. */
2453
2454    oc->n_symbols = hdr->NumberOfSymbols;
2455    oc->symbols   = stgMallocBytes(oc->n_symbols * sizeof(char*),
2456                                   "ocGetNames_PEi386(oc->symbols)");
2457    /* Call me paranoid; I don't care. */
2458    for (i = 0; i < oc->n_symbols; i++)
2459       oc->symbols[i] = NULL;
2460
2461    i = 0;
2462    while (1) {
2463       COFF_symbol* symtab_i;
2464       if (i >= (Int32)(hdr->NumberOfSymbols)) break;
2465       symtab_i = (COFF_symbol*)
2466                  myindex ( sizeof_COFF_symbol, symtab, i );
2467
2468       addr  = NULL;
2469
2470       if (symtab_i->StorageClass == MYIMAGE_SYM_CLASS_EXTERNAL
2471           && symtab_i->SectionNumber != MYIMAGE_SYM_UNDEFINED) {
2472          /* This symbol is global and defined, viz, exported */
2473          /* for MYIMAGE_SYMCLASS_EXTERNAL
2474                 && !MYIMAGE_SYM_UNDEFINED,
2475             the address of the symbol is:
2476                 address of relevant section + offset in section
2477          */
2478          COFF_section* sectabent
2479             = (COFF_section*) myindex ( sizeof_COFF_section,
2480                                         sectab,
2481                                         symtab_i->SectionNumber-1 );
2482          addr = ((UChar*)(oc->image))
2483                 + (sectabent->PointerToRawData
2484                    + symtab_i->Value);
2485       }
2486       else
2487       if (symtab_i->SectionNumber == MYIMAGE_SYM_UNDEFINED
2488           && symtab_i->Value > 0) {
2489          /* This symbol isn't in any section at all, ie, global bss.
2490             Allocate zeroed space for it. */
2491          addr = stgCallocBytes(1, symtab_i->Value,
2492                                "ocGetNames_PEi386(non-anonymous bss)");
2493          addSection(oc, SECTIONKIND_RWDATA, addr,
2494                         ((UChar*)addr) + symtab_i->Value - 1);
2495          addProddableBlock(oc, addr, symtab_i->Value);
2496          /* debugBelch("BSS      section at 0x%x\n", addr); */
2497       }
2498
2499       if (addr != NULL ) {
2500          sname = cstring_from_COFF_symbol_name ( symtab_i->Name, strtab );
2501          /* debugBelch("addSymbol %p `%s \n", addr,sname);  */
2502          IF_DEBUG(linker, debugBelch("addSymbol %p `%s'\n", addr,sname);)
2503          ASSERT(i >= 0 && i < oc->n_symbols);
2504          /* cstring_from_COFF_symbol_name always succeeds. */
2505          oc->symbols[i] = sname;
2506          ghciInsertStrHashTable(oc->fileName, symhash, sname, addr);
2507       } else {
2508 #        if 0
2509          debugBelch(
2510                    "IGNORING symbol %d\n"
2511                    "     name `",
2512                    i
2513                  );
2514          printName ( symtab_i->Name, strtab );
2515          debugBelch(
2516                    "'\n"
2517                    "    value 0x%x\n"
2518                    "   1+sec# %d\n"
2519                    "     type 0x%x\n"
2520                    "   sclass 0x%x\n"
2521                    "     nAux %d\n",
2522                    symtab_i->Value,
2523                    (Int32)(symtab_i->SectionNumber),
2524                    (UInt32)symtab_i->Type,
2525                    (UInt32)symtab_i->StorageClass,
2526                    (UInt32)symtab_i->NumberOfAuxSymbols
2527                  );
2528 #        endif
2529       }
2530
2531       i += symtab_i->NumberOfAuxSymbols;
2532       i++;
2533    }
2534
2535    return 1;
2536 }
2537
2538
2539 static int
2540 ocResolve_PEi386 ( ObjectCode* oc )
2541 {
2542    COFF_header*  hdr;
2543    COFF_section* sectab;
2544    COFF_symbol*  symtab;
2545    UChar*        strtab;
2546
2547    UInt32        A;
2548    UInt32        S;
2549    UInt32*       pP;
2550
2551    int i;
2552    UInt32 j, noRelocs;
2553
2554    /* ToDo: should be variable-sized?  But is at least safe in the
2555       sense of buffer-overrun-proof. */
2556    char symbol[1000];
2557    /* debugBelch("resolving for %s\n", oc->fileName); */
2558
2559    hdr = (COFF_header*)(oc->image);
2560    sectab = (COFF_section*) (
2561                ((UChar*)(oc->image))
2562                + sizeof_COFF_header + hdr->SizeOfOptionalHeader
2563             );
2564    symtab = (COFF_symbol*) (
2565                ((UChar*)(oc->image))
2566                + hdr->PointerToSymbolTable
2567             );
2568    strtab = ((UChar*)(oc->image))
2569             + hdr->PointerToSymbolTable
2570             + hdr->NumberOfSymbols * sizeof_COFF_symbol;
2571
2572    for (i = 0; i < hdr->NumberOfSections; i++) {
2573       COFF_section* sectab_i
2574          = (COFF_section*)
2575            myindex ( sizeof_COFF_section, sectab, i );
2576       COFF_reloc* reltab
2577          = (COFF_reloc*) (
2578               ((UChar*)(oc->image)) + sectab_i->PointerToRelocations
2579            );
2580
2581       /* Ignore sections called which contain stabs debugging
2582          information. */
2583       if (0 == strcmp(".stab", sectab_i->Name)
2584           || 0 == strcmp(".stabstr", sectab_i->Name)
2585           || 0 == strcmp(".ctors", sectab_i->Name))
2586          continue;
2587
2588       if ( sectab_i->Characteristics & MYIMAGE_SCN_LNK_NRELOC_OVFL ) {
2589         /* If the relocation field (a short) has overflowed, the
2590          * real count can be found in the first reloc entry.
2591          *
2592          * See Section 4.1 (last para) of the PE spec (rev6.0).
2593          *
2594          * Nov2003 update: the GNU linker still doesn't correctly
2595          * handle the generation of relocatable object files with
2596          * overflown relocations. Hence the output to warn of potential
2597          * troubles.
2598          */
2599         COFF_reloc* rel = (COFF_reloc*)
2600                            myindex ( sizeof_COFF_reloc, reltab, 0 );
2601         noRelocs = rel->VirtualAddress;
2602
2603         /* 10/05: we now assume (and check for) a GNU ld that is capable
2604          * of handling object files with (>2^16) of relocs.
2605          */
2606 #if 0
2607         debugBelch("WARNING: Overflown relocation field (# relocs found: %u)\n",
2608                    noRelocs);
2609 #endif
2610         j = 1;
2611       } else {
2612         noRelocs = sectab_i->NumberOfRelocations;
2613         j = 0;
2614       }
2615
2616
2617       for (; j < noRelocs; j++) {
2618          COFF_symbol* sym;
2619          COFF_reloc* reltab_j
2620             = (COFF_reloc*)
2621               myindex ( sizeof_COFF_reloc, reltab, j );
2622
2623          /* the location to patch */
2624          pP = (UInt32*)(
2625                  ((UChar*)(oc->image))
2626                  + (sectab_i->PointerToRawData
2627                     + reltab_j->VirtualAddress
2628                     - sectab_i->VirtualAddress )
2629               );
2630          /* the existing contents of pP */
2631          A = *pP;
2632          /* the symbol to connect to */
2633          sym = (COFF_symbol*)
2634                myindex ( sizeof_COFF_symbol,
2635                          symtab, reltab_j->SymbolTableIndex );
2636          IF_DEBUG(linker,
2637                   debugBelch(
2638                             "reloc sec %2d num %3d:  type 0x%-4x   "
2639                             "vaddr 0x%-8x   name `",
2640                             i, j,
2641                             (UInt32)reltab_j->Type,
2642                             reltab_j->VirtualAddress );
2643                             printName ( sym->Name, strtab );
2644                             debugBelch("'\n" ));
2645
2646          if (sym->StorageClass == MYIMAGE_SYM_CLASS_STATIC) {
2647             COFF_section* section_sym
2648                = findPEi386SectionCalled ( oc, sym->Name );
2649             if (!section_sym) {
2650                errorBelch("%s: can't find section `%s'", oc->fileName, sym->Name);
2651                return 0;
2652             }
2653             S = ((UInt32)(oc->image))
2654                 + (section_sym->PointerToRawData
2655                    + sym->Value);
2656          } else {
2657             copyName ( sym->Name, strtab, symbol, 1000-1 );
2658             S = (UInt32) lookupSymbol( symbol );
2659             if ((void*)S != NULL) goto foundit;
2660             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
2661             return 0;
2662            foundit:;
2663          }
2664          checkProddableBlock(oc, pP);
2665          switch (reltab_j->Type) {
2666             case MYIMAGE_REL_I386_DIR32:
2667                *pP = A + S;
2668                break;
2669             case MYIMAGE_REL_I386_REL32:
2670                /* Tricky.  We have to insert a displacement at
2671                   pP which, when added to the PC for the _next_
2672                   insn, gives the address of the target (S).
2673                   Problem is to know the address of the next insn
2674                   when we only know pP.  We assume that this
2675                   literal field is always the last in the insn,
2676                   so that the address of the next insn is pP+4
2677                   -- hence the constant 4.
2678                   Also I don't know if A should be added, but so
2679                   far it has always been zero.
2680
2681                   SOF 05/2005: 'A' (old contents of *pP) have been observed
2682                   to contain values other than zero (the 'wx' object file
2683                   that came with wxhaskell-0.9.4; dunno how it was compiled..).
2684                   So, add displacement to old value instead of asserting
2685                   A to be zero. Fixes wxhaskell-related crashes, and no other
2686                   ill effects have been observed.
2687
2688                   Update: the reason why we're seeing these more elaborate
2689                   relocations is due to a switch in how the NCG compiles SRTs
2690                   and offsets to them from info tables. SRTs live in .(ro)data,
2691                   while info tables live in .text, causing GAS to emit REL32/DISP32
2692                   relocations with non-zero values. Adding the displacement is
2693                   the right thing to do.
2694                */
2695                *pP = S - ((UInt32)pP) - 4 + A;
2696                break;
2697             default:
2698                debugBelch("%s: unhandled PEi386 relocation type %d",
2699                      oc->fileName, reltab_j->Type);
2700                return 0;
2701          }
2702
2703       }
2704    }
2705
2706    IF_DEBUG(linker, debugBelch("completed %s", oc->fileName));
2707    return 1;
2708 }
2709
2710 #endif /* defined(OBJFORMAT_PEi386) */
2711
2712
2713 /* --------------------------------------------------------------------------
2714  * ELF specifics
2715  * ------------------------------------------------------------------------*/
2716
2717 #if defined(OBJFORMAT_ELF)
2718
2719 #define FALSE 0
2720 #define TRUE  1
2721
2722 #if defined(sparc_HOST_ARCH)
2723 #  define ELF_TARGET_SPARC  /* Used inside <elf.h> */
2724 #elif defined(i386_HOST_ARCH)
2725 #  define ELF_TARGET_386    /* Used inside <elf.h> */
2726 #elif defined(x86_64_HOST_ARCH)
2727 #  define ELF_TARGET_X64_64
2728 #  define ELF_64BIT
2729 #elif defined (ia64_HOST_ARCH)
2730 #  define ELF_TARGET_IA64   /* Used inside <elf.h> */
2731 #  define ELF_64BIT
2732 #  define ELF_FUNCTION_DESC /* calling convention uses function descriptors */
2733 #  define ELF_NEED_GOT      /* needs Global Offset Table */
2734 #  define ELF_NEED_PLT      /* needs Procedure Linkage Tables */
2735 #endif
2736
2737 #if !defined(openbsd_HOST_OS)
2738 #  include <elf.h>
2739 #else
2740 /* openbsd elf has things in different places, with diff names */
2741 #  include <elf_abi.h>
2742 #  include <machine/reloc.h>
2743 #  define R_386_32    RELOC_32
2744 #  define R_386_PC32  RELOC_PC32
2745 #endif
2746
2747 /* If elf.h doesn't define it */
2748 #  ifndef R_X86_64_PC64
2749 #    define R_X86_64_PC64 24
2750 #  endif
2751
2752 /*
2753  * Define a set of types which can be used for both ELF32 and ELF64
2754  */
2755
2756 #ifdef ELF_64BIT
2757 #define ELFCLASS    ELFCLASS64
2758 #define Elf_Addr    Elf64_Addr
2759 #define Elf_Word    Elf64_Word
2760 #define Elf_Sword   Elf64_Sword
2761 #define Elf_Ehdr    Elf64_Ehdr
2762 #define Elf_Phdr    Elf64_Phdr
2763 #define Elf_Shdr    Elf64_Shdr
2764 #define Elf_Sym     Elf64_Sym
2765 #define Elf_Rel     Elf64_Rel
2766 #define Elf_Rela    Elf64_Rela
2767 #define ELF_ST_TYPE ELF64_ST_TYPE
2768 #define ELF_ST_BIND ELF64_ST_BIND
2769 #define ELF_R_TYPE  ELF64_R_TYPE
2770 #define ELF_R_SYM   ELF64_R_SYM
2771 #else
2772 #define ELFCLASS    ELFCLASS32
2773 #define Elf_Addr    Elf32_Addr
2774 #define Elf_Word    Elf32_Word
2775 #define Elf_Sword   Elf32_Sword
2776 #define Elf_Ehdr    Elf32_Ehdr
2777 #define Elf_Phdr    Elf32_Phdr
2778 #define Elf_Shdr    Elf32_Shdr
2779 #define Elf_Sym     Elf32_Sym
2780 #define Elf_Rel     Elf32_Rel
2781 #define Elf_Rela    Elf32_Rela
2782 #ifndef ELF_ST_TYPE
2783 #define ELF_ST_TYPE ELF32_ST_TYPE
2784 #endif
2785 #ifndef ELF_ST_BIND
2786 #define ELF_ST_BIND ELF32_ST_BIND
2787 #endif
2788 #ifndef ELF_R_TYPE
2789 #define ELF_R_TYPE  ELF32_R_TYPE
2790 #endif
2791 #ifndef ELF_R_SYM
2792 #define ELF_R_SYM   ELF32_R_SYM
2793 #endif
2794 #endif
2795
2796
2797 /*
2798  * Functions to allocate entries in dynamic sections.  Currently we simply
2799  * preallocate a large number, and we don't check if a entry for the given
2800  * target already exists (a linear search is too slow).  Ideally these
2801  * entries would be associated with symbols.
2802  */
2803
2804 /* These sizes sufficient to load HSbase + HShaskell98 + a few modules */
2805 #define GOT_SIZE            0x20000
2806 #define FUNCTION_TABLE_SIZE 0x10000
2807 #define PLT_SIZE            0x08000
2808
2809 #ifdef ELF_NEED_GOT
2810 static Elf_Addr got[GOT_SIZE];
2811 static unsigned int gotIndex;
2812 static Elf_Addr gp_val = (Elf_Addr)got;
2813
2814 static Elf_Addr
2815 allocateGOTEntry(Elf_Addr target)
2816 {
2817    Elf_Addr *entry;
2818
2819    if (gotIndex >= GOT_SIZE)
2820       barf("Global offset table overflow");
2821
2822    entry = &got[gotIndex++];
2823    *entry = target;
2824    return (Elf_Addr)entry;
2825 }
2826 #endif
2827
2828 #ifdef ELF_FUNCTION_DESC
2829 typedef struct {
2830    Elf_Addr ip;
2831    Elf_Addr gp;
2832 } FunctionDesc;
2833
2834 static FunctionDesc functionTable[FUNCTION_TABLE_SIZE];
2835 static unsigned int functionTableIndex;
2836
2837 static Elf_Addr
2838 allocateFunctionDesc(Elf_Addr target)
2839 {
2840    FunctionDesc *entry;
2841
2842    if (functionTableIndex >= FUNCTION_TABLE_SIZE)
2843       barf("Function table overflow");
2844
2845    entry = &functionTable[functionTableIndex++];
2846    entry->ip = target;
2847    entry->gp = (Elf_Addr)gp_val;
2848    return (Elf_Addr)entry;
2849 }
2850
2851 static Elf_Addr
2852 copyFunctionDesc(Elf_Addr target)
2853 {
2854    FunctionDesc *olddesc = (FunctionDesc *)target;
2855    FunctionDesc *newdesc;
2856
2857    newdesc = (FunctionDesc *)allocateFunctionDesc(olddesc->ip);
2858    newdesc->gp = olddesc->gp;
2859    return (Elf_Addr)newdesc;
2860 }
2861 #endif
2862
2863 #ifdef ELF_NEED_PLT
2864 #ifdef ia64_HOST_ARCH
2865 static void ia64_reloc_gprel22(Elf_Addr target, Elf_Addr value);
2866 static void ia64_reloc_pcrel21(Elf_Addr target, Elf_Addr value, ObjectCode *oc);
2867
2868 static unsigned char plt_code[] =
2869 {
2870    /* taken from binutils bfd/elfxx-ia64.c */
2871    0x0b, 0x78, 0x00, 0x02, 0x00, 0x24,  /*   [MMI]       addl r15=0,r1;;    */
2872    0x00, 0x41, 0x3c, 0x30, 0x28, 0xc0,  /*               ld8 r16=[r15],8    */
2873    0x01, 0x08, 0x00, 0x84,              /*               mov r14=r1;;       */
2874    0x11, 0x08, 0x00, 0x1e, 0x18, 0x10,  /*   [MIB]       ld8 r1=[r15]       */
2875    0x60, 0x80, 0x04, 0x80, 0x03, 0x00,  /*               mov b6=r16         */
2876    0x60, 0x00, 0x80, 0x00               /*               br.few b6;;        */
2877 };
2878
2879 /* If we can't get to the function descriptor via gp, take a local copy of it */
2880 #define PLT_RELOC(code, target) { \
2881    Elf64_Sxword rel_value = target - gp_val; \
2882    if ((rel_value > 0x1fffff) || (rel_value < -0x1fffff)) \
2883       ia64_reloc_gprel22((Elf_Addr)code, copyFunctionDesc(target)); \
2884    else \
2885       ia64_reloc_gprel22((Elf_Addr)code, target); \
2886    }
2887 #endif
2888
2889 typedef struct {
2890    unsigned char code[sizeof(plt_code)];
2891 } PLTEntry;
2892
2893 static Elf_Addr
2894 allocatePLTEntry(Elf_Addr target, ObjectCode *oc)
2895 {
2896    PLTEntry *plt = (PLTEntry *)oc->plt;
2897    PLTEntry *entry;
2898
2899    if (oc->pltIndex >= PLT_SIZE)
2900       barf("Procedure table overflow");
2901
2902    entry = &plt[oc->pltIndex++];
2903    memcpy(entry->code, plt_code, sizeof(entry->code));
2904    PLT_RELOC(entry->code, target);
2905    return (Elf_Addr)entry;
2906 }
2907
2908 static unsigned int
2909 PLTSize(void)
2910 {
2911    return (PLT_SIZE * sizeof(PLTEntry));
2912 }
2913 #endif
2914
2915
2916 /*
2917  * Generic ELF functions
2918  */
2919
2920 static char *
2921 findElfSection ( void* objImage, Elf_Word sh_type )
2922 {
2923    char* ehdrC = (char*)objImage;
2924    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
2925    Elf_Shdr* shdr = (Elf_Shdr*)(ehdrC + ehdr->e_shoff);
2926    char* sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
2927    char* ptr = NULL;
2928    int i;
2929
2930    for (i = 0; i < ehdr->e_shnum; i++) {
2931       if (shdr[i].sh_type == sh_type
2932           /* Ignore the section header's string table. */
2933           && i != ehdr->e_shstrndx
2934           /* Ignore string tables named .stabstr, as they contain
2935              debugging info. */
2936           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
2937          ) {
2938          ptr = ehdrC + shdr[i].sh_offset;
2939          break;
2940       }
2941    }
2942    return ptr;
2943 }
2944
2945 #if defined(ia64_HOST_ARCH)
2946 static Elf_Addr
2947 findElfSegment ( void* objImage, Elf_Addr vaddr )
2948 {
2949    char* ehdrC = (char*)objImage;
2950    Elf_Ehdr* ehdr = (Elf_Ehdr*)ehdrC;
2951    Elf_Phdr* phdr = (Elf_Phdr*)(ehdrC + ehdr->e_phoff);
2952    Elf_Addr segaddr = 0;
2953    int i;
2954
2955    for (i = 0; i < ehdr->e_phnum; i++) {
2956       segaddr = phdr[i].p_vaddr;
2957       if ((vaddr >= segaddr) && (vaddr < segaddr + phdr[i].p_memsz))
2958               break;
2959    }
2960    return segaddr;
2961 }
2962 #endif
2963
2964 static int
2965 ocVerifyImage_ELF ( ObjectCode* oc )
2966 {
2967    Elf_Shdr* shdr;
2968    Elf_Sym*  stab;
2969    int i, j, nent, nstrtab, nsymtabs;
2970    char* sh_strtab;
2971    char* strtab;
2972
2973    char*     ehdrC = (char*)(oc->image);
2974    Elf_Ehdr* ehdr  = (Elf_Ehdr*)ehdrC;
2975
2976    if (ehdr->e_ident[EI_MAG0] != ELFMAG0 ||
2977        ehdr->e_ident[EI_MAG1] != ELFMAG1 ||
2978        ehdr->e_ident[EI_MAG2] != ELFMAG2 ||
2979        ehdr->e_ident[EI_MAG3] != ELFMAG3) {
2980       errorBelch("%s: not an ELF object", oc->fileName);
2981       return 0;
2982    }
2983
2984    if (ehdr->e_ident[EI_CLASS] != ELFCLASS) {
2985       errorBelch("%s: unsupported ELF format", oc->fileName);
2986       return 0;
2987    }
2988
2989    if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
2990        IF_DEBUG(linker,debugBelch( "Is little-endian\n" ));
2991    } else
2992    if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
2993        IF_DEBUG(linker,debugBelch( "Is big-endian\n" ));
2994    } else {
2995        errorBelch("%s: unknown endiannness", oc->fileName);
2996        return 0;
2997    }
2998
2999    if (ehdr->e_type != ET_REL) {
3000       errorBelch("%s: not a relocatable object (.o) file", oc->fileName);
3001       return 0;
3002    }
3003    IF_DEBUG(linker, debugBelch( "Is a relocatable object (.o) file\n" ));
3004
3005    IF_DEBUG(linker,debugBelch( "Architecture is " ));
3006    switch (ehdr->e_machine) {
3007       case EM_386:   IF_DEBUG(linker,debugBelch( "x86" )); break;
3008 #ifdef EM_SPARC32PLUS
3009       case EM_SPARC32PLUS:
3010 #endif
3011       case EM_SPARC: IF_DEBUG(linker,debugBelch( "sparc" )); break;
3012 #ifdef EM_IA_64
3013       case EM_IA_64: IF_DEBUG(linker,debugBelch( "ia64" )); break;
3014 #endif
3015       case EM_PPC:   IF_DEBUG(linker,debugBelch( "powerpc32" )); break;
3016 #ifdef EM_X86_64
3017       case EM_X86_64: IF_DEBUG(linker,debugBelch( "x86_64" )); break;
3018 #elif defined(EM_AMD64)
3019       case EM_AMD64: IF_DEBUG(linker,debugBelch( "amd64" )); break;
3020 #endif
3021       default:       IF_DEBUG(linker,debugBelch( "unknown" ));
3022                      errorBelch("%s: unknown architecture (e_machine == %d)"
3023                                 , oc->fileName, ehdr->e_machine);
3024                      return 0;
3025    }
3026
3027    IF_DEBUG(linker,debugBelch(
3028              "\nSection header table: start %ld, n_entries %d, ent_size %d\n",
3029              (long)ehdr->e_shoff, ehdr->e_shnum, ehdr->e_shentsize  ));
3030
3031    ASSERT (ehdr->e_shentsize == sizeof(Elf_Shdr));
3032
3033    shdr = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3034
3035    if (ehdr->e_shstrndx == SHN_UNDEF) {
3036       errorBelch("%s: no section header string table", oc->fileName);
3037       return 0;
3038    } else {
3039       IF_DEBUG(linker,debugBelch( "Section header string table is section %d\n",
3040                           ehdr->e_shstrndx));
3041       sh_strtab = ehdrC + shdr[ehdr->e_shstrndx].sh_offset;
3042    }
3043
3044    for (i = 0; i < ehdr->e_shnum; i++) {
3045       IF_DEBUG(linker,debugBelch("%2d:  ", i ));
3046       IF_DEBUG(linker,debugBelch("type=%2d  ", (int)shdr[i].sh_type ));
3047       IF_DEBUG(linker,debugBelch("size=%4d  ", (int)shdr[i].sh_size ));
3048       IF_DEBUG(linker,debugBelch("offs=%4d  ", (int)shdr[i].sh_offset ));
3049       IF_DEBUG(linker,debugBelch("  (%p .. %p)  ",
3050                ehdrC + shdr[i].sh_offset,
3051                       ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1));
3052
3053       if (shdr[i].sh_type == SHT_REL) {
3054           IF_DEBUG(linker,debugBelch("Rel  " ));
3055       } else if (shdr[i].sh_type == SHT_RELA) {
3056           IF_DEBUG(linker,debugBelch("RelA " ));
3057       } else {
3058           IF_DEBUG(linker,debugBelch("     "));
3059       }
3060       if (sh_strtab) {
3061           IF_DEBUG(linker,debugBelch("sname=%s\n", sh_strtab + shdr[i].sh_name ));
3062       }
3063    }
3064
3065    IF_DEBUG(linker,debugBelch( "\nString tables" ));
3066    strtab = NULL;
3067    nstrtab = 0;
3068    for (i = 0; i < ehdr->e_shnum; i++) {
3069       if (shdr[i].sh_type == SHT_STRTAB
3070           /* Ignore the section header's string table. */
3071           && i != ehdr->e_shstrndx
3072           /* Ignore string tables named .stabstr, as they contain
3073              debugging info. */
3074           && 0 != memcmp(".stabstr", sh_strtab + shdr[i].sh_name, 8)
3075          ) {
3076          IF_DEBUG(linker,debugBelch("   section %d is a normal string table", i ));
3077          strtab = ehdrC + shdr[i].sh_offset;
3078          nstrtab++;
3079       }
3080    }
3081    if (nstrtab != 1) {
3082       errorBelch("%s: no string tables, or too many", oc->fileName);
3083       return 0;
3084    }
3085
3086    nsymtabs = 0;
3087    IF_DEBUG(linker,debugBelch( "\nSymbol tables" ));
3088    for (i = 0; i < ehdr->e_shnum; i++) {
3089       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3090       IF_DEBUG(linker,debugBelch( "section %d is a symbol table\n", i ));
3091       nsymtabs++;
3092       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3093       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3094       IF_DEBUG(linker,debugBelch( "   number of entries is apparently %d (%ld rem)\n",
3095                nent,
3096                (long)shdr[i].sh_size % sizeof(Elf_Sym)
3097              ));
3098       if (0 != shdr[i].sh_size % sizeof(Elf_Sym)) {
3099          errorBelch("%s: non-integral number of symbol table entries", oc->fileName);
3100          return 0;
3101       }
3102       for (j = 0; j < nent; j++) {
3103          IF_DEBUG(linker,debugBelch("   %2d  ", j ));
3104          IF_DEBUG(linker,debugBelch("  sec=%-5d  size=%-3d  val=%5p  ",
3105                              (int)stab[j].st_shndx,
3106                              (int)stab[j].st_size,
3107                              (char*)stab[j].st_value ));
3108
3109          IF_DEBUG(linker,debugBelch("type=" ));
3110          switch (ELF_ST_TYPE(stab[j].st_info)) {
3111             case STT_NOTYPE:  IF_DEBUG(linker,debugBelch("notype " )); break;
3112             case STT_OBJECT:  IF_DEBUG(linker,debugBelch("object " )); break;
3113             case STT_FUNC  :  IF_DEBUG(linker,debugBelch("func   " )); break;
3114             case STT_SECTION: IF_DEBUG(linker,debugBelch("section" )); break;
3115             case STT_FILE:    IF_DEBUG(linker,debugBelch("file   " )); break;
3116             default:          IF_DEBUG(linker,debugBelch("?      " )); break;
3117          }
3118          IF_DEBUG(linker,debugBelch("  " ));
3119
3120          IF_DEBUG(linker,debugBelch("bind=" ));
3121          switch (ELF_ST_BIND(stab[j].st_info)) {
3122             case STB_LOCAL :  IF_DEBUG(linker,debugBelch("local " )); break;
3123             case STB_GLOBAL:  IF_DEBUG(linker,debugBelch("global" )); break;
3124             case STB_WEAK  :  IF_DEBUG(linker,debugBelch("weak  " )); break;
3125             default:          IF_DEBUG(linker,debugBelch("?     " )); break;
3126          }
3127          IF_DEBUG(linker,debugBelch("  " ));
3128
3129          IF_DEBUG(linker,debugBelch("name=%s\n", strtab + stab[j].st_name ));
3130       }
3131    }
3132
3133    if (nsymtabs == 0) {
3134       errorBelch("%s: didn't find any symbol tables", oc->fileName);
3135       return 0;
3136    }
3137
3138    return 1;
3139 }
3140
3141 static int getSectionKind_ELF( Elf_Shdr *hdr, int *is_bss )
3142 {
3143     *is_bss = FALSE;
3144
3145     if (hdr->sh_type == SHT_PROGBITS
3146         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_EXECINSTR)) {
3147         /* .text-style section */
3148         return SECTIONKIND_CODE_OR_RODATA;
3149     }
3150
3151     if (hdr->sh_type == SHT_PROGBITS
3152             && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3153             /* .data-style section */
3154             return SECTIONKIND_RWDATA;
3155     }
3156
3157     if (hdr->sh_type == SHT_PROGBITS
3158         && (hdr->sh_flags & SHF_ALLOC) && !(hdr->sh_flags & SHF_WRITE)) {
3159         /* .rodata-style section */
3160         return SECTIONKIND_CODE_OR_RODATA;
3161     }
3162
3163     if (hdr->sh_type == SHT_NOBITS
3164         && (hdr->sh_flags & SHF_ALLOC) && (hdr->sh_flags & SHF_WRITE)) {
3165         /* .bss-style section */
3166         *is_bss = TRUE;
3167         return SECTIONKIND_RWDATA;
3168     }
3169
3170     return SECTIONKIND_OTHER;
3171 }
3172
3173
3174 static int
3175 ocGetNames_ELF ( ObjectCode* oc )
3176 {
3177    int i, j, k, nent;
3178    Elf_Sym* stab;
3179
3180    char*     ehdrC    = (char*)(oc->image);
3181    Elf_Ehdr* ehdr     = (Elf_Ehdr*)ehdrC;
3182    char*     strtab   = findElfSection ( ehdrC, SHT_STRTAB );
3183    Elf_Shdr* shdr     = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3184
3185    ASSERT(symhash != NULL);
3186
3187    if (!strtab) {
3188       errorBelch("%s: no strtab", oc->fileName);
3189       return 0;
3190    }
3191
3192    k = 0;
3193    for (i = 0; i < ehdr->e_shnum; i++) {
3194       /* Figure out what kind of section it is.  Logic derived from
3195          Figure 1.14 ("Special Sections") of the ELF document
3196          ("Portable Formats Specification, Version 1.1"). */
3197       int         is_bss = FALSE;
3198       SectionKind kind   = getSectionKind_ELF(&shdr[i], &is_bss);
3199
3200       if (is_bss && shdr[i].sh_size > 0) {
3201          /* This is a non-empty .bss section.  Allocate zeroed space for
3202             it, and set its .sh_offset field such that
3203             ehdrC + .sh_offset == addr_of_zeroed_space.  */
3204          char* zspace = stgCallocBytes(1, shdr[i].sh_size,
3205                                        "ocGetNames_ELF(BSS)");
3206          shdr[i].sh_offset = ((char*)zspace) - ((char*)ehdrC);
3207          /*
3208          debugBelch("BSS section at 0x%x, size %d\n",
3209                          zspace, shdr[i].sh_size);
3210          */
3211       }
3212
3213       /* fill in the section info */
3214       if (kind != SECTIONKIND_OTHER && shdr[i].sh_size > 0) {
3215          addProddableBlock(oc, ehdrC + shdr[i].sh_offset, shdr[i].sh_size);
3216          addSection(oc, kind, ehdrC + shdr[i].sh_offset,
3217                         ehdrC + shdr[i].sh_offset + shdr[i].sh_size - 1);
3218       }
3219
3220       if (shdr[i].sh_type != SHT_SYMTAB) continue;
3221
3222       /* copy stuff into this module's object symbol table */
3223       stab = (Elf_Sym*) (ehdrC + shdr[i].sh_offset);
3224       nent = shdr[i].sh_size / sizeof(Elf_Sym);
3225
3226       oc->n_symbols = nent;
3227       oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
3228                                    "ocGetNames_ELF(oc->symbols)");
3229
3230       for (j = 0; j < nent; j++) {
3231
3232          char  isLocal = FALSE; /* avoids uninit-var warning */
3233          char* ad      = NULL;
3234          char* nm      = strtab + stab[j].st_name;
3235          int   secno   = stab[j].st_shndx;
3236
3237          /* Figure out if we want to add it; if so, set ad to its
3238             address.  Otherwise leave ad == NULL. */
3239
3240          if (secno == SHN_COMMON) {
3241             isLocal = FALSE;
3242             ad = stgCallocBytes(1, stab[j].st_size, "ocGetNames_ELF(COMMON)");
3243             /*
3244             debugBelch("COMMON symbol, size %d name %s\n",
3245                             stab[j].st_size, nm);
3246             */
3247             /* Pointless to do addProddableBlock() for this area,
3248                since the linker should never poke around in it. */
3249          }
3250          else
3251          if ( ( ELF_ST_BIND(stab[j].st_info)==STB_GLOBAL
3252                 || ELF_ST_BIND(stab[j].st_info)==STB_LOCAL
3253               )
3254               /* and not an undefined symbol */
3255               && stab[j].st_shndx != SHN_UNDEF
3256               /* and not in a "special section" */
3257               && stab[j].st_shndx < SHN_LORESERVE
3258               &&
3259               /* and it's a not a section or string table or anything silly */
3260               ( ELF_ST_TYPE(stab[j].st_info)==STT_FUNC ||
3261                 ELF_ST_TYPE(stab[j].st_info)==STT_OBJECT ||
3262                 ELF_ST_TYPE(stab[j].st_info)==STT_NOTYPE
3263               )
3264             ) {
3265             /* Section 0 is the undefined section, hence > and not >=. */
3266             ASSERT(secno > 0 && secno < ehdr->e_shnum);
3267             /*
3268             if (shdr[secno].sh_type == SHT_NOBITS) {
3269                debugBelch("   BSS symbol, size %d off %d name %s\n",
3270                                stab[j].st_size, stab[j].st_value, nm);
3271             }
3272             */
3273             ad = ehdrC + shdr[ secno ].sh_offset + stab[j].st_value;
3274             if (ELF_ST_BIND(stab[j].st_info)==STB_LOCAL) {
3275                isLocal = TRUE;
3276             } else {
3277 #ifdef ELF_FUNCTION_DESC
3278                /* dlsym() and the initialisation table both give us function
3279                 * descriptors, so to be consistent we store function descriptors
3280                 * in the symbol table */
3281                if (ELF_ST_TYPE(stab[j].st_info) == STT_FUNC)
3282                    ad = (char *)allocateFunctionDesc((Elf_Addr)ad);
3283 #endif
3284                IF_DEBUG(linker,debugBelch( "addOTabName(GLOB): %10p  %s %s\n",
3285                                       ad, oc->fileName, nm ));
3286                isLocal = FALSE;
3287             }
3288          }
3289
3290          /* And the decision is ... */
3291
3292          if (ad != NULL) {
3293             ASSERT(nm != NULL);
3294             oc->symbols[j] = nm;
3295             /* Acquire! */
3296             if (isLocal) {
3297                /* Ignore entirely. */
3298             } else {
3299                ghciInsertStrHashTable(oc->fileName, symhash, nm, ad);
3300             }
3301          } else {
3302             /* Skip. */
3303             IF_DEBUG(linker,debugBelch( "skipping `%s'\n",
3304                                    strtab + stab[j].st_name ));
3305             /*
3306             debugBelch(
3307                     "skipping   bind = %d,  type = %d,  shndx = %d   `%s'\n",
3308                     (int)ELF_ST_BIND(stab[j].st_info),
3309                     (int)ELF_ST_TYPE(stab[j].st_info),
3310                     (int)stab[j].st_shndx,
3311                     strtab + stab[j].st_name
3312                    );
3313             */
3314             oc->symbols[j] = NULL;
3315          }
3316
3317       }
3318    }
3319
3320    return 1;
3321 }
3322
3323 /* Do ELF relocations which lack an explicit addend.  All x86-linux
3324    relocations appear to be of this form. */
3325 static int
3326 do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC,
3327                          Elf_Shdr* shdr, int shnum,
3328                          Elf_Sym*  stab, char* strtab )
3329 {
3330    int j;
3331    char *symbol;
3332    Elf_Word* targ;
3333    Elf_Rel*  rtab = (Elf_Rel*) (ehdrC + shdr[shnum].sh_offset);
3334    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rel);
3335    int target_shndx = shdr[shnum].sh_info;
3336    int symtab_shndx = shdr[shnum].sh_link;
3337
3338    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
3339    targ  = (Elf_Word*)(ehdrC + shdr[ target_shndx ].sh_offset);
3340    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
3341                           target_shndx, symtab_shndx ));
3342
3343    /* Skip sections that we're not interested in. */
3344    {
3345        int is_bss;
3346        SectionKind kind = getSectionKind_ELF(&shdr[target_shndx], &is_bss);
3347        if (kind == SECTIONKIND_OTHER) {
3348            IF_DEBUG(linker,debugBelch( "skipping (target section not loaded)"));
3349            return 1;
3350        }
3351    }
3352
3353    for (j = 0; j < nent; j++) {
3354       Elf_Addr offset = rtab[j].r_offset;
3355       Elf_Addr info   = rtab[j].r_info;
3356
3357       Elf_Addr  P  = ((Elf_Addr)targ) + offset;
3358       Elf_Word* pP = (Elf_Word*)P;
3359       Elf_Addr  A  = *pP;
3360       Elf_Addr  S;
3361       void*     S_tmp;
3362       Elf_Addr  value;
3363       StgStablePtr stablePtr;
3364       StgPtr stableVal;
3365
3366       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p)",
3367                              j, (void*)offset, (void*)info ));
3368       if (!info) {
3369          IF_DEBUG(linker,debugBelch( " ZERO" ));
3370          S = 0;
3371       } else {
3372          Elf_Sym sym = stab[ELF_R_SYM(info)];
3373          /* First see if it is a local symbol. */
3374          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
3375             /* Yes, so we can get the address directly from the ELF symbol
3376                table. */
3377             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
3378             S = (Elf_Addr)
3379                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
3380                        + stab[ELF_R_SYM(info)].st_value);
3381
3382          } else {
3383             symbol = strtab + sym.st_name;
3384             stablePtr = (StgStablePtr)lookupHashTable(stablehash, (StgWord)symbol);
3385             if (NULL == stablePtr) {
3386               /* No, so look up the name in our global table. */
3387               S_tmp = lookupSymbol( symbol );
3388               S = (Elf_Addr)S_tmp;
3389             } else {
3390               stableVal = deRefStablePtr( stablePtr );
3391               S_tmp = stableVal;
3392               S = (Elf_Addr)S_tmp;
3393             }
3394          }
3395          if (!S) {
3396             errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3397             return 0;
3398          }
3399          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p\n", symbol, (void*)S ));
3400       }
3401
3402       IF_DEBUG(linker,debugBelch( "Reloc: P = %p   S = %p   A = %p\n",
3403                              (void*)P, (void*)S, (void*)A ));
3404       checkProddableBlock ( oc, pP );
3405
3406       value = S + A;
3407
3408       switch (ELF_R_TYPE(info)) {
3409 #        ifdef i386_HOST_ARCH
3410          case R_386_32:   *pP = value;     break;
3411          case R_386_PC32: *pP = value - P; break;
3412 #        endif
3413          default:
3414             errorBelch("%s: unhandled ELF relocation(Rel) type %lu\n",
3415                   oc->fileName, (lnat)ELF_R_TYPE(info));
3416             return 0;
3417       }
3418
3419    }
3420    return 1;
3421 }
3422
3423 /* Do ELF relocations for which explicit addends are supplied.
3424    sparc-solaris relocations appear to be of this form. */
3425 static int
3426 do_Elf_Rela_relocations ( ObjectCode* oc, char* ehdrC,
3427                           Elf_Shdr* shdr, int shnum,
3428                           Elf_Sym*  stab, char* strtab )
3429 {
3430    int j;
3431    char *symbol = NULL;
3432    Elf_Addr targ;
3433    Elf_Rela* rtab = (Elf_Rela*) (ehdrC + shdr[shnum].sh_offset);
3434    int         nent = shdr[shnum].sh_size / sizeof(Elf_Rela);
3435    int target_shndx = shdr[shnum].sh_info;
3436    int symtab_shndx = shdr[shnum].sh_link;
3437
3438    stab  = (Elf_Sym*) (ehdrC + shdr[ symtab_shndx ].sh_offset);
3439    targ  = (Elf_Addr) (ehdrC + shdr[ target_shndx ].sh_offset);
3440    IF_DEBUG(linker,debugBelch( "relocations for section %d using symtab %d\n",
3441                           target_shndx, symtab_shndx ));
3442
3443    for (j = 0; j < nent; j++) {
3444 #if defined(DEBUG) || defined(sparc_HOST_ARCH) || defined(ia64_HOST_ARCH) || defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
3445       /* This #ifdef only serves to avoid unused-var warnings. */
3446       Elf_Addr  offset = rtab[j].r_offset;
3447       Elf_Addr  P      = targ + offset;
3448 #endif
3449       Elf_Addr  info   = rtab[j].r_info;
3450       Elf_Addr  A      = rtab[j].r_addend;
3451       Elf_Addr  S;
3452       void*     S_tmp;
3453       Elf_Addr  value;
3454 #     if defined(sparc_HOST_ARCH)
3455       Elf_Word* pP = (Elf_Word*)P;
3456       Elf_Word  w1, w2;
3457 #     elif defined(ia64_HOST_ARCH)
3458       Elf64_Xword *pP = (Elf64_Xword *)P;
3459       Elf_Addr addr;
3460 #     elif defined(powerpc_HOST_ARCH)
3461       Elf_Sword delta;
3462 #     endif
3463
3464       IF_DEBUG(linker,debugBelch( "Rel entry %3d is raw(%6p %6p %6p)   ",
3465                              j, (void*)offset, (void*)info,
3466                                 (void*)A ));
3467       if (!info) {
3468          IF_DEBUG(linker,debugBelch( " ZERO" ));
3469          S = 0;
3470       } else {
3471          Elf_Sym sym = stab[ELF_R_SYM(info)];
3472          /* First see if it is a local symbol. */
3473          if (ELF_ST_BIND(sym.st_info) == STB_LOCAL) {
3474             /* Yes, so we can get the address directly from the ELF symbol
3475                table. */
3476             symbol = sym.st_name==0 ? "(noname)" : strtab+sym.st_name;
3477             S = (Elf_Addr)
3478                 (ehdrC + shdr[ sym.st_shndx ].sh_offset
3479                        + stab[ELF_R_SYM(info)].st_value);
3480 #ifdef ELF_FUNCTION_DESC
3481             /* Make a function descriptor for this function */
3482             if (S && ELF_ST_TYPE(sym.st_info) == STT_FUNC) {
3483                S = allocateFunctionDesc(S + A);
3484                A = 0;
3485             }
3486 #endif
3487          } else {
3488             /* No, so look up the name in our global table. */
3489             symbol = strtab + sym.st_name;
3490             S_tmp = lookupSymbol( symbol );
3491             S = (Elf_Addr)S_tmp;
3492
3493 #ifdef ELF_FUNCTION_DESC
3494             /* If a function, already a function descriptor - we would
3495                have to copy it to add an offset. */
3496             if (S && (ELF_ST_TYPE(sym.st_info) == STT_FUNC) && (A != 0))
3497                errorBelch("%s: function %s with addend %p", oc->fileName, symbol, (void *)A);
3498 #endif
3499          }
3500          if (!S) {
3501            errorBelch("%s: unknown symbol `%s'", oc->fileName, symbol);
3502            return 0;
3503          }
3504          IF_DEBUG(linker,debugBelch( "`%s' resolves to %p", symbol, (void*)S ));
3505       }
3506
3507       IF_DEBUG(linker,debugBelch("Reloc: P = %p   S = %p   A = %p\n",
3508                                         (void*)P, (void*)S, (void*)A ));
3509       /* checkProddableBlock ( oc, (void*)P ); */
3510
3511       value = S + A;
3512
3513       switch (ELF_R_TYPE(info)) {
3514 #        if defined(sparc_HOST_ARCH)
3515          case R_SPARC_WDISP30:
3516             w1 = *pP & 0xC0000000;
3517             w2 = (Elf_Word)((value - P) >> 2);
3518             ASSERT((w2 & 0xC0000000) == 0);
3519             w1 |= w2;
3520             *pP = w1;
3521             break;
3522          case R_SPARC_HI22:
3523             w1 = *pP & 0xFFC00000;
3524             w2 = (Elf_Word)(value >> 10);
3525             ASSERT((w2 & 0xFFC00000) == 0);
3526             w1 |= w2;
3527             *pP = w1;
3528             break;
3529          case R_SPARC_LO10:
3530             w1 = *pP & ~0x3FF;
3531             w2 = (Elf_Word)(value & 0x3FF);
3532             ASSERT((w2 & ~0x3FF) == 0);
3533             w1 |= w2;
3534             *pP = w1;
3535             break;
3536          /* According to the Sun documentation:
3537             R_SPARC_UA32
3538             This relocation type resembles R_SPARC_32, except it refers to an
3539             unaligned word. That is, the word to be relocated must be treated
3540             as four separate bytes with arbitrary alignment, not as a word
3541             aligned according to the architecture requirements.
3542
3543             (JRS: which means that freeloading on the R_SPARC_32 case
3544             is probably wrong, but hey ...)
3545          */
3546          case R_SPARC_UA32:
3547          case R_SPARC_32:
3548             w2 = (Elf_Word)value;
3549             *pP = w2;
3550             break;
3551 #        elif defined(ia64_HOST_ARCH)
3552          case R_IA64_DIR64LSB:
3553          case R_IA64_FPTR64LSB:
3554             *pP = value;
3555             break;
3556          case R_IA64_PCREL64LSB:
3557             *pP = value - P;
3558             break;
3559          case R_IA64_SEGREL64LSB:
3560             addr = findElfSegment(ehdrC, value);
3561             *pP = value - addr;
3562             break;
3563          case R_IA64_GPREL22:
3564             ia64_reloc_gprel22(P, value);
3565             break;
3566          case R_IA64_LTOFF22:
3567          case R_IA64_LTOFF22X:
3568          case R_IA64_LTOFF_FPTR22:
3569             addr = allocateGOTEntry(value);
3570             ia64_reloc_gprel22(P, addr);
3571             break;
3572          case R_IA64_PCREL21B:
3573             ia64_reloc_pcrel21(P, S, oc);
3574             break;
3575          case R_IA64_LDXMOV:
3576             /* This goes with R_IA64_LTOFF22X and points to the load to
3577              * convert into a move.  We don't implement relaxation. */
3578             break;
3579 #        elif defined(powerpc_HOST_ARCH)
3580          case R_PPC_ADDR16_LO:
3581             *(Elf32_Half*) P = value;
3582             break;
3583
3584          case R_PPC_ADDR16_HI:
3585             *(Elf32_Half*) P = value >> 16;
3586             break;
3587
3588          case R_PPC_ADDR16_HA:
3589             *(Elf32_Half*) P = (value + 0x8000) >> 16;
3590             break;
3591
3592          case R_PPC_ADDR32:
3593             *(Elf32_Word *) P = value;
3594             break;
3595
3596          case R_PPC_REL32:
3597             *(Elf32_Word *) P = value - P;
3598             break;
3599
3600          case R_PPC_REL24:
3601             delta = value - P;
3602
3603             if( delta << 6 >> 6 != delta )
3604             {
3605                value = (Elf_Addr) (&makeSymbolExtra( oc, ELF_R_SYM(info), value )
3606                                         ->jumpIsland);
3607                delta = value - P;
3608
3609                if( value == 0 || delta << 6 >> 6 != delta )
3610                {
3611                   barf( "Unable to make SymbolExtra for #%d",
3612                         ELF_R_SYM(info) );
3613                   return 0;
3614                }
3615             }
3616
3617             *(Elf_Word *) P = (*(Elf_Word *) P & 0xfc000003)
3618                                           | (delta & 0x3fffffc);
3619             break;
3620 #        endif
3621
3622 #if x86_64_HOST_ARCH
3623       case R_X86_64_64:
3624           *(Elf64_Xword *)P = value;
3625           break;
3626
3627       case R_X86_64_PC32:
3628       {
3629           StgInt64 off = value - P;
3630           if (off >= 0x7fffffffL || off < -0x80000000L) {
3631 #if X86_64_ELF_NONPIC_HACK
3632               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3633                                                 -> jumpIsland;
3634               off = pltAddress + A - P;
3635 #else
3636               barf("R_X86_64_PC32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
3637                    symbol, off, oc->fileName );
3638 #endif
3639           }
3640           *(Elf64_Word *)P = (Elf64_Word)off;
3641           break;
3642       }
3643
3644       case R_X86_64_PC64:
3645       {
3646           StgInt64 off = value - P;
3647           *(Elf64_Word *)P = (Elf64_Word)off;
3648           break;
3649       }
3650
3651       case R_X86_64_32:
3652           if (value >= 0x7fffffffL) {
3653 #if X86_64_ELF_NONPIC_HACK
3654               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3655                                                 -> jumpIsland;
3656               value = pltAddress + A;
3657 #else
3658               barf("R_X86_64_32 relocation out of range: %s = %p\nRecompile %s with -fPIC.",
3659                    symbol, value, oc->fileName );
3660 #endif
3661           }
3662           *(Elf64_Word *)P = (Elf64_Word)value;
3663           break;
3664
3665       case R_X86_64_32S:
3666           if ((StgInt64)value > 0x7fffffffL || (StgInt64)value < -0x80000000L) {
3667 #if X86_64_ELF_NONPIC_HACK
3668               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3669                                                 -> jumpIsland;
3670               value = pltAddress + A;
3671 #else
3672               barf("R_X86_64_32S relocation out of range: %s = %p\nRecompile %s with -fPIC.",
3673                    symbol, value, oc->fileName );
3674 #endif
3675           }
3676           *(Elf64_Sword *)P = (Elf64_Sword)value;
3677           break;
3678
3679       case R_X86_64_GOTPCREL:
3680       {
3681           StgInt64 gotAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)->addr;
3682           StgInt64 off = gotAddress + A - P;
3683           *(Elf64_Word *)P = (Elf64_Word)off;
3684           break;
3685       }
3686
3687       case R_X86_64_PLT32:
3688       {
3689           StgInt64 off = value - P;
3690           if (off >= 0x7fffffffL || off < -0x80000000L) {
3691               StgInt64 pltAddress = (StgInt64) &makeSymbolExtra(oc, ELF_R_SYM(info), S)
3692                                                     -> jumpIsland;
3693               off = pltAddress + A - P;
3694           }
3695           *(Elf64_Word *)P = (Elf64_Word)off;
3696           break;
3697       }
3698 #endif
3699
3700          default:
3701             errorBelch("%s: unhandled ELF relocation(RelA) type %lu\n",
3702                   oc->fileName, (lnat)ELF_R_TYPE(info));
3703             return 0;
3704       }
3705
3706    }
3707    return 1;
3708 }
3709
3710 static int
3711 ocResolve_ELF ( ObjectCode* oc )
3712 {
3713    char *strtab;
3714    int   shnum, ok;
3715    Elf_Sym*  stab  = NULL;
3716    char*     ehdrC = (char*)(oc->image);
3717    Elf_Ehdr* ehdr  = (Elf_Ehdr*) ehdrC;
3718    Elf_Shdr* shdr  = (Elf_Shdr*) (ehdrC + ehdr->e_shoff);
3719
3720    /* first find "the" symbol table */
3721    stab = (Elf_Sym*) findElfSection ( ehdrC, SHT_SYMTAB );
3722
3723    /* also go find the string table */
3724    strtab = findElfSection ( ehdrC, SHT_STRTAB );
3725
3726    if (stab == NULL || strtab == NULL) {
3727       errorBelch("%s: can't find string or symbol table", oc->fileName);
3728       return 0;
3729    }
3730
3731    /* Process the relocation sections. */
3732    for (shnum = 0; shnum < ehdr->e_shnum; shnum++) {
3733       if (shdr[shnum].sh_type == SHT_REL) {
3734          ok = do_Elf_Rel_relocations ( oc, ehdrC, shdr,
3735                                        shnum, stab, strtab );
3736          if (!ok) return ok;
3737       }
3738       else
3739       if (shdr[shnum].sh_type == SHT_RELA) {
3740          ok = do_Elf_Rela_relocations ( oc, ehdrC, shdr,
3741                                         shnum, stab, strtab );
3742          if (!ok) return ok;
3743       }
3744    }
3745
3746 #if defined(powerpc_HOST_ARCH)
3747    ocFlushInstructionCache( oc );
3748 #endif
3749
3750    return 1;
3751 }
3752
3753 /*
3754  * IA64 specifics
3755  * Instructions are 41 bits long, packed into 128 bit bundles with a 5-bit template
3756  * at the front.  The following utility functions pack and unpack instructions, and
3757  * take care of the most common relocations.
3758  */
3759
3760 #ifdef ia64_HOST_ARCH
3761
3762 static Elf64_Xword
3763 ia64_extract_instruction(Elf64_Xword *target)
3764 {
3765    Elf64_Xword w1, w2;
3766    int slot = (Elf_Addr)target & 3;
3767    target = (Elf_Addr)target & ~3;
3768
3769    w1 = *target;
3770    w2 = *(target+1);
3771
3772    switch (slot)
3773    {
3774       case 0:
3775          return ((w1 >> 5) & 0x1ffffffffff);
3776       case 1:
3777          return (w1 >> 46) | ((w2 & 0x7fffff) << 18);
3778       case 2:
3779          return (w2 >> 23);
3780       default:
3781          barf("ia64_extract_instruction: invalid slot %p", target);
3782    }
3783 }
3784
3785 static void
3786 ia64_deposit_instruction(Elf64_Xword *target, Elf64_Xword value)
3787 {
3788    int slot = (Elf_Addr)target & 3;
3789    target = (Elf_Addr)target & ~3;
3790
3791    switch (slot)
3792    {
3793       case 0:
3794          *target |= value << 5;
3795          break;
3796       case 1:
3797          *target |= value << 46;
3798          *(target+1) |= value >> 18;
3799          break;
3800       case 2:
3801          *(target+1) |= value << 23;
3802          break;
3803    }
3804 }
3805
3806 static void
3807 ia64_reloc_gprel22(Elf_Addr target, Elf_Addr value)
3808 {
3809    Elf64_Xword instruction;
3810    Elf64_Sxword rel_value;
3811
3812    rel_value = value - gp_val;
3813    if ((rel_value > 0x1fffff) || (rel_value < -0x1fffff))
3814       barf("GP-relative data out of range (address = 0x%lx, gp = 0x%lx)", value, gp_val);
3815
3816    instruction = ia64_extract_instruction((Elf64_Xword *)target);
3817    instruction |= (((rel_value >> 0) & 0x07f) << 13)            /* imm7b */
3818                     | (((rel_value >> 7) & 0x1ff) << 27)        /* imm9d */
3819                     | (((rel_value >> 16) & 0x01f) << 22)       /* imm5c */
3820                     | ((Elf64_Xword)(rel_value < 0) << 36);     /* s */
3821    ia64_deposit_instruction((Elf64_Xword *)target, instruction);
3822 }
3823
3824 static void
3825 ia64_reloc_pcrel21(Elf_Addr target, Elf_Addr value, ObjectCode *oc)
3826 {
3827    Elf64_Xword instruction;
3828    Elf64_Sxword rel_value;
3829    Elf_Addr entry;
3830
3831    entry = allocatePLTEntry(value, oc);
3832
3833    rel_value = (entry >> 4) - (target >> 4);
3834    if ((rel_value > 0xfffff) || (rel_value < -0xfffff))
3835       barf("PLT entry too far away (entry = 0x%lx, target = 0x%lx)", entry, target);
3836
3837    instruction = ia64_extract_instruction((Elf64_Xword *)target);
3838    instruction |= ((rel_value & 0xfffff) << 13)                 /* imm20b */
3839                     | ((Elf64_Xword)(rel_value < 0) << 36);     /* s */
3840    ia64_deposit_instruction((Elf64_Xword *)target, instruction);
3841 }
3842
3843 #endif /* ia64 */
3844
3845 /*
3846  * PowerPC & X86_64 ELF specifics
3847  */
3848
3849 #if defined(powerpc_HOST_ARCH) || defined(x86_64_HOST_ARCH)
3850
3851 static int ocAllocateSymbolExtras_ELF( ObjectCode *oc )
3852 {
3853   Elf_Ehdr *ehdr;
3854   Elf_Shdr* shdr;
3855   int i;
3856
3857   ehdr = (Elf_Ehdr *) oc->image;
3858   shdr = (Elf_Shdr *) ( ((char *)oc->image) + ehdr->e_shoff );
3859
3860   for( i = 0; i < ehdr->e_shnum; i++ )
3861     if( shdr[i].sh_type == SHT_SYMTAB )
3862       break;
3863
3864   if( i == ehdr->e_shnum )
3865   {
3866     errorBelch( "This ELF file contains no symtab" );
3867     return 0;
3868   }
3869
3870   if( shdr[i].sh_entsize != sizeof( Elf_Sym ) )
3871   {
3872     errorBelch( "The entry size (%d) of the symtab isn't %d\n",
3873       (int) shdr[i].sh_entsize, (int) sizeof( Elf_Sym ) );
3874
3875     return 0;
3876   }
3877
3878   return ocAllocateSymbolExtras( oc, shdr[i].sh_size / sizeof( Elf_Sym ), 0 );
3879 }
3880
3881 #endif /* powerpc */
3882
3883 #endif /* ELF */
3884
3885 /* --------------------------------------------------------------------------
3886  * Mach-O specifics
3887  * ------------------------------------------------------------------------*/
3888
3889 #if defined(OBJFORMAT_MACHO)
3890
3891 /*
3892   Support for MachO linking on Darwin/MacOS X
3893   by Wolfgang Thaller (wolfgang.thaller@gmx.net)
3894
3895   I hereby formally apologize for the hackish nature of this code.
3896   Things that need to be done:
3897   *) implement ocVerifyImage_MachO
3898   *) add still more sanity checks.
3899 */
3900
3901 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
3902 #define mach_header mach_header_64
3903 #define segment_command segment_command_64
3904 #define section section_64
3905 #define nlist nlist_64
3906 #endif
3907
3908 #ifdef powerpc_HOST_ARCH
3909 static int ocAllocateSymbolExtras_MachO(ObjectCode* oc)
3910 {
3911     struct mach_header *header = (struct mach_header *) oc->image;
3912     struct load_command *lc = (struct load_command *) (header + 1);
3913     unsigned i;
3914
3915     for( i = 0; i < header->ncmds; i++ )
3916     {
3917         if( lc->cmd == LC_SYMTAB )
3918         {
3919                 // Find out the first and last undefined external
3920                 // symbol, so we don't have to allocate too many
3921                 // jump islands.
3922             struct symtab_command *symLC = (struct symtab_command *) lc;
3923             unsigned min = symLC->nsyms, max = 0;
3924             struct nlist *nlist =
3925                 symLC ? (struct nlist*) ((char*) oc->image + symLC->symoff)
3926                       : NULL;
3927             for(i=0;i<symLC->nsyms;i++)
3928             {
3929                 if(nlist[i].n_type & N_STAB)
3930                     ;
3931                 else if(nlist[i].n_type & N_EXT)
3932                 {
3933                     if((nlist[i].n_type & N_TYPE) == N_UNDF
3934                         && (nlist[i].n_value == 0))
3935                     {
3936                         if(i < min)
3937                             min = i;
3938                         if(i > max)
3939                             max = i;
3940                     }
3941                 }
3942             }
3943             if(max >= min)
3944                 return ocAllocateSymbolExtras(oc, max - min + 1, min);
3945
3946             break;
3947         }
3948
3949         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
3950     }
3951     return ocAllocateSymbolExtras(oc,0,0);
3952 }
3953 #endif
3954 #ifdef x86_64_HOST_ARCH
3955 static int ocAllocateSymbolExtras_MachO(ObjectCode* oc)
3956 {
3957     struct mach_header *header = (struct mach_header *) oc->image;
3958     struct load_command *lc = (struct load_command *) (header + 1);
3959     unsigned i;
3960
3961     for( i = 0; i < header->ncmds; i++ )
3962     {
3963         if( lc->cmd == LC_SYMTAB )
3964         {
3965                 // Just allocate one entry for every symbol
3966             struct symtab_command *symLC = (struct symtab_command *) lc;
3967
3968             return ocAllocateSymbolExtras(oc, symLC->nsyms, 0);
3969         }
3970
3971         lc = (struct load_command *) ( ((char *)lc) + lc->cmdsize );
3972     }
3973     return ocAllocateSymbolExtras(oc,0,0);
3974 }
3975 #endif
3976
3977 static int ocVerifyImage_MachO(ObjectCode* oc)
3978 {
3979     char *image = (char*) oc->image;
3980     struct mach_header *header = (struct mach_header*) image;
3981
3982 #if x86_64_TARGET_ARCH || powerpc64_TARGET_ARCH
3983     if(header->magic != MH_MAGIC_64)
3984         return 0;
3985 #else
3986     if(header->magic != MH_MAGIC)
3987         return 0;
3988 #endif
3989     // FIXME: do some more verifying here
3990     return 1;
3991 }
3992
3993 static int resolveImports(
3994     ObjectCode* oc,
3995     char *image,
3996     struct symtab_command *symLC,
3997     struct section *sect,    // ptr to lazy or non-lazy symbol pointer section
3998     unsigned long *indirectSyms,
3999     struct nlist *nlist)
4000 {
4001     unsigned i;
4002     size_t itemSize = 4;
4003
4004 #if i386_HOST_ARCH
4005     int isJumpTable = 0;
4006     if(!strcmp(sect->sectname,"__jump_table"))
4007     {
4008         isJumpTable = 1;
4009         itemSize = 5;
4010         ASSERT(sect->reserved2 == itemSize);
4011     }
4012 #endif
4013
4014     for(i=0; i*itemSize < sect->size;i++)
4015     {
4016         // according to otool, reserved1 contains the first index into the indirect symbol table
4017         struct nlist *symbol = &nlist[indirectSyms[sect->reserved1+i]];
4018         char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4019         void *addr = NULL;
4020
4021         if((symbol->n_type & N_TYPE) == N_UNDF
4022             && (symbol->n_type & N_EXT) && (symbol->n_value != 0))
4023             addr = (void*) (symbol->n_value);
4024         else
4025             addr = lookupSymbol(nm);
4026         if(!addr)
4027         {
4028             errorBelch("\n%s: unknown symbol `%s'", oc->fileName, nm);
4029             return 0;
4030         }
4031         ASSERT(addr);
4032
4033 #if i386_HOST_ARCH
4034         if(isJumpTable)
4035         {
4036             checkProddableBlock(oc,image + sect->offset + i*itemSize);
4037             *(image + sect->offset + i*itemSize) = 0xe9; // jmp
4038             *(unsigned*)(image + sect->offset + i*itemSize + 1)
4039                 = (char*)addr - (image + sect->offset + i*itemSize + 5);
4040         }
4041         else
4042 #endif
4043         {
4044             checkProddableBlock(oc,((void**)(image + sect->offset)) + i);
4045             ((void**)(image + sect->offset))[i] = addr;
4046         }
4047     }
4048
4049     return 1;
4050 }
4051
4052 static unsigned long relocateAddress(
4053     ObjectCode* oc,
4054     int nSections,
4055     struct section* sections,
4056     unsigned long address)
4057 {
4058     int i;
4059     for(i = 0; i < nSections; i++)
4060     {
4061         if(sections[i].addr <= address
4062             && address < sections[i].addr + sections[i].size)
4063         {
4064             return (unsigned long)oc->image
4065                     + sections[i].offset + address - sections[i].addr;
4066         }
4067     }
4068     barf("Invalid Mach-O file:"
4069          "Address out of bounds while relocating object file");
4070     return 0;
4071 }
4072
4073 static int relocateSection(
4074     ObjectCode* oc,
4075     char *image,
4076     struct symtab_command *symLC, struct nlist *nlist,
4077     int nSections, struct section* sections, struct section *sect)
4078 {
4079     struct relocation_info *relocs;
4080     int i,n;
4081
4082     if(!strcmp(sect->sectname,"__la_symbol_ptr"))
4083         return 1;
4084     else if(!strcmp(sect->sectname,"__nl_symbol_ptr"))
4085         return 1;
4086     else if(!strcmp(sect->sectname,"__la_sym_ptr2"))
4087         return 1;
4088     else if(!strcmp(sect->sectname,"__la_sym_ptr3"))
4089         return 1;
4090
4091     n = sect->nreloc;
4092     relocs = (struct relocation_info*) (image + sect->reloff);
4093
4094     for(i=0;i<n;i++)
4095     {
4096 #ifdef x86_64_HOST_ARCH
4097         struct relocation_info *reloc = &relocs[i];
4098
4099         char    *thingPtr = image + sect->offset + reloc->r_address;
4100         uint64_t thing;
4101         uint64_t value;
4102         uint64_t baseValue;
4103         int type = reloc->r_type;
4104
4105         checkProddableBlock(oc,thingPtr);
4106         switch(reloc->r_length)
4107         {
4108             case 0:
4109                 thing = *(uint8_t*)thingPtr;
4110                 baseValue = (uint64_t)thingPtr + 1;
4111                 break;
4112             case 1:
4113                 thing = *(uint16_t*)thingPtr;
4114                 baseValue = (uint64_t)thingPtr + 2;
4115                 break;
4116             case 2:
4117                 thing = *(uint32_t*)thingPtr;
4118                 baseValue = (uint64_t)thingPtr + 4;
4119                 break;
4120             case 3:
4121                 thing = *(uint64_t*)thingPtr;
4122                 baseValue = (uint64_t)thingPtr + 8;
4123                 break;
4124             default:
4125                 barf("Unknown size.");
4126         }
4127
4128         if(type == X86_64_RELOC_GOT
4129            || type == X86_64_RELOC_GOT_LOAD)
4130         {
4131             ASSERT(reloc->r_extern);
4132             value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)->addr;
4133
4134             type = X86_64_RELOC_SIGNED;
4135         }
4136         else if(reloc->r_extern)
4137         {
4138             struct nlist *symbol = &nlist[reloc->r_symbolnum];
4139             char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4140             if(symbol->n_value == 0)
4141                 value = (uint64_t) lookupSymbol(nm);
4142             else
4143                 value = relocateAddress(oc, nSections, sections,
4144                                         symbol->n_value);
4145         }
4146         else
4147         {
4148             value = sections[reloc->r_symbolnum-1].offset
4149                   - sections[reloc->r_symbolnum-1].addr
4150                   + (uint64_t) image;
4151         }
4152
4153         if(type == X86_64_RELOC_BRANCH)
4154         {
4155             if((int32_t)(value - baseValue) != (int64_t)(value - baseValue))
4156             {
4157                 ASSERT(reloc->r_extern);
4158                 value = (uint64_t) &makeSymbolExtra(oc, reloc->r_symbolnum, value)
4159                                         -> jumpIsland;
4160             }
4161             ASSERT((int32_t)(value - baseValue) == (int64_t)(value - baseValue));
4162             type = X86_64_RELOC_SIGNED;
4163         }
4164
4165         switch(type)
4166         {
4167             case X86_64_RELOC_UNSIGNED:
4168                 ASSERT(!reloc->r_pcrel);
4169                 thing += value;
4170                 break;
4171             case X86_64_RELOC_SIGNED:
4172                 ASSERT(reloc->r_pcrel);
4173                 thing += value - baseValue;
4174                 break;
4175             case X86_64_RELOC_SUBTRACTOR:
4176                 ASSERT(!reloc->r_pcrel);
4177                 thing -= value;
4178                 break;
4179             default:
4180                 barf("unkown relocation");
4181         }
4182
4183         switch(reloc->r_length)
4184         {
4185             case 0:
4186                 *(uint8_t*)thingPtr = thing;
4187                 break;
4188             case 1:
4189                 *(uint16_t*)thingPtr = thing;
4190                 break;
4191             case 2:
4192                 *(uint32_t*)thingPtr = thing;
4193                 break;
4194             case 3:
4195                 *(uint64_t*)thingPtr = thing;
4196                 break;
4197         }
4198 #else
4199         if(relocs[i].r_address & R_SCATTERED)
4200         {
4201             struct scattered_relocation_info *scat =
4202                 (struct scattered_relocation_info*) &relocs[i];
4203
4204             if(!scat->r_pcrel)
4205             {
4206                 if(scat->r_length == 2)
4207                 {
4208                     unsigned long word = 0;
4209                     unsigned long* wordPtr = (unsigned long*) (image + sect->offset + scat->r_address);
4210                     checkProddableBlock(oc,wordPtr);
4211
4212                     // Note on relocation types:
4213                     // i386 uses the GENERIC_RELOC_* types,
4214                     // while ppc uses special PPC_RELOC_* types.
4215                     // *_RELOC_VANILLA and *_RELOC_PAIR have the same value
4216                     // in both cases, all others are different.
4217                     // Therefore, we use GENERIC_RELOC_VANILLA
4218                     // and GENERIC_RELOC_PAIR instead of the PPC variants,
4219                     // and use #ifdefs for the other types.
4220
4221                     // Step 1: Figure out what the relocated value should be
4222                     if(scat->r_type == GENERIC_RELOC_VANILLA)
4223                     {
4224                         word = *wordPtr + (unsigned long) relocateAddress(
4225                                                                 oc,
4226                                                                 nSections,
4227                                                                 sections,
4228                                                                 scat->r_value)
4229                                         - scat->r_value;
4230                     }
4231 #ifdef powerpc_HOST_ARCH
4232                     else if(scat->r_type == PPC_RELOC_SECTDIFF
4233                         || scat->r_type == PPC_RELOC_LO16_SECTDIFF
4234                         || scat->r_type == PPC_RELOC_HI16_SECTDIFF
4235                         || scat->r_type == PPC_RELOC_HA16_SECTDIFF)
4236 #else
4237                     else if(scat->r_type == GENERIC_RELOC_SECTDIFF)
4238 #endif
4239                     {
4240                         struct scattered_relocation_info *pair =
4241                                 (struct scattered_relocation_info*) &relocs[i+1];
4242
4243                         if(!pair->r_scattered || pair->r_type != GENERIC_RELOC_PAIR)
4244                             barf("Invalid Mach-O file: "
4245                                  "RELOC_*_SECTDIFF not followed by RELOC_PAIR");
4246
4247                         word = (unsigned long)
4248                                (relocateAddress(oc, nSections, sections, scat->r_value)
4249                               - relocateAddress(oc, nSections, sections, pair->r_value));
4250                         i++;
4251                     }
4252 #ifdef powerpc_HOST_ARCH
4253                     else if(scat->r_type == PPC_RELOC_HI16
4254                          || scat->r_type == PPC_RELOC_LO16
4255                          || scat->r_type == PPC_RELOC_HA16
4256                          || scat->r_type == PPC_RELOC_LO14)
4257                     {   // these are generated by label+offset things
4258                         struct relocation_info *pair = &relocs[i+1];
4259                         if((pair->r_address & R_SCATTERED) || pair->r_type != PPC_RELOC_PAIR)
4260                             barf("Invalid Mach-O file: "
4261                                  "PPC_RELOC_* not followed by PPC_RELOC_PAIR");
4262
4263                         if(scat->r_type == PPC_RELOC_LO16)
4264                         {
4265                             word = ((unsigned short*) wordPtr)[1];
4266                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4267                         }
4268                         else if(scat->r_type == PPC_RELOC_LO14)
4269                         {
4270                             barf("Unsupported Relocation: PPC_RELOC_LO14");
4271                             word = ((unsigned short*) wordPtr)[1] & 0xFFFC;
4272                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4273                         }
4274                         else if(scat->r_type == PPC_RELOC_HI16)
4275                         {
4276                             word = ((unsigned short*) wordPtr)[1] << 16;
4277                             word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4278                         }
4279                         else if(scat->r_type == PPC_RELOC_HA16)
4280                         {
4281                             word = ((unsigned short*) wordPtr)[1] << 16;
4282                             word += ((short)relocs[i+1].r_address & (short)0xFFFF);
4283                         }
4284
4285
4286                         word += (unsigned long) relocateAddress(oc, nSections, sections, scat->r_value)
4287                                                 - scat->r_value;
4288
4289                         i++;
4290                     }
4291  #endif
4292                     else
4293                         continue;  // ignore the others
4294
4295 #ifdef powerpc_HOST_ARCH
4296                     if(scat->r_type == GENERIC_RELOC_VANILLA
4297                         || scat->r_type == PPC_RELOC_SECTDIFF)
4298 #else
4299                     if(scat->r_type == GENERIC_RELOC_VANILLA
4300                         || scat->r_type == GENERIC_RELOC_SECTDIFF)
4301 #endif
4302                     {
4303                         *wordPtr = word;
4304                     }
4305 #ifdef powerpc_HOST_ARCH
4306                     else if(scat->r_type == PPC_RELOC_LO16_SECTDIFF || scat->r_type == PPC_RELOC_LO16)
4307                     {
4308                         ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
4309                     }
4310                     else if(scat->r_type == PPC_RELOC_HI16_SECTDIFF || scat->r_type == PPC_RELOC_HI16)
4311                     {
4312                         ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
4313                     }
4314                     else if(scat->r_type == PPC_RELOC_HA16_SECTDIFF || scat->r_type == PPC_RELOC_HA16)
4315                     {
4316                         ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
4317                             + ((word & (1<<15)) ? 1 : 0);
4318                     }
4319 #endif
4320                 }
4321             }
4322
4323             continue; // FIXME: I hope it's OK to ignore all the others.
4324         }
4325         else
4326         {
4327             struct relocation_info *reloc = &relocs[i];
4328             if(reloc->r_pcrel && !reloc->r_extern)
4329                 continue;
4330
4331             if(reloc->r_length == 2)
4332             {
4333                 unsigned long word = 0;
4334 #ifdef powerpc_HOST_ARCH
4335                 unsigned long jumpIsland = 0;
4336                 long offsetToJumpIsland = 0xBADBAD42; // initialise to bad value
4337                                                       // to avoid warning and to catch
4338                                                       // bugs.
4339 #endif
4340
4341                 unsigned long* wordPtr = (unsigned long*) (image + sect->offset + reloc->r_address);
4342                 checkProddableBlock(oc,wordPtr);
4343
4344                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
4345                 {
4346                     word = *wordPtr;
4347                 }
4348 #ifdef powerpc_HOST_ARCH
4349                 else if(reloc->r_type == PPC_RELOC_LO16)
4350                 {
4351                     word = ((unsigned short*) wordPtr)[1];
4352                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF) << 16;
4353                 }
4354                 else if(reloc->r_type == PPC_RELOC_HI16)
4355                 {
4356                     word = ((unsigned short*) wordPtr)[1] << 16;
4357                     word |= ((unsigned long) relocs[i+1].r_address & 0xFFFF);
4358                 }
4359                 else if(reloc->r_type == PPC_RELOC_HA16)
4360                 {
4361                     word = ((unsigned short*) wordPtr)[1] << 16;
4362                     word += ((short)relocs[i+1].r_address & (short)0xFFFF);
4363                 }
4364                 else if(reloc->r_type == PPC_RELOC_BR24)
4365                 {
4366                     word = *wordPtr;
4367                     word = (word & 0x03FFFFFC) | ((word & 0x02000000) ? 0xFC000000 : 0);
4368                 }
4369 #endif
4370
4371                 if(!reloc->r_extern)
4372                 {
4373                     long delta =
4374                         sections[reloc->r_symbolnum-1].offset
4375                         - sections[reloc->r_symbolnum-1].addr
4376                         + ((long) image);
4377
4378                     word += delta;
4379                 }
4380                 else
4381                 {
4382                     struct nlist *symbol = &nlist[reloc->r_symbolnum];
4383                     char *nm = image + symLC->stroff + symbol->n_un.n_strx;
4384                     void *symbolAddress = lookupSymbol(nm);
4385                     if(!symbolAddress)
4386                     {
4387                         errorBelch("\nunknown symbol `%s'", nm);
4388                         return 0;
4389                     }
4390
4391                     if(reloc->r_pcrel)
4392                     {
4393 #ifdef powerpc_HOST_ARCH
4394                             // In the .o file, this should be a relative jump to NULL
4395                             // and we'll change it to a relative jump to the symbol
4396                         ASSERT(word + reloc->r_address == 0);
4397                         jumpIsland = (unsigned long)
4398                                         &makeSymbolExtra(oc,
4399                                                          reloc->r_symbolnum,
4400                                                          (unsigned long) symbolAddress)
4401                                          -> jumpIsland;
4402                         if(jumpIsland != 0)
4403                         {
4404                             offsetToJumpIsland = word + jumpIsland
4405                                 - (((long)image) + sect->offset - sect->addr);
4406                         }
4407 #endif
4408                         word += (unsigned long) symbolAddress
4409                                 - (((long)image) + sect->offset - sect->addr);
4410                     }
4411                     else
4412                     {
4413                         word += (unsigned long) symbolAddress;
4414                     }
4415                 }
4416
4417                 if(reloc->r_type == GENERIC_RELOC_VANILLA)
4418                 {
4419                     *wordPtr = word;
4420                     continue;
4421                 }
4422 #ifdef powerpc_HOST_ARCH
4423                 else if(reloc->r_type == PPC_RELOC_LO16)
4424                 {
4425                     ((unsigned short*) wordPtr)[1] = word & 0xFFFF;
4426                     i++; continue;
4427                 }
4428                 else if(reloc->r_type == PPC_RELOC_HI16)
4429                 {
4430                     ((unsigned short*) wordPtr)[1] = (word >> 16) & 0xFFFF;
4431                     i++; continue;
4432                 }
4433                 else if(reloc->r_type == PPC_RELOC_HA16)
4434                 {
4435                     ((unsigned short*) wordPtr)[1] = ((word >> 16) & 0xFFFF)
4436                         + ((word & (1<<15)) ? 1 : 0);
4437                     i++; continue;
4438                 }
4439                 else if(reloc->r_type == PPC_RELOC_BR24)
4440                 {
4441                     if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
4442                     {
4443                         // The branch offset is too large.
4444                         // Therefore, we try to use a jump island.
4445                         if(jumpIsland == 0)
4446                         {
4447                             barf("unconditional relative branch out of range: "
4448                                  "no jump island available");
4449                         }
4450
4451                         word = offsetToJumpIsland;
4452                         if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
4453                             barf("unconditional relative branch out of range: "
4454                                  "jump island out of range");
4455                     }
4456                     *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
4457                     continue;
4458                 }
4459 #endif
4460             }
4461             barf("\nunknown relocation %d",reloc->r_type);
4462             return 0;
4463         }
4464 #endif
4465     }
4466     return 1;
4467 }
4468
4469 static int ocGetNames_MachO(ObjectCode* oc)
4470 {
4471     char *image = (char*) oc->image;
4472     struct mach_header *header = (struct mach_header*) image;
4473     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
4474     unsigned i,curSymbol = 0;
4475     struct segment_command *segLC = NULL;
4476     struct section *sections;
4477     struct symtab_command *symLC = NULL;
4478     struct nlist *nlist;
4479     unsigned long commonSize = 0;
4480     char    *commonStorage = NULL;
4481     unsigned long commonCounter;
4482
4483     for(i=0;i<header->ncmds;i++)
4484     {
4485         if(lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64)
4486             segLC = (struct segment_command*) lc;
4487         else if(lc->cmd == LC_SYMTAB)
4488             symLC = (struct symtab_command*) lc;
4489         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
4490     }
4491
4492     sections = (struct section*) (segLC+1);
4493     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
4494                   : NULL;
4495
4496     if(!segLC)
4497         barf("ocGetNames_MachO: no segment load command");
4498
4499     for(i=0;i<segLC->nsects;i++)
4500     {
4501         if(sections[i].size == 0)
4502             continue;
4503
4504         if((sections[i].flags & SECTION_TYPE) == S_ZEROFILL)
4505         {
4506             char * zeroFillArea = stgCallocBytes(1,sections[i].size,
4507                                       "ocGetNames_MachO(common symbols)");
4508             sections[i].offset = zeroFillArea - image;
4509         }
4510
4511         if(!strcmp(sections[i].sectname,"__text"))
4512             addSection(oc, SECTIONKIND_CODE_OR_RODATA,
4513                 (void*) (image + sections[i].offset),
4514                 (void*) (image + sections[i].offset + sections[i].size));
4515         else if(!strcmp(sections[i].sectname,"__const"))
4516             addSection(oc, SECTIONKIND_RWDATA,
4517                 (void*) (image + sections[i].offset),
4518                 (void*) (image + sections[i].offset + sections[i].size));
4519         else if(!strcmp(sections[i].sectname,"__data"))
4520             addSection(oc, SECTIONKIND_RWDATA,
4521                 (void*) (image + sections[i].offset),
4522                 (void*) (image + sections[i].offset + sections[i].size));
4523         else if(!strcmp(sections[i].sectname,"__bss")
4524                 || !strcmp(sections[i].sectname,"__common"))
4525             addSection(oc, SECTIONKIND_RWDATA,
4526                 (void*) (image + sections[i].offset),
4527                 (void*) (image + sections[i].offset + sections[i].size));
4528
4529         addProddableBlock(oc, (void*) (image + sections[i].offset),
4530                                         sections[i].size);
4531     }
4532
4533         // count external symbols defined here
4534     oc->n_symbols = 0;
4535     if(symLC)
4536     {
4537         for(i=0;i<symLC->nsyms;i++)
4538         {
4539             if(nlist[i].n_type & N_STAB)
4540                 ;
4541             else if(nlist[i].n_type & N_EXT)
4542             {
4543                 if((nlist[i].n_type & N_TYPE) == N_UNDF
4544                     && (nlist[i].n_value != 0))
4545                 {
4546                     commonSize += nlist[i].n_value;
4547                     oc->n_symbols++;
4548                 }
4549                 else if((nlist[i].n_type & N_TYPE) == N_SECT)
4550                     oc->n_symbols++;
4551             }
4552         }
4553     }
4554     oc->symbols = stgMallocBytes(oc->n_symbols * sizeof(char*),
4555                                    "ocGetNames_MachO(oc->symbols)");
4556
4557     if(symLC)
4558     {
4559         for(i=0;i<symLC->nsyms;i++)
4560         {
4561             if(nlist[i].n_type & N_STAB)
4562                 ;
4563             else if((nlist[i].n_type & N_TYPE) == N_SECT)
4564             {
4565                 if(nlist[i].n_type & N_EXT)
4566                 {
4567                     char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
4568                     if((nlist[i].n_desc & N_WEAK_DEF) && lookupSymbol(nm))
4569                         ; // weak definition, and we already have a definition
4570                     else
4571                     {
4572                             ghciInsertStrHashTable(oc->fileName, symhash, nm,
4573                                                     image
4574                                                     + sections[nlist[i].n_sect-1].offset
4575                                                     - sections[nlist[i].n_sect-1].addr
4576                                                     + nlist[i].n_value);
4577                             oc->symbols[curSymbol++] = nm;
4578                     }
4579                 }
4580             }
4581         }
4582     }
4583
4584     commonStorage = stgCallocBytes(1,commonSize,"ocGetNames_MachO(common symbols)");
4585     commonCounter = (unsigned long)commonStorage;
4586     if(symLC)
4587     {
4588         for(i=0;i<symLC->nsyms;i++)
4589         {
4590             if((nlist[i].n_type & N_TYPE) == N_UNDF
4591                     && (nlist[i].n_type & N_EXT) && (nlist[i].n_value != 0))
4592             {
4593                 char *nm = image + symLC->stroff + nlist[i].n_un.n_strx;
4594                 unsigned long sz = nlist[i].n_value;
4595
4596                 nlist[i].n_value = commonCounter;
4597
4598                 ghciInsertStrHashTable(oc->fileName, symhash, nm,
4599                                        (void*)commonCounter);
4600                 oc->symbols[curSymbol++] = nm;
4601
4602                 commonCounter += sz;
4603             }
4604         }
4605     }
4606     return 1;
4607 }
4608
4609 static int ocResolve_MachO(ObjectCode* oc)
4610 {
4611     char *image = (char*) oc->image;
4612     struct mach_header *header = (struct mach_header*) image;
4613     struct load_command *lc = (struct load_command*) (image + sizeof(struct mach_header));
4614     unsigned i;
4615     struct segment_command *segLC = NULL;
4616     struct section *sections;
4617     struct symtab_command *symLC = NULL;
4618     struct dysymtab_command *dsymLC = NULL;
4619     struct nlist *nlist;
4620
4621     for(i=0;i<header->ncmds;i++)
4622     {
4623         if(lc->cmd == LC_SEGMENT || lc->cmd == LC_SEGMENT_64)
4624             segLC = (struct segment_command*) lc;
4625         else if(lc->cmd == LC_SYMTAB)
4626             symLC = (struct symtab_command*) lc;
4627         else if(lc->cmd == LC_DYSYMTAB)
4628             dsymLC = (struct dysymtab_command*) lc;
4629         lc = (struct load_command *) ( ((char*)lc) + lc->cmdsize );
4630     }
4631
4632     sections = (struct section*) (segLC+1);
4633     nlist = symLC ? (struct nlist*) (image + symLC->symoff)
4634                   : NULL;
4635
4636     if(dsymLC)
4637     {
4638         unsigned long *indirectSyms
4639             = (unsigned long*) (image + dsymLC->indirectsymoff);
4640
4641         for(i=0;i<segLC->nsects;i++)
4642         {
4643             if(    !strcmp(sections[i].sectname,"__la_symbol_ptr")
4644                 || !strcmp(sections[i].sectname,"__la_sym_ptr2")
4645                 || !strcmp(sections[i].sectname,"__la_sym_ptr3"))
4646             {
4647                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
4648                     return 0;
4649             }
4650             else if(!strcmp(sections[i].sectname,"__nl_symbol_ptr")
4651                 ||  !strcmp(sections[i].sectname,"__pointers"))
4652             {
4653                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
4654                     return 0;
4655             }
4656             else if(!strcmp(sections[i].sectname,"__jump_table"))
4657             {
4658                 if(!resolveImports(oc,image,symLC,&sections[i],indirectSyms,nlist))
4659                     return 0;
4660             }
4661         }
4662     }
4663
4664     for(i=0;i<segLC->nsects;i++)
4665     {
4666         if(!relocateSection(oc,image,symLC,nlist,segLC->nsects,sections,&sections[i]))
4667             return 0;
4668     }
4669
4670 #if defined (powerpc_HOST_ARCH)
4671     ocFlushInstructionCache( oc );
4672 #endif
4673
4674     return 1;
4675 }
4676
4677 #ifdef powerpc_HOST_ARCH
4678 /*
4679  * The Mach-O object format uses leading underscores. But not everywhere.
4680  * There is a small number of runtime support functions defined in
4681  * libcc_dynamic.a whose name does not have a leading underscore.
4682  * As a consequence, we can't get their address from C code.
4683  * We have to use inline assembler just to take the address of a function.
4684  * Yuck.
4685  */
4686
4687 static void machoInitSymbolsWithoutUnderscore()
4688 {
4689     extern void* symbolsWithoutUnderscore[];
4690     void **p = symbolsWithoutUnderscore;
4691     __asm__ volatile(".globl _symbolsWithoutUnderscore\n.data\n_symbolsWithoutUnderscore:");
4692
4693 #undef SymI_NeedsProto
4694 #define SymI_NeedsProto(x)  \
4695     __asm__ volatile(".long " # x);
4696
4697     RTS_MACHO_NOUNDERLINE_SYMBOLS
4698
4699     __asm__ volatile(".text");
4700
4701 #undef SymI_NeedsProto
4702 #define SymI_NeedsProto(x)  \
4703     ghciInsertStrHashTable("(GHCi built-in symbols)", symhash, #x, *p++);
4704
4705     RTS_MACHO_NOUNDERLINE_SYMBOLS
4706
4707 #undef SymI_NeedsProto
4708 }
4709 #endif
4710
4711 /*
4712  * Figure out by how much to shift the entire Mach-O file in memory
4713  * when loading so that its single segment ends up 16-byte-aligned
4714  */
4715 static int machoGetMisalignment( FILE * f )
4716 {
4717     struct mach_header header;
4718     int misalignment;
4719
4720     fread(&header, sizeof(header), 1, f);
4721     rewind(f);
4722
4723 #if x86_64_TARGET_ARCH || powerpc64_TARGET_ARCH
4724     if(header.magic != MH_MAGIC_64)
4725         return 0;
4726 #else
4727     if(header.magic != MH_MAGIC)
4728         return 0;
4729 #endif
4730
4731     misalignment = (header.sizeofcmds + sizeof(header))
4732                     & 0xF;
4733
4734     return misalignment ? (16 - misalignment) : 0;
4735 }
4736
4737 #endif
4738