/* BFD back-end for IBM RS/6000 "XCOFF" files.
- Copyright 1990, 91, 92, 93, 94, 95, 96, 1997 Free Software Foundation, Inc.
+ Copyright 1990, 91, 92, 93, 94, 95, 96, 97, 98, 1999
+ Free Software Foundation, Inc.
FIXME: Can someone provide a transliteration of this name into ASCII?
Using the following chars caused a compiler warning on HIUX (so I replaced
them with octal escapes), and isn't useful without an understanding of what
static boolean xcoff_write_armap
PARAMS ((bfd *, unsigned int, struct orl *, unsigned int, int));
static boolean xcoff_write_archive_contents PARAMS ((bfd *));
+static int _bfd_xcoff_sizeof_headers PARAMS ((bfd *, boolean));
\f
/* We use our own tdata type. Its first field is the COFF tdata type,
so the COFF routines are compatible. */
static boolean
xcoff_is_local_label_name (abfd, name)
- bfd *abfd;
- const char *name;
+ bfd *abfd ATTRIBUTE_UNUSED;
+ const char *name ATTRIBUTE_UNUSED;
{
return false;
}
0xffff, /* dst_mask */
false), /* pcrel_offset */
- { 7 },
+ EMPTY_HOWTO (7),
/* Non modifiable absolute branch. */
HOWTO (8, /* type */
0x3fffffc, /* dst_mask */
false), /* pcrel_offset */
- { 9 },
+ EMPTY_HOWTO (9),
/* Non modifiable relative branch. */
HOWTO (0xa, /* type */
0x3fffffc, /* dst_mask */
false), /* pcrel_offset */
- { 0xb },
+ EMPTY_HOWTO (0xb),
/* Indirect load. */
HOWTO (0xc, /* type */
0xffff, /* dst_mask */
false), /* pcrel_offset */
- { 0xe },
+ EMPTY_HOWTO (0xe),
/* Non-relocating reference. */
HOWTO (0xf, /* type */
0, /* dst_mask */
false), /* pcrel_offset */
- { 0x10 },
- { 0x11 },
+ EMPTY_HOWTO (0x10),
+ EMPTY_HOWTO (0x11),
/* TOC relative indirect load. */
HOWTO (0x12, /* type */
false) /* pcrel_offset */
};
+/* These are the first two like the above but for 16-bit relocs. */
+static reloc_howto_type xcoff_howto_table_16[] =
+{
+ /* Standard 16 bit relocation. */
+ HOWTO (0, /* type */
+ 0, /* rightshift */
+ 2, /* size (0 = byte, 1 = short, 2 = long) */
+ 16, /* bitsize */
+ false, /* pc_relative */
+ 0, /* bitpos */
+ complain_overflow_bitfield, /* complain_on_overflow */
+ 0, /* special_function */
+ "R_POS_16", /* name */
+ true, /* partial_inplace */
+ 0xffffffff, /* src_mask */
+ 0xffffffff, /* dst_mask */
+ false), /* pcrel_offset */
+
+ /* 16 bit relocation, but store negative value. */
+ HOWTO (1, /* type */
+ 0, /* rightshift */
+ -2, /* size (0 = byte, 1 = short, 2 = long) */
+ 16, /* bitsize */
+ false, /* pc_relative */
+ 0, /* bitpos */
+ complain_overflow_bitfield, /* complain_on_overflow */
+ 0, /* special_function */
+ "R_NEG_16", /* name */
+ true, /* partial_inplace */
+ 0xffffffff, /* src_mask */
+ 0xffffffff, /* dst_mask */
+ false), /* pcrel_offset */
+
+ /* 16 bit PC relative relocation. */
+ HOWTO (2, /* type */
+ 0, /* rightshift */
+ 2, /* size (0 = byte, 1 = short, 2 = long) */
+ 32, /* bitsize */
+ true, /* pc_relative */
+ 0, /* bitpos */
+ complain_overflow_signed, /* complain_on_overflow */
+ 0, /* special_function */
+ "R_REL_16", /* name */
+ true, /* partial_inplace */
+ 0xffffffff, /* src_mask */
+ 0xffffffff, /* dst_mask */
+ false), /* pcrel_offset */
+};
+
static void
xcoff_rtype2howto (relent, internal)
arelent *relent;
{
relent->howto = xcoff_howto_table + internal->r_type;
+ if (relent->howto->bitsize != ((unsigned int) internal->r_size & 0x1f) + 1
+ && (internal->r_type
+ < sizeof(xcoff_howto_table_16)/sizeof(xcoff_howto_table_16[0])))
+ relent->howto = xcoff_howto_table_16 + internal->r_type;
+
/* The r_size field of an XCOFF reloc encodes the bitsize of the
relocation, as well as indicating whether it is signed or not.
Doublecheck that the relocation information gathered from the
static reloc_howto_type *
xcoff_reloc_type_lookup (abfd, code)
- bfd *abfd;
+ bfd *abfd ATTRIBUTE_UNUSED;
bfd_reloc_code_real_type code;
{
switch (code)
binary number which is the number of entries. This is followed by
a that many four byte binary numbers; each is the file offset of an
entry in the archive. These numbers are followed by a series of
- null terminated strings, which are symbol names. */
+ null terminated strings, which are symbol names.
-/* XCOFF archives use this as a magic string. */
+ AIX 4.3 introduced a new archive format which can handle larger
+ files and also 32- and 64-bit objects in the same archive. The
+ things said above remain true except that there is now more than
+ one global symbol table. The one is used to index 32-bit objects,
+ the other for 64-bit objects.
-#define XCOFFARMAG "<aiaff>\012"
-#define SXCOFFARMAG 8
+ The new archives (recognizable by the new ARMAG string) has larger
+ field lengths so that we cannot really share any code. Also we have
+ to take care that we are not generating the new form of archives
+ on AIX 4.2 or earlier systems. */
+
+/* XCOFF archives use this as a magic string. Note that both strings
+ have the same length. */
+
+#define XCOFFARMAG "<aiaff>\012"
+#define XCOFFARMAGBIG "<bigaf>\012"
+#define SXCOFFARMAG 8
/* This terminates an XCOFF archive member name. */
#define SIZEOF_AR_FILE_HDR (5 * 12 + SXCOFFARMAG)
+/* This is the equivalent data structure for the big archive format. */
+
+struct xcoff_ar_file_hdr_big
+{
+ /* Magic string. */
+ char magic[SXCOFFARMAG];
+
+ /* Offset of the member table (decimal ASCII string). */
+ char memoff[20];
+
+ /* Offset of the global symbol table for 32-bit objects (decimal ASCII
+ string). */
+ char symoff[20];
+
+ /* Offset of the global symbol table for 64-bit objects (decimal ASCII
+ string). */
+ char symoff64[20];
+
+ /* Offset of the first member in the archive (decimal ASCII string). */
+ char firstmemoff[20];
+
+ /* Offset of the last member in the archive (decimal ASCII string). */
+ char lastmemoff[20];
+
+ /* Offset of the first member on the free list (decimal ASCII
+ string). */
+ char freeoff[20];
+};
+
+#define SIZEOF_AR_FILE_HDR_BIG (6 * 20 + SXCOFFARMAG)
+
+
/* Each XCOFF archive member starts with this (printable) structure. */
struct xcoff_ar_hdr
#define SIZEOF_AR_HDR (7 * 12 + 4)
+/* The equivalent for the big archive format. */
+
+struct xcoff_ar_hdr_big
+{
+ /* File size not including the header (decimal ASCII string). */
+ char size[20];
+
+ /* File offset of next archive member (decimal ASCII string). */
+ char nextoff[20];
+
+ /* File offset of previous archive member (decimal ASCII string). */
+ char prevoff[20];
+
+ /* File mtime (decimal ASCII string). */
+ char date[12];
+
+ /* File UID (decimal ASCII string). */
+ char uid[12];
+
+ /* File GID (decimal ASCII string). */
+ char gid[12];
+
+ /* File mode (octal ASCII string). */
+ char mode[12];
+
+ /* Length of file name (decimal ASCII string). */
+ char namlen[4];
+
+ /* This structure is followed by the file name. The length of the
+ name is given in the namlen field. If the length of the name is
+ odd, the name is followed by a null byte. The name and optional
+ null byte are followed by XCOFFARFMAG, which is not included in
+ namlen. The contents of the archive member follow; the number of
+ bytes is given in the size field. */
+};
+
+#define SIZEOF_AR_HDR_BIG (3 * 20 + 4 * 12 + 4)
+
+
+/* We often have to distinguish between the old and big file format.
+ Make it a bit cleaner. We can use `xcoff_ardata' here because the
+ `hdr' member has the same size and position in both formats. */
+#define xcoff_big_format_p(abfd) \
+ (xcoff_ardata (abfd)->magic[1] == 'b')
+
/* We store a copy of the xcoff_ar_file_hdr in the tdata field of the
- artdata structure. */
+ artdata structure. Similar for the big archive. */
#define xcoff_ardata(abfd) \
((struct xcoff_ar_file_hdr *) bfd_ardata (abfd)->tdata)
+#define xcoff_ardata_big(abfd) \
+ ((struct xcoff_ar_file_hdr_big *) bfd_ardata (abfd)->tdata)
/* We store a copy of the xcoff_ar_hdr in the arelt_data field of an
- archive element. */
+ archive element. Similar for the big archive. */
#define arch_eltdata(bfd) ((struct areltdata *) ((bfd)->arelt_data))
#define arch_xhdr(bfd) \
((struct xcoff_ar_hdr *) arch_eltdata (bfd)->arch_header)
+#define arch_xhdr_big(bfd) \
+ ((struct xcoff_ar_hdr_big *) arch_eltdata (bfd)->arch_header)
/* XCOFF archives do not have anything which corresponds to an
extended name table. */
bfd *abfd;
{
file_ptr off;
- struct xcoff_ar_hdr hdr;
size_t namlen;
bfd_size_type sz;
bfd_byte *contents, *cend;
return true;
}
- off = strtol (xcoff_ardata (abfd)->symoff, (char **) NULL, 10);
- if (off == 0)
+ if (! xcoff_big_format_p (abfd))
{
- bfd_has_map (abfd) = false;
- return true;
+ /* This is for the old format. */
+ struct xcoff_ar_hdr hdr;
+
+ off = strtol (xcoff_ardata (abfd)->symoff, (char **) NULL, 10);
+ if (off == 0)
+ {
+ bfd_has_map (abfd) = false;
+ return true;
+ }
+
+ if (bfd_seek (abfd, off, SEEK_SET) != 0)
+ return false;
+
+ /* The symbol table starts with a normal archive header. */
+ if (bfd_read ((PTR) &hdr, SIZEOF_AR_HDR, 1, abfd) != SIZEOF_AR_HDR)
+ return false;
+
+ /* Skip the name (normally empty). */
+ namlen = strtol (hdr.namlen, (char **) NULL, 10);
+ if (bfd_seek (abfd, ((namlen + 1) & ~1) + SXCOFFARFMAG, SEEK_CUR) != 0)
+ return false;
+
+ sz = strtol (hdr.size, (char **) NULL, 10);
}
+ else
+ {
+ /* This is for the new format. */
+ struct xcoff_ar_hdr_big hdr;
- if (bfd_seek (abfd, off, SEEK_SET) != 0)
- return false;
+ off = strtol (xcoff_ardata_big (abfd)->symoff, (char **) NULL, 10);
+ if (off == 0)
+ {
+ bfd_has_map (abfd) = false;
+ return true;
+ }
- /* The symbol table starts with a normal archive header. */
- if (bfd_read ((PTR) &hdr, SIZEOF_AR_HDR, 1, abfd) != SIZEOF_AR_HDR)
- return false;
+ if (bfd_seek (abfd, off, SEEK_SET) != 0)
+ return false;
- /* Skip the name (normally empty). */
- namlen = strtol (hdr.namlen, (char **) NULL, 10);
- if (bfd_seek (abfd, ((namlen + 1) & ~1) + SXCOFFARFMAG, SEEK_CUR) != 0)
- return false;
+ /* The symbol table starts with a normal archive header. */
+ if (bfd_read ((PTR) &hdr, SIZEOF_AR_HDR_BIG, 1, abfd)
+ != SIZEOF_AR_HDR_BIG)
+ return false;
+
+ /* Skip the name (normally empty). */
+ namlen = strtol (hdr.namlen, (char **) NULL, 10);
+ if (bfd_seek (abfd, ((namlen + 1) & ~1) + SXCOFFARFMAG, SEEK_CUR) != 0)
+ return false;
+
+ /* XXX This actually has to be a call to strtoll (at least on 32-bit
+ machines) since the field width is 20 and there numbers with more
+ than 32 bits can be represented. */
+ sz = strtol (hdr.size, (char **) NULL, 10);
+ }
/* Read in the entire symbol table. */
- sz = strtol (hdr.size, (char **) NULL, 10);
contents = (bfd_byte *) bfd_alloc (abfd, sz);
if (contents == NULL)
return false;
xcoff_archive_p (abfd)
bfd *abfd;
{
- struct xcoff_ar_file_hdr hdr;
+ char magic[SXCOFFARMAG];
- if (bfd_read ((PTR) &hdr, SIZEOF_AR_FILE_HDR, 1, abfd)
- != SIZEOF_AR_FILE_HDR)
+ if (bfd_read ((PTR) magic, SXCOFFARMAG, 1, abfd) != SXCOFFARMAG)
{
if (bfd_get_error () != bfd_error_system_call)
bfd_set_error (bfd_error_wrong_format);
return NULL;
}
- if (strncmp (hdr.magic, XCOFFARMAG, SXCOFFARMAG) != 0)
+ if (strncmp (magic, XCOFFARMAG, SXCOFFARMAG) != 0
+ && strncmp (magic, XCOFFARMAGBIG, SXCOFFARMAG) != 0)
{
bfd_set_error (bfd_error_wrong_format);
return NULL;
if (bfd_ardata (abfd) == (struct artdata *) NULL)
return NULL;
- bfd_ardata (abfd)->first_file_filepos = strtol (hdr.firstmemoff,
- (char **) NULL, 10);
bfd_ardata (abfd)->cache = NULL;
bfd_ardata (abfd)->archive_head = NULL;
bfd_ardata (abfd)->symdefs = NULL;
bfd_ardata (abfd)->extended_names = NULL;
- bfd_ardata (abfd)->tdata = bfd_zalloc (abfd, SIZEOF_AR_FILE_HDR);
- if (bfd_ardata (abfd)->tdata == NULL)
- return NULL;
+ /* Now handle the two formats. */
+ if (magic[1] != 'b')
+ {
+ /* This is the old format. */
+ struct xcoff_ar_file_hdr hdr;
+
+ /* Copy over the magic string. */
+ memcpy (hdr.magic, magic, SXCOFFARMAG);
+
+ /* Now read the rest of the file header. */
+ if (bfd_read ((PTR) &hdr.memoff, SIZEOF_AR_FILE_HDR - SXCOFFARMAG, 1,
+ abfd) != SIZEOF_AR_FILE_HDR - SXCOFFARMAG)
+ {
+ if (bfd_get_error () != bfd_error_system_call)
+ bfd_set_error (bfd_error_wrong_format);
+ return NULL;
+ }
+
+ bfd_ardata (abfd)->first_file_filepos = strtol (hdr.firstmemoff,
+ (char **) NULL, 10);
+
+ bfd_ardata (abfd)->tdata = bfd_zalloc (abfd, SIZEOF_AR_FILE_HDR);
+ if (bfd_ardata (abfd)->tdata == NULL)
+ return NULL;
+
+ memcpy (bfd_ardata (abfd)->tdata, &hdr, SIZEOF_AR_FILE_HDR);
+ }
+ else
+ {
+ /* This is the new format. */
+ struct xcoff_ar_file_hdr_big hdr;
+
+ /* Copy over the magic string. */
+ memcpy (hdr.magic, magic, SXCOFFARMAG);
+
+ /* Now read the rest of the file header. */
+ if (bfd_read ((PTR) &hdr.memoff, SIZEOF_AR_FILE_HDR_BIG - SXCOFFARMAG, 1,
+ abfd) != SIZEOF_AR_FILE_HDR_BIG - SXCOFFARMAG)
+ {
+ if (bfd_get_error () != bfd_error_system_call)
+ bfd_set_error (bfd_error_wrong_format);
+ return NULL;
+ }
- memcpy (bfd_ardata (abfd)->tdata, &hdr, SIZEOF_AR_FILE_HDR);
+ /* XXX This actually has to be a call to strtoll (at least on 32-bit
+ machines) since the field width is 20 and there numbers with more
+ than 32 bits can be represented. */
+ bfd_ardata (abfd)->first_file_filepos = strtol (hdr.firstmemoff,
+ (char **) NULL, 10);
+
+ bfd_ardata (abfd)->tdata = bfd_zalloc (abfd, SIZEOF_AR_FILE_HDR_BIG);
+ if (bfd_ardata (abfd)->tdata == NULL)
+ return NULL;
+
+ memcpy (bfd_ardata (abfd)->tdata, &hdr, SIZEOF_AR_FILE_HDR_BIG);
+ }
if (! xcoff_slurp_armap (abfd))
{
xcoff_read_ar_hdr (abfd)
bfd *abfd;
{
- struct xcoff_ar_hdr hdr;
size_t namlen;
- struct xcoff_ar_hdr *hdrp;
struct areltdata *ret;
- if (bfd_read ((PTR) &hdr, SIZEOF_AR_HDR, 1, abfd) != SIZEOF_AR_HDR)
- return NULL;
-
- namlen = strtol (hdr.namlen, (char **) NULL, 10);
- hdrp = bfd_alloc (abfd, SIZEOF_AR_HDR + namlen + 1);
- if (hdrp == NULL)
- return NULL;
- memcpy (hdrp, &hdr, SIZEOF_AR_HDR);
- if (bfd_read ((char *) hdrp + SIZEOF_AR_HDR, 1, namlen, abfd) != namlen)
- return NULL;
- ((char *) hdrp)[SIZEOF_AR_HDR + namlen] = '\0';
-
ret = (struct areltdata *) bfd_alloc (abfd, sizeof (struct areltdata));
if (ret == NULL)
return NULL;
- ret->arch_header = (char *) hdrp;
- ret->parsed_size = strtol (hdr.size, (char **) NULL, 10);
- ret->filename = (char *) hdrp + SIZEOF_AR_HDR;
+
+ if (! xcoff_big_format_p (abfd))
+ {
+ struct xcoff_ar_hdr hdr;
+ struct xcoff_ar_hdr *hdrp;
+
+ if (bfd_read ((PTR) &hdr, SIZEOF_AR_HDR, 1, abfd) != SIZEOF_AR_HDR)
+ {
+ free (ret);
+ return NULL;
+ }
+
+ namlen = strtol (hdr.namlen, (char **) NULL, 10);
+ hdrp = (struct xcoff_ar_hdr *) bfd_alloc (abfd,
+ SIZEOF_AR_HDR + namlen + 1);
+ if (hdrp == NULL)
+ {
+ free (ret);
+ return NULL;
+ }
+ memcpy (hdrp, &hdr, SIZEOF_AR_HDR);
+ if (bfd_read ((char *) hdrp + SIZEOF_AR_HDR, 1, namlen, abfd) != namlen)
+ {
+ free (ret);
+ return NULL;
+ }
+ ((char *) hdrp)[SIZEOF_AR_HDR + namlen] = '\0';
+
+ ret->arch_header = (char *) hdrp;
+ ret->parsed_size = strtol (hdr.size, (char **) NULL, 10);
+ ret->filename = (char *) hdrp + SIZEOF_AR_HDR;
+ }
+ else
+ {
+ struct xcoff_ar_hdr_big hdr;
+ struct xcoff_ar_hdr_big *hdrp;
+
+ if (bfd_read ((PTR) &hdr, SIZEOF_AR_HDR_BIG, 1, abfd)
+ != SIZEOF_AR_HDR_BIG)
+ {
+ free (ret);
+ return NULL;
+ }
+
+ namlen = strtol (hdr.namlen, (char **) NULL, 10);
+ hdrp = (struct xcoff_ar_hdr_big *) bfd_alloc (abfd,
+ SIZEOF_AR_HDR_BIG
+ + namlen + 1);
+ if (hdrp == NULL)
+ {
+ free (ret);
+ return NULL;
+ }
+ memcpy (hdrp, &hdr, SIZEOF_AR_HDR_BIG);
+ if (bfd_read ((char *) hdrp + SIZEOF_AR_HDR_BIG, 1, namlen, abfd) != namlen)
+ {
+ free (ret);
+ return NULL;
+ }
+ ((char *) hdrp)[SIZEOF_AR_HDR_BIG + namlen] = '\0';
+
+ ret->arch_header = (char *) hdrp;
+ /* XXX This actually has to be a call to strtoll (at least on 32-bit
+ machines) since the field width is 20 and there numbers with more
+ than 32 bits can be represented. */
+ ret->parsed_size = strtol (hdr.size, (char **) NULL, 10);
+ ret->filename = (char *) hdrp + SIZEOF_AR_HDR_BIG;
+ }
/* Skip over the XCOFFARFMAG at the end of the file name. */
if (bfd_seek (abfd, (namlen & 1) + SXCOFFARFMAG, SEEK_CUR) != 0)
return NULL;
}
- if (last_file == NULL)
- filestart = bfd_ardata (archive)->first_file_filepos;
+ if (! xcoff_big_format_p (archive))
+ {
+ if (last_file == NULL)
+ filestart = bfd_ardata (archive)->first_file_filepos;
+ else
+ filestart = strtol (arch_xhdr (last_file)->nextoff, (char **) NULL,
+ 10);
+
+ if (filestart == 0
+ || filestart == strtol (xcoff_ardata (archive)->memoff,
+ (char **) NULL, 10)
+ || filestart == strtol (xcoff_ardata (archive)->symoff,
+ (char **) NULL, 10))
+ {
+ bfd_set_error (bfd_error_no_more_archived_files);
+ return NULL;
+ }
+ }
else
- filestart = strtol (arch_xhdr (last_file)->nextoff, (char **) NULL, 10);
-
- if (filestart == 0
- || filestart == strtol (xcoff_ardata (archive)->memoff,
- (char **) NULL, 10)
- || filestart == strtol (xcoff_ardata (archive)->symoff,
- (char **) NULL, 10))
{
- bfd_set_error (bfd_error_no_more_archived_files);
- return NULL;
+ if (last_file == NULL)
+ filestart = bfd_ardata (archive)->first_file_filepos;
+ else
+ /* XXX These actually have to be a calls to strtoll (at least
+ on 32-bit machines) since the fields's width is 20 and
+ there numbers with more than 32 bits can be represented. */
+ filestart = strtol (arch_xhdr_big (last_file)->nextoff, (char **) NULL,
+ 10);
+
+ /* XXX These actually have to be calls to strtoll (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with more
+ than 32 bits can be represented. */
+ if (filestart == 0
+ || filestart == strtol (xcoff_ardata_big (archive)->memoff,
+ (char **) NULL, 10)
+ || filestart == strtol (xcoff_ardata_big (archive)->symoff,
+ (char **) NULL, 10))
+ {
+ bfd_set_error (bfd_error_no_more_archived_files);
+ return NULL;
+ }
}
return _bfd_get_elt_at_filepos (archive, filestart);
bfd *abfd;
struct stat *s;
{
- struct xcoff_ar_hdr *hdrp;
-
if (abfd->arelt_data == NULL)
{
bfd_set_error (bfd_error_invalid_operation);
return -1;
}
- hdrp = arch_xhdr (abfd);
+ if (! xcoff_big_format_p (abfd))
+ {
+ struct xcoff_ar_hdr *hdrp = arch_xhdr (abfd);
+
+ s->st_mtime = strtol (hdrp->date, (char **) NULL, 10);
+ s->st_uid = strtol (hdrp->uid, (char **) NULL, 10);
+ s->st_gid = strtol (hdrp->gid, (char **) NULL, 10);
+ s->st_mode = strtol (hdrp->mode, (char **) NULL, 8);
+ s->st_size = arch_eltdata (abfd)->parsed_size;
+ }
+ else
+ {
+ struct xcoff_ar_hdr_big *hdrp = arch_xhdr_big (abfd);
- s->st_mtime = strtol (hdrp->date, (char **) NULL, 10);
- s->st_uid = strtol (hdrp->uid, (char **) NULL, 10);
- s->st_gid = strtol (hdrp->gid, (char **) NULL, 10);
- s->st_mode = strtol (hdrp->mode, (char **) NULL, 8);
- s->st_size = arch_eltdata (abfd)->parsed_size;
+ s->st_mtime = strtol (hdrp->date, (char **) NULL, 10);
+ s->st_uid = strtol (hdrp->uid, (char **) NULL, 10);
+ s->st_gid = strtol (hdrp->gid, (char **) NULL, 10);
+ s->st_mode = strtol (hdrp->mode, (char **) NULL, 8);
+ s->st_size = arch_eltdata (abfd)->parsed_size;
+ }
return 0;
}
/*ARGSUSED*/
static boolean
-xcoff_write_armap (abfd, elength, map, orl_count, stridx)
+xcoff_write_armap_old (abfd, elength, map, orl_count, stridx)
bfd *abfd;
- unsigned int elength;
+ unsigned int elength ATTRIBUTE_UNUSED;
struct orl *map;
unsigned int orl_count;
int stridx;
if (bfd_write ((PTR) &hdr, SIZEOF_AR_HDR, 1, abfd) != SIZEOF_AR_HDR
|| bfd_write (XCOFFARFMAG, 1, SXCOFFARFMAG, abfd) != SXCOFFARFMAG)
return false;
-
+
bfd_h_put_32 (abfd, orl_count, buf);
if (bfd_write (buf, 1, 4, abfd) != 4)
return false;
return true;
}
-/* Write out an XCOFF archive. We always write an entire archive,
- rather than fussing with the freelist and so forth. */
-
+/*ARGSUSED*/
static boolean
-xcoff_write_archive_contents (abfd)
+xcoff_write_armap_big (abfd, elength, map, orl_count, stridx)
bfd *abfd;
+ unsigned int elength ATTRIBUTE_UNUSED;
+ struct orl *map;
+ unsigned int orl_count;
+ int stridx;
{
- struct xcoff_ar_file_hdr fhdr;
- size_t count;
- size_t total_namlen;
- file_ptr *offsets;
- boolean makemap;
- boolean hasobjects;
- file_ptr prevoff, nextoff;
+ struct xcoff_ar_hdr_big hdr;
+ char *p;
+ unsigned char buf[4];
bfd *sub;
+ file_ptr fileoff;
unsigned int i;
- struct xcoff_ar_hdr ahdr;
- bfd_size_type size;
- char *p;
- char decbuf[13];
- memset (&fhdr, 0, sizeof fhdr);
- strncpy (fhdr.magic, XCOFFARMAG, SXCOFFARMAG);
- sprintf (fhdr.firstmemoff, "%d", SIZEOF_AR_FILE_HDR);
- sprintf (fhdr.freeoff, "%d", 0);
+ memset (&hdr, 0, sizeof hdr);
+ /* XXX This call actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (hdr.size, "%ld", (long) (4 + orl_count * 4 + stridx));
+ sprintf (hdr.nextoff, "%d", 0);
+ memcpy (hdr.prevoff, xcoff_ardata (abfd)->memoff, 12);
+ sprintf (hdr.date, "%d", 0);
+ sprintf (hdr.uid, "%d", 0);
+ sprintf (hdr.gid, "%d", 0);
+ sprintf (hdr.mode, "%d", 0);
+ sprintf (hdr.namlen, "%d", 0);
- count = 0;
- total_namlen = 0;
- for (sub = abfd->archive_head; sub != NULL; sub = sub->next)
- {
- ++count;
- total_namlen += strlen (normalize_filename (sub)) + 1;
- }
- offsets = (file_ptr *) bfd_alloc (abfd, count * sizeof (file_ptr));
- if (offsets == NULL)
+ /* We need spaces, not null bytes, in the header. */
+ for (p = (char *) &hdr; p < (char *) &hdr + SIZEOF_AR_HDR_BIG; p++)
+ if (*p == '\0')
+ *p = ' ';
+
+ if (bfd_write ((PTR) &hdr, SIZEOF_AR_HDR_BIG, 1, abfd) != SIZEOF_AR_HDR_BIG
+ || bfd_write (XCOFFARFMAG, 1, SXCOFFARFMAG, abfd) != SXCOFFARFMAG)
return false;
- if (bfd_seek (abfd, SIZEOF_AR_FILE_HDR, SEEK_SET) != 0)
+ bfd_h_put_32 (abfd, orl_count, buf);
+ if (bfd_write (buf, 1, 4, abfd) != 4)
return false;
- makemap = bfd_has_map (abfd);
- hasobjects = false;
- prevoff = 0;
- nextoff = SIZEOF_AR_FILE_HDR;
- for (sub = abfd->archive_head, i = 0; sub != NULL; sub = sub->next, i++)
+ sub = abfd->archive_head;
+ fileoff = SIZEOF_AR_FILE_HDR_BIG;
+ i = 0;
+ while (sub != NULL && i < orl_count)
+ {
+ size_t namlen;
+
+ while (((bfd *) (map[i]).pos) == sub)
+ {
+ bfd_h_put_32 (abfd, fileoff, buf);
+ if (bfd_write (buf, 1, 4, abfd) != 4)
+ return false;
+ ++i;
+ }
+ namlen = strlen (normalize_filename (sub));
+ namlen = (namlen + 1) &~ 1;
+ fileoff += (SIZEOF_AR_HDR_BIG
+ + namlen
+ + SXCOFFARFMAG
+ + arelt_size (sub));
+ fileoff = (fileoff + 1) &~ 1;
+ sub = sub->next;
+ }
+
+ for (i = 0; i < orl_count; i++)
+ {
+ const char *name;
+ size_t namlen;
+
+ name = *map[i].name;
+ namlen = strlen (name);
+ if (bfd_write (name, 1, namlen + 1, abfd) != namlen + 1)
+ return false;
+ }
+
+ if ((stridx & 1) != 0)
+ {
+ char b;
+
+ b = '\0';
+ if (bfd_write (&b, 1, 1, abfd) != 1)
+ return false;
+ }
+
+ return true;
+}
+
+/*ARGSUSED*/
+static boolean
+xcoff_write_armap (abfd, elength, map, orl_count, stridx)
+ bfd *abfd;
+ unsigned int elength ATTRIBUTE_UNUSED;
+ struct orl *map;
+ unsigned int orl_count;
+ int stridx;
+{
+ if (! xcoff_big_format_p (abfd))
+ return xcoff_write_armap_old (abfd, elength, map, orl_count, stridx);
+ else
+ return xcoff_write_armap_big (abfd, elength, map, orl_count, stridx);
+}
+
+/* Write out an XCOFF archive. We always write an entire archive,
+ rather than fussing with the freelist and so forth. */
+
+static boolean
+xcoff_write_archive_contents_old (abfd)
+ bfd *abfd;
+{
+ struct xcoff_ar_file_hdr fhdr;
+ size_t count;
+ size_t total_namlen;
+ file_ptr *offsets;
+ boolean makemap;
+ boolean hasobjects;
+ file_ptr prevoff, nextoff;
+ bfd *sub;
+ unsigned int i;
+ struct xcoff_ar_hdr ahdr;
+ bfd_size_type size;
+ char *p;
+ char decbuf[13];
+
+ memset (&fhdr, 0, sizeof fhdr);
+ strncpy (fhdr.magic, XCOFFARMAG, SXCOFFARMAG);
+ sprintf (fhdr.firstmemoff, "%d", SIZEOF_AR_FILE_HDR);
+ sprintf (fhdr.freeoff, "%d", 0);
+
+ count = 0;
+ total_namlen = 0;
+ for (sub = abfd->archive_head; sub != NULL; sub = sub->next)
+ {
+ ++count;
+ total_namlen += strlen (normalize_filename (sub)) + 1;
+ }
+ offsets = (file_ptr *) bfd_alloc (abfd, count * sizeof (file_ptr));
+ if (offsets == NULL)
+ return false;
+
+ if (bfd_seek (abfd, SIZEOF_AR_FILE_HDR, SEEK_SET) != 0)
+ return false;
+
+ makemap = bfd_has_map (abfd);
+ hasobjects = false;
+ prevoff = 0;
+ nextoff = SIZEOF_AR_FILE_HDR;
+ for (sub = abfd->archive_head, i = 0; sub != NULL; sub = sub->next, i++)
{
const char *name;
size_t namlen;
if (sub->arelt_data == NULL)
{
- sub->arelt_data = ((struct areltdata *)
- bfd_alloc (sub, sizeof (struct areltdata)));
+ sub->arelt_data = bfd_alloc (sub, sizeof (struct areltdata));
if (sub->arelt_data == NULL)
return false;
}
return true;
}
+
+static boolean
+xcoff_write_archive_contents_big (abfd)
+ bfd *abfd;
+{
+ struct xcoff_ar_file_hdr_big fhdr;
+ size_t count;
+ size_t total_namlen;
+ file_ptr *offsets;
+ boolean makemap;
+ boolean hasobjects;
+ file_ptr prevoff, nextoff;
+ bfd *sub;
+ unsigned int i;
+ struct xcoff_ar_hdr_big ahdr;
+ bfd_size_type size;
+ char *p;
+ char decbuf[13];
+
+ memset (&fhdr, 0, sizeof fhdr);
+ strncpy (fhdr.magic, XCOFFARMAGBIG, SXCOFFARMAG);
+ sprintf (fhdr.firstmemoff, "%d", SIZEOF_AR_FILE_HDR_BIG);
+ sprintf (fhdr.freeoff, "%d", 0);
+
+ count = 0;
+ total_namlen = 0;
+ for (sub = abfd->archive_head; sub != NULL; sub = sub->next)
+ {
+ ++count;
+ total_namlen += strlen (normalize_filename (sub)) + 1;
+ }
+ offsets = (file_ptr *) bfd_alloc (abfd, count * sizeof (file_ptr));
+ if (offsets == NULL)
+ return false;
+
+ if (bfd_seek (abfd, SIZEOF_AR_FILE_HDR_BIG, SEEK_SET) != 0)
+ return false;
+
+ makemap = bfd_has_map (abfd);
+ hasobjects = false;
+ prevoff = 0;
+ nextoff = SIZEOF_AR_FILE_HDR_BIG;
+ for (sub = abfd->archive_head, i = 0; sub != NULL; sub = sub->next, i++)
+ {
+ const char *name;
+ size_t namlen;
+ struct xcoff_ar_hdr_big *ahdrp;
+ bfd_size_type remaining;
+
+ if (makemap && ! hasobjects)
+ {
+ if (bfd_check_format (sub, bfd_object))
+ hasobjects = true;
+ }
+
+ name = normalize_filename (sub);
+ namlen = strlen (name);
+
+ if (sub->arelt_data != NULL)
+ ahdrp = arch_xhdr_big (sub);
+ else
+ ahdrp = NULL;
+
+ if (ahdrp == NULL)
+ {
+ struct stat s;
+
+ memset (&ahdr, 0, sizeof ahdr);
+ ahdrp = &ahdr;
+ /* XXX This should actually be a call to stat64 (at least on
+ 32-bit machines). */
+ if (stat (bfd_get_filename (sub), &s) != 0)
+ {
+ bfd_set_error (bfd_error_system_call);
+ return false;
+ }
+
+ /* XXX This call actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (ahdrp->size, "%ld", (long) s.st_size);
+ sprintf (ahdrp->date, "%ld", (long) s.st_mtime);
+ sprintf (ahdrp->uid, "%ld", (long) s.st_uid);
+ sprintf (ahdrp->gid, "%ld", (long) s.st_gid);
+ sprintf (ahdrp->mode, "%o", (unsigned int) s.st_mode);
+
+ if (sub->arelt_data == NULL)
+ {
+ sub->arelt_data = bfd_alloc (sub, sizeof (struct areltdata));
+ if (sub->arelt_data == NULL)
+ return false;
+ }
+
+ arch_eltdata (sub)->parsed_size = s.st_size;
+ }
+
+ /* XXX These calls actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (ahdrp->prevoff, "%ld", (long) prevoff);
+ sprintf (ahdrp->namlen, "%ld", (long) namlen);
+
+ /* If the length of the name is odd, we write out the null byte
+ after the name as well. */
+ namlen = (namlen + 1) &~ 1;
+
+ remaining = arelt_size (sub);
+ size = (SIZEOF_AR_HDR_BIG
+ + namlen
+ + SXCOFFARFMAG
+ + remaining);
+
+ BFD_ASSERT (nextoff == bfd_tell (abfd));
+
+ offsets[i] = nextoff;
+
+ prevoff = nextoff;
+ nextoff += size + (size & 1);
+
+ sprintf (ahdrp->nextoff, "%ld", (long) nextoff);
+
+ /* We need spaces, not null bytes, in the header. */
+ for (p = (char *) ahdrp; p < (char *) ahdrp + SIZEOF_AR_HDR_BIG; p++)
+ if (*p == '\0')
+ *p = ' ';
+
+ if (bfd_write ((PTR) ahdrp, 1, SIZEOF_AR_HDR_BIG, abfd)
+ != SIZEOF_AR_HDR_BIG
+ || bfd_write ((PTR) name, 1, namlen, abfd) != namlen
+ || (bfd_write ((PTR) XCOFFARFMAG, 1, SXCOFFARFMAG, abfd)
+ != SXCOFFARFMAG))
+ return false;
+
+ if (bfd_seek (sub, (file_ptr) 0, SEEK_SET) != 0)
+ return false;
+ while (remaining != 0)
+ {
+ bfd_size_type amt;
+ bfd_byte buffer[DEFAULT_BUFFERSIZE];
+
+ amt = sizeof buffer;
+ if (amt > remaining)
+ amt = remaining;
+ if (bfd_read (buffer, 1, amt, sub) != amt
+ || bfd_write (buffer, 1, amt, abfd) != amt)
+ return false;
+ remaining -= amt;
+ }
+
+ if ((size & 1) != 0)
+ {
+ bfd_byte b;
+
+ b = '\0';
+ if (bfd_write (&b, 1, 1, abfd) != 1)
+ return false;
+ }
+ }
+
+ /* XXX This call actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (fhdr.lastmemoff, "%ld", (long) prevoff);
+
+ /* Write out the member table. */
+
+ BFD_ASSERT (nextoff == bfd_tell (abfd));
+ /* XXX This call actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (fhdr.memoff, "%ld", (long) nextoff);
+
+ memset (&ahdr, 0, sizeof ahdr);
+ /* XXX The next two calls actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (ahdr.size, "%ld", (long) (12 + count * 12 + total_namlen));
+ sprintf (ahdr.prevoff, "%ld", (long) prevoff);
+ sprintf (ahdr.date, "%d", 0);
+ sprintf (ahdr.uid, "%d", 0);
+ sprintf (ahdr.gid, "%d", 0);
+ sprintf (ahdr.mode, "%d", 0);
+ sprintf (ahdr.namlen, "%d", 0);
+
+ size = (SIZEOF_AR_HDR_BIG
+ + 12
+ + count * 12
+ + total_namlen
+ + SXCOFFARFMAG);
+
+ prevoff = nextoff;
+ nextoff += size + (size & 1);
+
+ if (makemap && hasobjects)
+ /* XXX This call actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (ahdr.nextoff, "%ld", (long) nextoff);
+ else
+ sprintf (ahdr.nextoff, "%d", 0);
+
+ /* We need spaces, not null bytes, in the header. */
+ for (p = (char *) &ahdr; p < (char *) &ahdr + SIZEOF_AR_HDR_BIG; p++)
+ if (*p == '\0')
+ *p = ' ';
+
+ if (bfd_write ((PTR) &ahdr, 1, SIZEOF_AR_HDR_BIG, abfd) != SIZEOF_AR_HDR_BIG
+ || (bfd_write ((PTR) XCOFFARFMAG, 1, SXCOFFARFMAG, abfd)
+ != SXCOFFARFMAG))
+ return false;
+
+ sprintf (decbuf, "%-12ld", (long) count);
+ if (bfd_write ((PTR) decbuf, 1, 12, abfd) != 12)
+ return false;
+ for (i = 0; i < count; i++)
+ {
+ sprintf (decbuf, "%-12ld", (long) offsets[i]);
+ if (bfd_write ((PTR) decbuf, 1, 12, abfd) != 12)
+ return false;
+ }
+ for (sub = abfd->archive_head; sub != NULL; sub = sub->next)
+ {
+ const char *name;
+ size_t namlen;
+
+ name = normalize_filename (sub);
+ namlen = strlen (name);
+ if (bfd_write ((PTR) name, 1, namlen + 1, abfd) != namlen + 1)
+ return false;
+ }
+ if ((size & 1) != 0)
+ {
+ bfd_byte b;
+
+ b = '\0';
+ if (bfd_write ((PTR) &b, 1, 1, abfd) != 1)
+ return false;
+ }
+
+ /* Write out the armap, if appropriate. */
+
+ if (! makemap || ! hasobjects)
+ sprintf (fhdr.symoff, "%d", 0);
+ else
+ {
+ BFD_ASSERT (nextoff == bfd_tell (abfd));
+ /* XXX This call actually should use %lld (at least on 32-bit
+ machines) since the fields's width is 20 and there numbers with
+ more than 32 bits can be represented. */
+ sprintf (fhdr.symoff, "%ld", (long) nextoff);
+ bfd_ardata (abfd)->tdata = (PTR) &fhdr;
+ if (! _bfd_compute_and_write_armap (abfd, 0))
+ return false;
+ }
+
+ /* Write out the archive file header. */
+
+ /* We need spaces, not null bytes, in the header. */
+ for (p = (char *) &fhdr; p < (char *) &fhdr + SIZEOF_AR_FILE_HDR_BIG; p++)
+ if (*p == '\0')
+ *p = ' ';
+
+ if (bfd_seek (abfd, (file_ptr) 0, SEEK_SET) != 0
+ || (bfd_write ((PTR) &fhdr, SIZEOF_AR_FILE_HDR_BIG, 1, abfd) !=
+ SIZEOF_AR_FILE_HDR_BIG))
+ return false;
+
+ return true;
+}
+
+static boolean
+xcoff_write_archive_contents (abfd)
+ bfd *abfd;
+{
+ if (! xcoff_big_format_p (abfd))
+ return xcoff_write_archive_contents_old (abfd);
+ else
+ return xcoff_write_archive_contents_big (abfd);
+}
\f
/* We can't use the usual coff_sizeof_headers routine, because AIX
always uses an a.out header. */
static int
_bfd_xcoff_sizeof_headers (abfd, reloc)
bfd *abfd;
- boolean reloc;
+ boolean reloc ATTRIBUTE_UNUSED;
{
int size;
#define _bfd_xcoff_bfd_get_relocated_section_contents \
coff_bfd_get_relocated_section_contents
#define _bfd_xcoff_bfd_relax_section coff_bfd_relax_section
+#define _bfd_xcoff_bfd_gc_sections coff_bfd_gc_sections
#define _bfd_xcoff_bfd_link_split_section coff_bfd_link_split_section
/* The transfer vector that leads the outside world to all of the above. */
#else
"aixcoff-rs6000", /* name */
#endif
- bfd_target_coff_flavour,
+ bfd_target_coff_flavour,
BFD_ENDIAN_BIG, /* data byte order is big */
BFD_ENDIAN_BIG, /* header byte order is big */
BFD_JUMP_TABLE_LINK (_bfd_xcoff),
BFD_JUMP_TABLE_DYNAMIC (_bfd_xcoff),
- COFF_SWAP_TABLE,
+ NULL,
+
+ COFF_SWAP_TABLE
};
projects / deliverable / binutils-gdb.git / blobdiff