/* Assorted BFD support routines, only used internally. Copyright 1990, 91, 92, 93, 94, 95, 96, 97, 98, 1999 Free Software Foundation, Inc. Written by Cygnus Support. This file is part of BFD, the Binary File Descriptor library. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ #include "bfd.h" #include "sysdep.h" #include "libbfd.h" #ifndef HAVE_GETPAGESIZE #define getpagesize() 2048 #endif static int real_read PARAMS ((PTR, size_t, size_t, FILE *)); /* SECTION Internal functions DESCRIPTION These routines are used within BFD. They are not intended for export, but are documented here for completeness. */ /* A routine which is used in target vectors for unsupported operations. */ /*ARGSUSED*/ boolean bfd_false (ignore) bfd *ignore ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_invalid_operation); return false; } /* A routine which is used in target vectors for supported operations which do not actually do anything. */ /*ARGSUSED*/ boolean bfd_true (ignore) bfd *ignore ATTRIBUTE_UNUSED; { return true; } /* A routine which is used in target vectors for unsupported operations which return a pointer value. */ /*ARGSUSED*/ PTR bfd_nullvoidptr (ignore) bfd *ignore ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_invalid_operation); return NULL; } /*ARGSUSED*/ int bfd_0 (ignore) bfd *ignore ATTRIBUTE_UNUSED; { return 0; } /*ARGSUSED*/ unsigned int bfd_0u (ignore) bfd *ignore ATTRIBUTE_UNUSED; { return 0; } /*ARGUSED*/ long bfd_0l (ignore) bfd *ignore ATTRIBUTE_UNUSED; { return 0; } /* A routine which is used in target vectors for unsupported operations which return -1 on error. */ /*ARGSUSED*/ long _bfd_n1 (ignore_abfd) bfd *ignore_abfd ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_invalid_operation); return -1; } /*ARGSUSED*/ void bfd_void (ignore) bfd *ignore ATTRIBUTE_UNUSED; { } /*ARGSUSED*/ boolean _bfd_nocore_core_file_matches_executable_p (ignore_core_bfd, ignore_exec_bfd) bfd *ignore_core_bfd ATTRIBUTE_UNUSED; bfd *ignore_exec_bfd ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_invalid_operation); return false; } /* Routine to handle core_file_failing_command entry point for targets without core file support. */ /*ARGSUSED*/ char * _bfd_nocore_core_file_failing_command (ignore_abfd) bfd *ignore_abfd ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_invalid_operation); return (char *)NULL; } /* Routine to handle core_file_failing_signal entry point for targets without core file support. */ /*ARGSUSED*/ int _bfd_nocore_core_file_failing_signal (ignore_abfd) bfd *ignore_abfd ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_invalid_operation); return 0; } /*ARGSUSED*/ const bfd_target * _bfd_dummy_target (ignore_abfd) bfd *ignore_abfd ATTRIBUTE_UNUSED; { bfd_set_error (bfd_error_wrong_format); return 0; } /* Allocate memory using malloc. */ PTR bfd_malloc (size) size_t size; { PTR ptr; ptr = (PTR) malloc (size); if (ptr == NULL && size != 0) bfd_set_error (bfd_error_no_memory); return ptr; } /* Reallocate memory using realloc. */ PTR bfd_realloc (ptr, size) PTR ptr; size_t size; { PTR ret; if (ptr == NULL) ret = malloc (size); else ret = realloc (ptr, size); if (ret == NULL) bfd_set_error (bfd_error_no_memory); return ret; } /* Allocate memory using malloc and clear it. */ PTR bfd_zmalloc (size) size_t size; { PTR ptr; ptr = (PTR) malloc (size); if (size != 0) { if (ptr == NULL) bfd_set_error (bfd_error_no_memory); else memset (ptr, 0, size); } return ptr; } /* Some IO code */ /* Note that archive entries don't have streams; they share their parent's. This allows someone to play with the iostream behind BFD's back. Also, note that the origin pointer points to the beginning of a file's contents (0 for non-archive elements). For archive entries this is the first octet in the file, NOT the beginning of the archive header. */ static int real_read (where, a,b, file) PTR where; size_t a; size_t b; FILE *file; { /* FIXME - this looks like an optimization, but it's really to cover up for a feature of some OSs (not solaris - sigh) that ld/pe-dll.c takes advantage of (apparently) when it creates BFDs internally and tries to link against them. BFD seems to be smart enough to realize there are no symbol records in the "file" that doesn't exist but attempts to read them anyway. On Solaris, attempting to read zero bytes from a NULL file results in a core dump, but on other platforms it just returns zero bytes read. This makes it to something reasonable. - DJ */ if (a == 0 || b == 0) return 0; #if defined (__VAX) && defined (VMS) /* Apparently fread on Vax VMS does not keep the record length information. */ return read (fileno (file), where, a * b); #else return fread (where, a, b, file); #endif } /* Return value is amount read (FIXME: how are errors and end of file dealt with? We never call bfd_set_error, which is probably a mistake). */ bfd_size_type bfd_read (ptr, size, nitems, abfd) PTR ptr; bfd_size_type size; bfd_size_type nitems; bfd *abfd; { int nread; if ((abfd->flags & BFD_IN_MEMORY) != 0) { struct bfd_in_memory *bim; bfd_size_type get; bim = (struct bfd_in_memory *) abfd->iostream; get = size * nitems; if (abfd->where + get > bim->size) { if (bim->size < abfd->where) get = 0; else get = bim->size - abfd->where; bfd_set_error (bfd_error_file_truncated); } memcpy (ptr, bim->buffer + abfd->where, get); abfd->where += get; return get; } nread = real_read (ptr, 1, (size_t)(size*nitems), bfd_cache_lookup(abfd)); if (nread > 0) abfd->where += nread; /* Set bfd_error if we did not read as much data as we expected. If the read failed due to an error set the bfd_error_system_call, else set bfd_error_file_truncated. A BFD backend may wish to override bfd_error_file_truncated to provide something more useful (eg. no_symbols or wrong_format). */ if (nread != (int) (size * nitems)) { if (ferror (bfd_cache_lookup (abfd))) bfd_set_error (bfd_error_system_call); else bfd_set_error (bfd_error_file_truncated); } return nread; } /* The window support stuff should probably be broken out into another file.... */ /* The idea behind the next and refcount fields is that one mapped region can suffice for multiple read-only windows or multiple non-overlapping read-write windows. It's not implemented yet though. */ struct _bfd_window_internal { struct _bfd_window_internal *next; PTR data; bfd_size_type size; int refcount : 31; /* should be enough... */ unsigned mapped : 1; /* 1 = mmap, 0 = malloc */ }; void bfd_init_window (windowp) bfd_window *windowp; { windowp->data = 0; windowp->i = 0; windowp->size = 0; } /* Currently, if USE_MMAP is undefined, none if the window stuff is used. Okay, so it's mis-named. At least the command-line option "--without-mmap" is more obvious than "--without-windows" or some such. */ #ifdef USE_MMAP #undef HAVE_MPROTECT /* code's not tested yet */ #if HAVE_MMAP || HAVE_MPROTECT || HAVE_MADVISE #include #endif #ifndef MAP_FILE #define MAP_FILE 0 #endif static int debug_windows; void bfd_free_window (windowp) bfd_window *windowp; { bfd_window_internal *i = windowp->i; windowp->i = 0; windowp->data = 0; if (i == 0) return; i->refcount--; if (debug_windows) fprintf (stderr, "freeing window @%p<%p,%lx,%p>\n", windowp, windowp->data, windowp->size, windowp->i); if (i->refcount != 0) return; if (i->mapped) { #ifdef HAVE_MMAP munmap (i->data, i->size); goto no_free; #else abort (); #endif } #ifdef HAVE_MPROTECT mprotect (i->data, i->size, PROT_READ | PROT_WRITE); #endif free (i->data); #ifdef HAVE_MMAP no_free: #endif i->data = 0; /* There should be no more references to i at this point. */ free (i); } static int ok_to_map = 1; boolean bfd_get_file_window (abfd, offset, size, windowp, writable) bfd *abfd; file_ptr offset; bfd_size_type size; bfd_window *windowp; boolean writable; { static size_t pagesize; bfd_window_internal *i = windowp->i; size_t size_to_alloc = size; if (debug_windows) fprintf (stderr, "bfd_get_file_window (%p, %6ld, %6ld, %p<%p,%lx,%p>, %d)", abfd, (long) offset, (long) size, windowp, windowp->data, (unsigned long) windowp->size, windowp->i, writable); /* Make sure we know the page size, so we can be friendly to mmap. */ if (pagesize == 0) pagesize = getpagesize (); if (pagesize == 0) abort (); if (i == 0) { windowp->i = i = (bfd_window_internal *) bfd_zmalloc (sizeof (bfd_window_internal)); if (i == 0) return false; i->data = 0; } #ifdef HAVE_MMAP if (ok_to_map && (i->data == 0 || i->mapped == 1) && (abfd->flags & BFD_IN_MEMORY) == 0) { file_ptr file_offset, offset2; size_t real_size; int fd; FILE *f; /* Find the real file and the real offset into it. */ while (abfd->my_archive != NULL) { offset += abfd->origin; abfd = abfd->my_archive; } f = bfd_cache_lookup (abfd); fd = fileno (f); /* Compute offsets and size for mmap and for the user's data. */ offset2 = offset % pagesize; if (offset2 < 0) abort (); file_offset = offset - offset2; real_size = offset + size - file_offset; real_size = real_size + pagesize - 1; real_size -= real_size % pagesize; /* If we're re-using a memory region, make sure it's big enough. */ if (i->data && i->size < size) { munmap (i->data, i->size); i->data = 0; } i->data = mmap (i->data, real_size, writable ? PROT_WRITE | PROT_READ : PROT_READ, (writable ? MAP_FILE | MAP_PRIVATE : MAP_FILE | MAP_SHARED), fd, file_offset); if (i->data == (PTR) -1) { /* An error happened. Report it, or try using malloc, or something. */ bfd_set_error (bfd_error_system_call); i->data = 0; windowp->data = 0; if (debug_windows) fprintf (stderr, "\t\tmmap failed!\n"); return false; } if (debug_windows) fprintf (stderr, "\n\tmapped %ld at %p, offset is %ld\n", (long) real_size, i->data, (long) offset2); i->size = real_size; windowp->data = (PTR) ((bfd_byte *) i->data + offset2); windowp->size = size; i->mapped = 1; return true; } else if (debug_windows) { if (ok_to_map) fprintf (stderr, _("not mapping: data=%lx mapped=%d\n"), (unsigned long) i->data, (int) i->mapped); else fprintf (stderr, _("not mapping: env var not set\n")); } #else ok_to_map = 0; #endif #ifdef HAVE_MPROTECT if (!writable) { size_to_alloc += pagesize - 1; size_to_alloc -= size_to_alloc % pagesize; } #endif if (debug_windows) fprintf (stderr, "\n\t%s(%6ld)", i->data ? "realloc" : " malloc", (long) size_to_alloc); i->data = (PTR) bfd_realloc (i->data, size_to_alloc); if (debug_windows) fprintf (stderr, "\t-> %p\n", i->data); i->refcount = 1; if (i->data == NULL) { if (size_to_alloc == 0) return true; bfd_set_error (bfd_error_no_memory); return false; } if (bfd_seek (abfd, offset, SEEK_SET) != 0) return false; i->size = bfd_read (i->data, size, 1, abfd); if (i->size != size) return false; i->mapped = 0; #ifdef HAVE_MPROTECT if (!writable) { if (debug_windows) fprintf (stderr, "\tmprotect (%p, %ld, PROT_READ)\n", i->data, (long) i->size); mprotect (i->data, i->size, PROT_READ); } #endif windowp->data = i->data; windowp->size = i->size; return true; } #endif /* USE_MMAP */ bfd_size_type bfd_write (ptr, size, nitems, abfd) CONST PTR ptr; bfd_size_type size; bfd_size_type nitems; bfd *abfd; { long nwrote; if ((abfd->flags & BFD_IN_MEMORY) != 0) { struct bfd_in_memory *bim = (struct bfd_in_memory *) (abfd->iostream); size *= nitems; if (abfd->where + size > bim->size) { long newsize, oldsize = (bim->size + 127) & ~127; bim->size = abfd->where + size; /* Round up to cut down on memory fragmentation */ newsize = (bim->size + 127) & ~127; if (newsize > oldsize) { bim->buffer = bfd_realloc (bim->buffer, newsize); if (bim->buffer == 0) { bim->size = 0; return 0; } } } memcpy (bim->buffer + abfd->where, ptr, size); abfd->where += size; return size; } nwrote = fwrite (ptr, 1, (size_t) (size * nitems), bfd_cache_lookup (abfd)); if (nwrote > 0) abfd->where += nwrote; if ((bfd_size_type) nwrote != size * nitems) { #ifdef ENOSPC if (nwrote >= 0) errno = ENOSPC; #endif bfd_set_error (bfd_error_system_call); } return nwrote; } /* INTERNAL_FUNCTION bfd_write_bigendian_4byte_int SYNOPSIS void bfd_write_bigendian_4byte_int(bfd *abfd, int i); DESCRIPTION Write a 4 byte integer @var{i} to the output BFD @var{abfd}, in big endian order regardless of what else is going on. This is useful in archives. */ void bfd_write_bigendian_4byte_int (abfd, i) bfd *abfd; int i; { bfd_byte buffer[4]; bfd_putb32(i, buffer); if (bfd_write((PTR)buffer, 4, 1, abfd) != 4) abort (); } long bfd_tell (abfd) bfd *abfd; { file_ptr ptr; if ((abfd->flags & BFD_IN_MEMORY) != 0) return abfd->where; ptr = ftell (bfd_cache_lookup(abfd)); if (abfd->my_archive) ptr -= abfd->origin; abfd->where = ptr; return ptr; } int bfd_flush (abfd) bfd *abfd; { if ((abfd->flags & BFD_IN_MEMORY) != 0) return 0; return fflush (bfd_cache_lookup(abfd)); } /* Returns 0 for success, negative value for failure (in which case bfd_get_error can retrieve the error code). */ int bfd_stat (abfd, statbuf) bfd *abfd; struct stat *statbuf; { FILE *f; int result; if ((abfd->flags & BFD_IN_MEMORY) != 0) abort (); f = bfd_cache_lookup (abfd); if (f == NULL) { bfd_set_error (bfd_error_system_call); return -1; } result = fstat (fileno (f), statbuf); if (result < 0) bfd_set_error (bfd_error_system_call); return result; } /* Returns 0 for success, nonzero for failure (in which case bfd_get_error can retrieve the error code). */ int bfd_seek (abfd, position, direction) bfd *abfd; file_ptr position; int direction; { int result; FILE *f; file_ptr file_position; /* For the time being, a BFD may not seek to it's end. The problem is that we don't easily have a way to recognize the end of an element in an archive. */ BFD_ASSERT (direction == SEEK_SET || direction == SEEK_CUR); if (direction == SEEK_CUR && position == 0) return 0; if ((abfd->flags & BFD_IN_MEMORY) != 0) { struct bfd_in_memory *bim; bim = (struct bfd_in_memory *) abfd->iostream; if (direction == SEEK_SET) abfd->where = position; else abfd->where += position; if (abfd->where > bim->size) { abfd->where = bim->size; bfd_set_error (bfd_error_file_truncated); return -1; } return 0; } if (abfd->format != bfd_archive && abfd->my_archive == 0) { #if 0 /* Explanation for this code: I'm only about 95+% sure that the above conditions are sufficient and that all i/o calls are properly adjusting the `where' field. So this is sort of an `assert' that the `where' field is correct. If we can go a while without tripping the abort, we can probably safely disable this code, so that the real optimizations happen. */ file_ptr where_am_i_now; where_am_i_now = ftell (bfd_cache_lookup (abfd)); if (abfd->my_archive) where_am_i_now -= abfd->origin; if (where_am_i_now != abfd->where) abort (); #endif if (direction == SEEK_SET && position == abfd->where) return 0; } else { /* We need something smarter to optimize access to archives. Currently, anything inside an archive is read via the file handle for the archive. Which means that a bfd_seek on one component affects the `current position' in the archive, as well as in any other component. It might be sufficient to put a spike through the cache abstraction, and look to the archive for the file position, but I think we should try for something cleaner. In the meantime, no optimization for archives. */ } f = bfd_cache_lookup (abfd); file_position = position; if (direction == SEEK_SET && abfd->my_archive != NULL) file_position += abfd->origin; result = fseek (f, file_position, direction); if (result != 0) { int hold_errno = errno; /* Force redetermination of `where' field. */ bfd_tell (abfd); /* An EINVAL error probably means that the file offset was absurd. */ if (hold_errno == EINVAL) bfd_set_error (bfd_error_file_truncated); else { bfd_set_error (bfd_error_system_call); errno = hold_errno; } } else { /* Adjust `where' field. */ if (direction == SEEK_SET) abfd->where = position; else abfd->where += position; } return result; } /** The do-it-yourself (byte) sex-change kit */ /* The middle letter e.g. getshort indicates Big or Little endian target machine. It doesn't matter what the byte order of the host machine is; these routines work for either. */ /* FIXME: Should these take a count argument? Answer (gnu@cygnus.com): No, but perhaps they should be inline functions in swap.h #ifdef __GNUC__. Gprof them later and find out. */ /* FUNCTION bfd_put_size FUNCTION bfd_get_size DESCRIPTION These macros as used for reading and writing raw data in sections; each access (except for bytes) is vectored through the target format of the BFD and mangled accordingly. The mangling performs any necessary endian translations and removes alignment restrictions. Note that types accepted and returned by these macros are identical so they can be swapped around in macros---for example, @file{libaout.h} defines <> to either <> or <>. In the put routines, @var{val} must be a <>. If we are on a system without prototypes, the caller is responsible for making sure that is true, with a cast if necessary. We don't cast them in the macro definitions because that would prevent <> or <> from detecting sins such as passing a pointer. To detect calling these with less than a <>, use <> on a host with 64 bit <>'s. . .{* Byte swapping macros for user section data. *} . .#define bfd_put_8(abfd, val, ptr) \ . ((void) (*((unsigned char *)(ptr)) = (unsigned char)(val))) .#define bfd_put_signed_8 \ . bfd_put_8 .#define bfd_get_8(abfd, ptr) \ . (*(unsigned char *)(ptr)) .#define bfd_get_signed_8(abfd, ptr) \ . ((*(unsigned char *)(ptr) ^ 0x80) - 0x80) . .#define bfd_put_16(abfd, val, ptr) \ . BFD_SEND(abfd, bfd_putx16, ((val),(ptr))) .#define bfd_put_signed_16 \ . bfd_put_16 .#define bfd_get_16(abfd, ptr) \ . BFD_SEND(abfd, bfd_getx16, (ptr)) .#define bfd_get_signed_16(abfd, ptr) \ . BFD_SEND (abfd, bfd_getx_signed_16, (ptr)) . .#define bfd_put_32(abfd, val, ptr) \ . BFD_SEND(abfd, bfd_putx32, ((val),(ptr))) .#define bfd_put_signed_32 \ . bfd_put_32 .#define bfd_get_32(abfd, ptr) \ . BFD_SEND(abfd, bfd_getx32, (ptr)) .#define bfd_get_signed_32(abfd, ptr) \ . BFD_SEND(abfd, bfd_getx_signed_32, (ptr)) . .#define bfd_put_64(abfd, val, ptr) \ . BFD_SEND(abfd, bfd_putx64, ((val), (ptr))) .#define bfd_put_signed_64 \ . bfd_put_64 .#define bfd_get_64(abfd, ptr) \ . BFD_SEND(abfd, bfd_getx64, (ptr)) .#define bfd_get_signed_64(abfd, ptr) \ . BFD_SEND(abfd, bfd_getx_signed_64, (ptr)) . .#define bfd_get(bits, abfd, ptr) \ . ((bits) == 8 ? bfd_get_8 (abfd, ptr) \ . : (bits) == 16 ? bfd_get_16 (abfd, ptr) \ . : (bits) == 32 ? bfd_get_32 (abfd, ptr) \ . : (bits) == 64 ? bfd_get_64 (abfd, ptr) \ . : (abort (), (bfd_vma) - 1)) . .#define bfd_put(bits, abfd, val, ptr) \ . ((bits) == 8 ? bfd_put_8 (abfd, val, ptr) \ . : (bits) == 16 ? bfd_put_16 (abfd, val, ptr) \ . : (bits) == 32 ? bfd_put_32 (abfd, val, ptr) \ . : (bits) == 64 ? bfd_put_64 (abfd, val, ptr) \ . : (abort (), (void) 0)) . */ /* FUNCTION bfd_h_put_size bfd_h_get_size DESCRIPTION These macros have the same function as their <> bretheren, except that they are used for removing information for the header records of object files. Believe it or not, some object files keep their header records in big endian order and their data in little endian order. . .{* Byte swapping macros for file header data. *} . .#define bfd_h_put_8(abfd, val, ptr) \ . bfd_put_8 (abfd, val, ptr) .#define bfd_h_put_signed_8(abfd, val, ptr) \ . bfd_put_8 (abfd, val, ptr) .#define bfd_h_get_8(abfd, ptr) \ . bfd_get_8 (abfd, ptr) .#define bfd_h_get_signed_8(abfd, ptr) \ . bfd_get_signed_8 (abfd, ptr) . .#define bfd_h_put_16(abfd, val, ptr) \ . BFD_SEND(abfd, bfd_h_putx16,(val,ptr)) .#define bfd_h_put_signed_16 \ . bfd_h_put_16 .#define bfd_h_get_16(abfd, ptr) \ . BFD_SEND(abfd, bfd_h_getx16,(ptr)) .#define bfd_h_get_signed_16(abfd, ptr) \ . BFD_SEND(abfd, bfd_h_getx_signed_16, (ptr)) . .#define bfd_h_put_32(abfd, val, ptr) \ . BFD_SEND(abfd, bfd_h_putx32,(val,ptr)) .#define bfd_h_put_signed_32 \ . bfd_h_put_32 .#define bfd_h_get_32(abfd, ptr) \ . BFD_SEND(abfd, bfd_h_getx32,(ptr)) .#define bfd_h_get_signed_32(abfd, ptr) \ . BFD_SEND(abfd, bfd_h_getx_signed_32, (ptr)) . .#define bfd_h_put_64(abfd, val, ptr) \ . BFD_SEND(abfd, bfd_h_putx64,(val, ptr)) .#define bfd_h_put_signed_64 \ . bfd_h_put_64 .#define bfd_h_get_64(abfd, ptr) \ . BFD_SEND(abfd, bfd_h_getx64,(ptr)) .#define bfd_h_get_signed_64(abfd, ptr) \ . BFD_SEND(abfd, bfd_h_getx_signed_64, (ptr)) . */ /* Sign extension to bfd_signed_vma. */ #define COERCE16(x) (((bfd_signed_vma) (x) ^ 0x8000) - 0x8000) #define COERCE32(x) \ ((bfd_signed_vma) (long) (((unsigned long) (x) ^ 0x80000000) - 0x80000000)) #define EIGHT_GAZILLION (((BFD_HOST_64_BIT)0x80000000) << 32) #define COERCE64(x) \ (((bfd_signed_vma) (x) ^ EIGHT_GAZILLION) - EIGHT_GAZILLION) bfd_vma bfd_getb16 (addr) register const bfd_byte *addr; { return (addr[0] << 8) | addr[1]; } bfd_vma bfd_getl16 (addr) register const bfd_byte *addr; { return (addr[1] << 8) | addr[0]; } bfd_signed_vma bfd_getb_signed_16 (addr) register const bfd_byte *addr; { return COERCE16((addr[0] << 8) | addr[1]); } bfd_signed_vma bfd_getl_signed_16 (addr) register const bfd_byte *addr; { return COERCE16((addr[1] << 8) | addr[0]); } void bfd_putb16 (data, addr) bfd_vma data; register bfd_byte *addr; { addr[0] = (bfd_byte)(data >> 8); addr[1] = (bfd_byte )data; } void bfd_putl16 (data, addr) bfd_vma data; register bfd_byte *addr; { addr[0] = (bfd_byte )data; addr[1] = (bfd_byte)(data >> 8); } bfd_vma bfd_getb32 (addr) register const bfd_byte *addr; { unsigned long v; v = (unsigned long) addr[0] << 24; v |= (unsigned long) addr[1] << 16; v |= (unsigned long) addr[2] << 8; v |= (unsigned long) addr[3]; return (bfd_vma) v; } bfd_vma bfd_getl32 (addr) register const bfd_byte *addr; { unsigned long v; v = (unsigned long) addr[0]; v |= (unsigned long) addr[1] << 8; v |= (unsigned long) addr[2] << 16; v |= (unsigned long) addr[3] << 24; return (bfd_vma) v; } bfd_signed_vma bfd_getb_signed_32 (addr) register const bfd_byte *addr; { unsigned long v; v = (unsigned long) addr[0] << 24; v |= (unsigned long) addr[1] << 16; v |= (unsigned long) addr[2] << 8; v |= (unsigned long) addr[3]; return COERCE32 (v); } bfd_signed_vma bfd_getl_signed_32 (addr) register const bfd_byte *addr; { unsigned long v; v = (unsigned long) addr[0]; v |= (unsigned long) addr[1] << 8; v |= (unsigned long) addr[2] << 16; v |= (unsigned long) addr[3] << 24; return COERCE32 (v); } bfd_vma bfd_getb64 (addr) register const bfd_byte *addr ATTRIBUTE_UNUSED; { #ifdef BFD64 bfd_vma low, high; high= ((((((((addr[0]) << 8) | addr[1]) << 8) | addr[2]) << 8) | addr[3]) ); low = (((((((((bfd_vma)addr[4]) << 8) | addr[5]) << 8) | addr[6]) << 8) | addr[7])); return high << 32 | low; #else BFD_FAIL(); return 0; #endif } bfd_vma bfd_getl64 (addr) register const bfd_byte *addr ATTRIBUTE_UNUSED; { #ifdef BFD64 bfd_vma low, high; high= (((((((addr[7] << 8) | addr[6]) << 8) | addr[5]) << 8) | addr[4])); low = ((((((((bfd_vma)addr[3] << 8) | addr[2]) << 8) | addr[1]) << 8) | addr[0]) ); return high << 32 | low; #else BFD_FAIL(); return 0; #endif } bfd_signed_vma bfd_getb_signed_64 (addr) register const bfd_byte *addr ATTRIBUTE_UNUSED; { #ifdef BFD64 bfd_vma low, high; high= ((((((((addr[0]) << 8) | addr[1]) << 8) | addr[2]) << 8) | addr[3]) ); low = (((((((((bfd_vma)addr[4]) << 8) | addr[5]) << 8) | addr[6]) << 8) | addr[7])); return COERCE64(high << 32 | low); #else BFD_FAIL(); return 0; #endif } bfd_signed_vma bfd_getl_signed_64 (addr) register const bfd_byte *addr ATTRIBUTE_UNUSED; { #ifdef BFD64 bfd_vma low, high; high= (((((((addr[7] << 8) | addr[6]) << 8) | addr[5]) << 8) | addr[4])); low = ((((((((bfd_vma)addr[3] << 8) | addr[2]) << 8) | addr[1]) << 8) | addr[0]) ); return COERCE64(high << 32 | low); #else BFD_FAIL(); return 0; #endif } void bfd_putb32 (data, addr) bfd_vma data; register bfd_byte *addr; { addr[0] = (bfd_byte)(data >> 24); addr[1] = (bfd_byte)(data >> 16); addr[2] = (bfd_byte)(data >> 8); addr[3] = (bfd_byte)data; } void bfd_putl32 (data, addr) bfd_vma data; register bfd_byte *addr; { addr[0] = (bfd_byte)data; addr[1] = (bfd_byte)(data >> 8); addr[2] = (bfd_byte)(data >> 16); addr[3] = (bfd_byte)(data >> 24); } void bfd_putb64 (data, addr) bfd_vma data ATTRIBUTE_UNUSED; register bfd_byte *addr ATTRIBUTE_UNUSED; { #ifdef BFD64 addr[0] = (bfd_byte)(data >> (7*8)); addr[1] = (bfd_byte)(data >> (6*8)); addr[2] = (bfd_byte)(data >> (5*8)); addr[3] = (bfd_byte)(data >> (4*8)); addr[4] = (bfd_byte)(data >> (3*8)); addr[5] = (bfd_byte)(data >> (2*8)); addr[6] = (bfd_byte)(data >> (1*8)); addr[7] = (bfd_byte)(data >> (0*8)); #else BFD_FAIL(); #endif } void bfd_putl64 (data, addr) bfd_vma data ATTRIBUTE_UNUSED; register bfd_byte *addr ATTRIBUTE_UNUSED; { #ifdef BFD64 addr[7] = (bfd_byte)(data >> (7*8)); addr[6] = (bfd_byte)(data >> (6*8)); addr[5] = (bfd_byte)(data >> (5*8)); addr[4] = (bfd_byte)(data >> (4*8)); addr[3] = (bfd_byte)(data >> (3*8)); addr[2] = (bfd_byte)(data >> (2*8)); addr[1] = (bfd_byte)(data >> (1*8)); addr[0] = (bfd_byte)(data >> (0*8)); #else BFD_FAIL(); #endif } /* Default implementation */ boolean _bfd_generic_get_section_contents (abfd, section, location, offset, count) bfd *abfd; sec_ptr section; PTR location; file_ptr offset; bfd_size_type count; { if (count == 0) return true; if ((bfd_size_type) (offset + count) > section->_raw_size) { bfd_set_error (bfd_error_invalid_operation); return false; } if (bfd_seek (abfd, section->filepos + offset, SEEK_SET) != 0 || bfd_read (location, (bfd_size_type) 1, count, abfd) != count) return false; return true; } boolean _bfd_generic_get_section_contents_in_window (abfd, section, w, offset, count) bfd *abfd ATTRIBUTE_UNUSED; sec_ptr section ATTRIBUTE_UNUSED; bfd_window *w ATTRIBUTE_UNUSED; file_ptr offset ATTRIBUTE_UNUSED; bfd_size_type count ATTRIBUTE_UNUSED; { #ifdef USE_MMAP if (count == 0) return true; if (abfd->xvec->_bfd_get_section_contents != _bfd_generic_get_section_contents) { /* We don't know what changes the bfd's get_section_contents method may have to make. So punt trying to map the file window, and let get_section_contents do its thing. */ /* @@ FIXME : If the internal window has a refcount of 1 and was allocated with malloc instead of mmap, just reuse it. */ bfd_free_window (w); w->i = (bfd_window_internal *) bfd_zmalloc (sizeof (bfd_window_internal)); if (w->i == NULL) return false; w->i->data = (PTR) bfd_malloc ((size_t) count); if (w->i->data == NULL) { free (w->i); w->i = NULL; return false; } w->i->mapped = 0; w->i->refcount = 1; w->size = w->i->size = count; w->data = w->i->data; return bfd_get_section_contents (abfd, section, w->data, offset, count); } if ((bfd_size_type) (offset+count) > section->_raw_size || (bfd_get_file_window (abfd, section->filepos + offset, count, w, true) == false)) return false; return true; #else abort (); #endif } /* This generic function can only be used in implementations where creating NEW sections is disallowed. It is useful in patching existing sections in read-write files, though. See other set_section_contents functions to see why it doesn't work for new sections. */ boolean _bfd_generic_set_section_contents (abfd, section, location, offset, count) bfd *abfd; sec_ptr section; PTR location; file_ptr offset; bfd_size_type count; { if (count == 0) return true; if (bfd_seek (abfd, (file_ptr) (section->filepos + offset), SEEK_SET) == -1 || bfd_write (location, (bfd_size_type) 1, count, abfd) != count) return false; return true; } /* INTERNAL_FUNCTION bfd_log2 SYNOPSIS unsigned int bfd_log2(bfd_vma x); DESCRIPTION Return the log base 2 of the value supplied, rounded up. E.g., an @var{x} of 1025 returns 11. */ unsigned int bfd_log2 (x) bfd_vma x; { unsigned int result = 0; while ((x = (x >> 1)) != 0) ++result; return result; } boolean bfd_generic_is_local_label_name (abfd, name) bfd *abfd; const char *name; { char locals_prefix = (bfd_get_symbol_leading_char (abfd) == '_') ? 'L' : '.'; return (name[0] == locals_prefix); } /* Can be used from / for bfd_merge_private_bfd_data to check that endianness matches between input and output file. Returns true for a match, otherwise returns false and emits an error. */ boolean _bfd_generic_verify_endian_match (ibfd, obfd) bfd *ibfd; bfd *obfd; { if (ibfd->xvec->byteorder != obfd->xvec->byteorder && obfd->xvec->byteorder != BFD_ENDIAN_UNKNOWN) { (*_bfd_error_handler) ("%s: compiled for a %s endian system and target is %s endian", bfd_get_filename (ibfd), bfd_big_endian (ibfd) ? "big" : "little", bfd_big_endian (obfd) ? "big" : "little"); bfd_set_error (bfd_error_wrong_format); return false; } return true; }