* cpu-i860.c: Added comments.
* elf32-i860.c (TARGET_LITTLE_SYM): Defined to
bfd_elf32_i860_little_vec.
(TARGET_LITTLE_NAME): Defined to "elf32-i860-little".
(ELF_MAXPAGESIZE): Changed to 4096.
* targets.c (bfd_elf32_i860_little_vec): Declaration of
new target.
(bfd_target_vector): Added bfd_elf32_i860_little_vec.
* config.bfd (i860-stardent-sysv4*, i860-stardent-elf*): Added
config for little endian elf32 i860.
(targ_defvec): Define for the new config above
as "bfd_elf32_i860_little_vec".
(targ_selvecs): Define for the new config above
as "bfd_elf32_i860_vec bfd_elf32_i860_little_vec"
* configure.in (bfd_elf32_i860_little_vec): Added recognition
of new target vec.
* configure: Regenerated.
* include/opcode/i860.h (btne, bte, bla): Changed these opcodes
to use sbroff ('r') instead of split16 ('s').
(J, K, L, M): New operand types for 16-bit aligned fields.
(ld.x, {p}fld.x, fst.x, pst.d): Changed these opcodes to
use I, J, K, L, M instead of just I.
(T, U): New operand types for split 16-bit aligned fields.
(st.x): Changed these opcodes to use S, T, U instead of just S.
(andh, andnoth, orh, xorh): Deleted 3-register forms as they do not
exist on the i860.
(pfgt.sd, pfle.sd): Deleted these as they do not exist on the i860.
(pfeq.ss, pfeq.dd): New opcodes.
(st.s): Fixed incorrect mask bits.
(fmlow): Fixed incorrect mask bits.
(fzchkl, pfzchkl): Fixed incorrect mask bits.
(faddz, pfaddz): Fixed incorrect mask bits.
(form, pform): Fixed incorrect mask bits.
(pfld.l): Fixed incorrect mask bits.
(fst.q): Fixed incorrect mask bits.
(all floating point opcodes): Fixed incorrect mask bits for
handling of dual bit.
* include/elf/i860.h: New file.
(elf_i860_reloc_type): Defined ELF32 i860 relocations.
* bfd/cpu-i860.c: Added comments.
* bfd/elf32-i860.c (TARGET_LITTLE_SYM): Defined to
bfd_elf32_i860_little_vec.
(TARGET_LITTLE_NAME): Defined to "elf32-i860-little".
(ELF_MAXPAGESIZE): Changed to 4096.
* bfd/targets.c (bfd_elf32_i860_little_vec): Declaration of
new target.
(bfd_target_vector): Added bfd_elf32_i860_little_vec.
* bfd/config.bfd (i860-stardent-sysv4*, i860-stardent-elf*): Added
config for little endian elf32 i860.
(targ_defvec): Define for the new config above
as "bfd_elf32_i860_little_vec".
(targ_selvecs): Define for the new config above
as "bfd_elf32_i860_vec bfd_elf32_i860_little_vec"
* bfd/configure.in (bfd_elf32_i860_little_vec): Added recognition
of new target vec.
* bfd/configure: Regenerated.
* opcodes/i860-dis.c: New file.
(print_insn_i860): New function.
(print_br_address): New function.
(sign_extend): New function.
(BITWISE_OP): New macro.
(I860_REG_PREFIX): New macro.
(grnames, frnames, crnames): New structures.
* opcodes/disassemble.c (ARCH_i860): Define.
(disassembler): Add check for bfd_arch_i860 to set disassemble
function to print_insn_i860.
* include/dis-asm.h (print_insn_i860): Add prototype.
* opcodes/Makefile.in (CFILES): Added i860-dis.c.
(ALL_MACHINES): Added i860-dis.lo.
(i860-dis.lo): New dependences.
* opcodes/configure.in: New bits for bfd_i860_arch.
* opcodes/configure: Regenerated.
* elflink.h (elf_merge_symbol): Take one more argument,
dt_needed, to indicate if the symbol comes from a DT_NEEDED
entry. Don't overide the existing weak definition if dt_needed
is true.
(elf_link_add_object_symbols): Pass dt_needed to
elf_merge_symbol ().
(xcoff_write_armap_big): Write both 32-bit and 64-bit armaps.
(xcoff_write_archive_contents_big): Don't update the offset
of the symbol table, xcoff_write_armap will do it.
* elflink.h (NAME(bfd_elf,size_dynamic_sections)): Don't check
info->new_dtags when setting DT_FLAGS_1. It will only be set
by the new linker options. It shouldn't break anything.
* elf32-arm.h (elf32_arm_size_dynamic_sections): Also set
DF_TEXTREL if DT_TEXTREL is set.
* elf32-i370.c (i370_elf_size_dynamic_sections): Likewise.
* elf32-i386.c (elf_i386_size_dynamic_sections): Likewise.
* elf32-m68k.c (elf_m68k_size_dynamic_sections): Likewise.
* elf32-mips.c (_bfd_mips_elf_size_dynamic_sections): Likewise.
* elf32-ppc.c (ppc_elf_size_dynamic_sections): Likewise.
* elf32-sparc.c (elf32_sparc_size_dynamic_sections): Likewise.
* elf64-alpha.c (elf64_alpha_size_dynamic_sections): Likewise.
* elf64-hppa.c (elf64_hppa_size_dynamic_sections): Likewise.
* elf64-ia64.c (elf64_ia64_size_dynamic_sections): Likewise.
* elf64-sparc.c (sparc64_elf_size_dynamic_sections): Likewise.
* bfd/elflink.h (NAME(bfd_elf,size_dynamic_sections)): Also
set DF_SYMBOLIC for symbolic link. Also set DT_RUNPATH if
DT_RPATH is set.
Set the DT_FLAGS and DT_FLAGS_1 entries if necessary.
* bfd-in.h (bfd_elf_set_dt_needed_soname): New.
* bfd-in2.h: Rebuild.
* elf-bfd.h (elf_obj_tdata): Add dt_soname.
(elf_dt_soname): New.
* elf.c (bfd_elf_set_dt_needed_soname): New.
* elflink.h (elf_link_add_object_symbols): Add the DT_NEEDED
entry if the shared object loaded by DT_NEEDED is used to
resolve the reference in a regular object.
Enable the support for Traditional MIPS.
* elf32-mips.c (IRIX_COMPAT): Recognize bfd_elf32_tradbigmips_vecand
return ict_none appropriately for traditional mips targets.
(STUB_LW): Change 0x8f998000 to 0x8f998010 for traditional mips.
(STUB_MOVE): Conditionalize for traditonal mips.
(STUB_LI16): Likewise.
(_bfd_mips_elf_modify_segment_map): Conditionalize to avoid making
room for RTPROC header.
(_bfd_mips_elf_modify_segment_map): For a normal mips executable set
the permission for the PT_DYNAMIC as read, write and execute.
(mips_elf_calculate_relocation): Check for the symbol _DYNAMIC_LINKING
for traditonal mips.
(_bfd_mips_elf_create_dynamic_sections): Add the symbol
_DYNAMIC_LINKING for traditonal mips.
(_bfd_mips_elf_create_dynamic_sections): Add the symbol __RLD_MAP
in case of traditonal mips.
(_bfd_mips_elf_adjust_dynamic_symbol): Create a stub only if a PLT
entry is required. For a function if PLT is not required then set the
corresponding hash table entry to 0.
(_bfd_mips_elf_size_dynamic_sections): Add DT_DEBUG entry for
traditonal mips.
(_bfd_mips_elf_finish_dynamic_symbol): for a undefined symbol in a
shared object set the value to 0.
(_bfd_mips_elf_finish_dynamic_symbol): Check for the symbol
_DYNAMIC_LINKING for traditonal mips.
(_bfd_mips_elf_finish_dynamic_symbol): Check for the symbol __RLD_MAP
for traditonal mips.
* elf32-mips.c (sort_dynamic_relocs): New Function.
(_bfd_mips_elf_finish_dynamic_sections): Call sort_dynamic_relocs
via qsort to sort the dynamic relocations in increasing r_symndx
value.
* config.bfd: Change targ_defvec and targ_selvecs for mips*-*-sysv4*
to add a new target for traditional mips i.e
bfd_elf32_tradbigmips_vec and bfd_elf32_tradlittlemips_vec.
* configure.in: Likewise.
* configure: Rebuild.
* targets.c (bfd_elf32_tradbigmips_vec): Declare and put in
bfd_target_vector.
(bfd_elf32_tradlittlemips_vec): Likewise.
* elfxx-target.h: Add macro INCLUDED_TARGET_FILE which is more a test
to see that elfNN_bed does not get redefined even if the target file
is included twice for a chip. See elf32-mips.c.
* elf-bfd.h (struct elf_obj_tdata): Define per BFD Irix 5 virtual
sections elf_{text,data}_{section,symbol}.
* elf32-mips.c: mips_elf_{text,data}_{section,symbol}{,_ptr}: Remove.
(_bfd_mips_elf_hide_symbol): New function.
(elf_backend_hide_symbol): Map to the new function.
(_bfd_mips_elf_add_symbol_hook): Change to use new per BFD
definitions of mips_elf_{text,data}_{section,symbol}.
(mips_elf_local_relocation_p): Try to find the direct symbol
based on new check_forced argument.
(mips_elf_calculate_relocation): Use new version of
mips_elf_local_relocation_p.
(mips_elf_relocate_section): Likewise.
(_bfd_mips_elf_relocate_section): Likewise.
(mips_elf_sort_hash_table): Only assert that have enough GOT
space.
(mips_elf_got16_entry): Match all 32 bits to the existing GOT
entry if the relocation based on the new external argument.
(mips_elf_create_dynamic_relocation): Assert that we have a
section contents allocated where we can swap out the dynamic
relocations.
(mips_elf_calculate_relocation): Find the real hash-table entry
correctly by using h->root.root.type. Only create a dynamic
relocation entry if the symbol is defined in a shared library.
Create an external GOT entry for the GOT16 relocation if the
symbol was forced local.
(_bfd_mips_elf_finish_dynamic_symbol): Don't assert there is a
dynamic index if the symbol was forced local.
2000-06-20 Maciej W. Rozycki <macro@ds2.pg.gda.pl>
* elf32-mips.c: Fix typos in comments.
* coff-i386.c (coff_i386_reloc): Don't return in case of
output_bfd == (bfd *) NULL if COFF_WITH_PE is defined.
Compensate PE relocations when linking with non-PE object
files to generate a non-PE executable.
* elflink.h (elf_link_adjust_relocs): Check for and call backend
specifific swap_reloc_{in,out} and swap_reloca_{in,out} if
available.
(elf_link_output_relocs): Likewise.
(elf_reloc_link_order): Likewise.
PowerPc and RS6000 machine numbers.
* bfd-in2.h: Regenerate.
* coffcode.h (coff_set_arch_mach_hook): #ifdef XCOFF64, set arch
to bfd_arch_powerpc instead of bfd_arch_rs6000. Refer to PowerPc
and RS6000 machine numbers using #defined constants from
archures.c.
* cpu-powerpc.c (arch_info_struct): Refer to PowerPc and RS6000
machine numbers using #defined constants from archures.c. Add
entries for EC603e, 630, A35, RS64II, RS64III, 7400. Specify
64-bit words in 620 entry.
* cpu-rs6000.c (arch_info_struct): Create with entries for RS1,
RSC, and RS2.
(bfd_rs6000_arch): Change default machine to 0 (bfd_mach_rs6k).
files on AIX 4.3.
(read_hdr): New function.
(rs6000coff_core_p): Store mstsave or __context64 struct instead
of trying to extract individual registers. Set abfd->arch_info
to match the architecture that created the core file.
(rs6000coff_get_section_contents): Delete.
* xcoff-target.h (rs6000coff_get_section_contents): Delete.
* elf.c (elfcore_grok_prstatus, elfcore_grok_pstatus,
elfcore_grok_psinfo): Add code to allow debugging a 32-bit
corefile on a 64-bit (Sparc Solaris) host. Also clean up
a few old comments.
* elflink.c (_bfd_elf_link_record_dynamic_symbol): Don't flag
an error when seeing an undefined symbol with hidden/internal
attribute. It is handled in *_relocate_section ().
* elflink.h (elf_fix_symbol_flags): Follow the link for the
indirect symbol for the ELF_LINK_NON_ELF bit.
(elf_link_output_extsym): Don't output the indirect symbol even
if the ELF_LINK_NON_ELF bit is set.
* elflink.c (_bfd_elf_link_record_dynamic_symbol): Clear the
visibility bits if the symbol is undefined. Correctly handle
weak undefined symbols with hidden and internal attributes.
* elflink.h (elf_link_add_object_symbols): Always turn the
symbol into local if it has the hidden or internal attribute.
Remove unusued variables.
(sparc64_elf_relocate_section): Change r_symndx type to unsigned long.
(sparc64_elf_merge_private_bfd_data): Fix shared library case from
previous fix, so that shared libs really don't influence targets
extension mask and memory model.
* elflink.h (elf_link_add_object_symbols): Reset dynindx for
hidden and internal symbols.
(elf_fix_symbol_flags): Clear NEEDS_PLT for symbols with
visibility.
* elflink.c (_bfd_elf_link_record_dynamic_symbol): Do not
assign a PLT or GOT entry to symbols with hidden and
internal visibility.
coffdu-rs6000.{lo,c}.
(coff-pmac.lo, coff-rs6000.lo, coff64-rs6000.lo): Add dependency
on xcoff.h
* Makefile.in: Regenerate.
* xcoff.h: New file.
* coff-pmac.c: Use xcoff.h instead of coff-rs6000.c.
* coff-rs6000.c: Move all declarations and defines that are
common to the xcoff backends into xcoff.h
* coff64-rs6000.c: Ditto,
bfd:
* Makefile.am (coff64-rs6000.lo): New rule.
* Makefile.in: Regenerate.
* coff-rs6000.c (xcoff_mkobject, xcoff_copy_private_bfd_data,
xcoff_is_local_label_name, xcoff_rtype2howto,
xcoff_reloc_type_lookup, xcoff_slurp_armap, xcoff_archive_p,
xcoff_read_ar_hdr, xcoff_openr_next_archived_file, xcoff_write_armap,
xcoff_write_archive_contents): No longer static, and prefix with _bfd_.
(NO_COFF_SYMBOLS): Define.
(xcoff64_swap_sym_in, xcoff64_swap_sym_out, xcoff64_swap_aux_in,
xcoff64_swap_aux_out): New functions; handle xcoff symbol tables
internally.
(MINUS_ONE): New macro.
(xcoff_howto_tabl, xcoff_reloc_type_lookup): Add 64 bit POS
relocation.
(coff_SWAP_sym_in, coff_SWAP_sym_out, coff_SWAP_aux_in,
coff_SWAP_aux_out): Map to the new functions.
* coff64-rs6000.c: New file.
* libcoff.h (bfd_coff_backend_data): Add new fields
_bfd_coff_force_symnames_in_strings and
_bfd_coff_debug_string_prefix_length.
(bfd_coff_force_symnames_in_strings,
bfd_coff_debug_string_prefix_length): New macros for above fields.
* coffcode.h (coff_set_arch_mach_hook): Handle XCOFF64 magic.
Set machine to 620 for XCOFF64. Use bfd_coff_swap_sym_in instead
of using coff_swap_sym_in directly.
(FORCE_SYMNAMES_IN_STRINGS): New macro, defined for XCOFF64.
(coff_set_flags) Set magic for XCOFF64.
(coff_compute_section_file_positions): Add symbol name length to
string section length if bfd_coff_debug_string_prefix_length is
true.
(coff_write_object_contents): Don't do reloc overflow for XCOFF64.
(coff_slurp_line_table): Use bfd_coff_swap_lineno_in instead of
using coff_swap_lineno_in directly.
(bfd_coff_backend_data): Add _bfd_coff_force_symnames_in_strings
and _bfd_coff_debug_string_prefix_length fields.
* coffgen.c (coff_fix_symbol_name, coff_write_symbols): Force
symbol names into strings table when
bfd_coff_force_symnames_in_strings is true.
* coffswap.h (MAX_SCNHDR_NRELOC, MAX_SCNHDR_NLNNO, GET_RELOC_VADDR,
SET_RELOC_VADDR): New macros.
(coff_swap_reloc_in, coff_swap_reloc_out): Use above macros.
(coff_swap_aux_in, coff_swap_aux_out): Remove RS6000COFF_C
code.
(coff_swap_aouthdr_in, coff_swap_aouthdr_out): Handle XCOFF64
changes within RS6000COFF_C specific code.
(coff_swap_scnhdr_out): Use PUT_SCNHDR_NLNNO, PUT_SCNHDR_NRELOC,
MAX_SCNHDR_NRELOC, and MAX_SCNHDR_NLNNO.
* reloc.c (bfd_perform_relocation, bfd_install_relocation):
Extend existing hack on target name.
* xcofflink.c (XCOFF_XVECP): Extend existing hack on
target name.
* coff-tic54x.c (ticof): Keep up to date with new fields
in bfd_coff_backend_data.
* config.bfd: Add bfd_powerpc_64_arch to targ_arch and define
targ_selvecs to include rs6000coff64_vec for rs6000.
* configure.in: Add rs6000coff64_vec case.
* cpu-powerpc.c: New bfd_arch_info_type.
gas:
* as.c (parse_args): Allow md_parse_option to override -a listing
option.
* config/obj-coff.c (add_lineno): Change type of offset parameter
from "int" to "bfd_vma."
* config/tc-ppc.c (md_pseudo_table): Add "llong" and "machine."
(ppc_mach, ppc_subseg_align, ppc_target_format): New.
(ppc_change_csect): Align correctly for XCOFF64.
(ppc_machine): New function, which discards "ppc_machine" line.
(ppc_tc): Cons for 8 when code is 64 bit.
(md_apply_fix3): Don't check operand->insert. Handle 64 bit
relocations.
(md_parse_option): Handle -a64 and -a32.
(ppc_xcoff64): New.
* config/tc-ppc.h (TARGET_MACH): Define.
(TARGET_FORMAT): Move to function.
(SUB_SEGMENT_ALIGN): Use ppc_subseg_align.
include:
* include/coff/rs6k64.h: New file.
opcodes:
* configure.in: Add bfd_powerpc_64_arch.
* disassemble.c (disassembler): Use print_insn_big_powerpc for
64 bit code.
bfd_elf32_slurp_reloc_table, bfd_elf64_write_relocs, and
bfd_elf64_slurp_reloc_table.
* elfcode.h (elf_write_relocs, elf_slurp_reloc_table): New
definitions to get external names.
(elf_write_relocs): Renamed from write_relocs and make global.
(elf_slurp_reloc_table): Make global.
(_bfd_elf,size_info): Use elf_write_relocs instead of write_relocs.
(bfd_ar_hdr_from_filesystem): Use it if HPUX_LARGE_AR_IDS is
defined and the ID is greater than 99999.
(bfd_generic_stat_arch_elt): If HPUX_LARGE_AR_IDS is defined decode
special uid/gid fields into 32 bit values.
Add ``-W -Wall'' to sub-directories bfd, binutils, gas gprof, ld and
opcodes by the addition of WARN_CFLAGS to Makefile.am and configury to
set it. Add configure option --enable-build-warnings.
Re-generate all and sundry using auto*-000227.
* dwarf2.c (struct dwarf2_debug): New field dwarf_line_size.
(decode_line_info): Set it. Report error if unit->line_offset is
equal to or larger than it.
opcodes:
* sh-opc.h (sh_nibble_type): Remove DISP_8 and DISP_4.
Split IMM_[48]{,BY[24]} into IMM[01]_[48]{,BY[24]}. Add REPEAT.
(sh_arg_type): Add A_PC.
(sh_table): Update entries using immediates. Add repeat.
* sh-dis.c (print_insn_shx): Remove DISP_8 and DISP_4.
Split IMM_[48]{,BY[24]} into IMM[01]_[48]{,BY[24]}. Add REPEAT.
gas:
* config/tc-sh.c (immediate): Delete.
(sh_operand_info): Add immediate member.
(parse_reg): Use A_PC for pc.
(parse_exp): Add second argument 'op'. All callers changed.
(parse_at): Expect pc to be coded as A_PC.
Use immediate field in *op.
(insert): Add fourth argument 'op'. All callers changed.
(build_relax): Add second argument 'op'. All callers changed.
(insert_loop_bounds): New function.
(build_Mytes): Remove DISP_4.
Split IMM_[48]{,BY[24]} into IMM[01]_[48]{,BY[24]}. Add REPEAT.
(assemble_ppi): Use immediate field in *operand.
(sh_force_relocation): Handle BFD_RELOC_SH_LOOP_{START,END}.
(md_apply_fix): Likewise.
(tc_gen_reloc): Likewise. Check for a pcrel BFD_RELOC_SH_LABEL.
include/coff:
* sh.h (R_SH_LOOP_START, R_SH_LOOP_END): Define.
include/elf:
* sh.h (R_SH_LOOP_START, R_SH_LOOP_END): New RELOC_NUMBERs.
bfd:
* reloc.c (_bfd_relocate_contents): Add BFD_RELOC_SH_LOOP_START and
BFD_RELOC_SH_LOOP_END.
* elf32-sh.c (sh_elf_howto_tab): Change special_func to
sh_elf_ignore_reloc for all entries that sh_elf_reloc used to ignore.
Add entries for R_SH_LOOP_START and R_SH_LOOP_END.
(sh_elf_reloc_loop): New function.
(sh_elf_reloc): No need to test for always-to-be-ignored relocs
any more.
(sh_rel): Add entries for BFD_RELOC_SH_LOOP_{START,END}.
(sh_elf_relocate_section): Handle BFD_RELOC_SH_LOOP_{START,END}.
* bfd-in2.h, libbfd.h: Regenerate.
BFD_RELOC_16 to switch for extended relocs.
(MY_swap_ext_reloc_in): New.
(MY_swap_ext_reloc_out): New.
(NAME(aout,slurp_reloc_table)): Use MY_swap_ext_reloc_in rather
than NAME(aout,swap_ext_reloc_in) for extended relocs.
(NAME(aout,squirt_out_relocs)): Similarly use
MY_swap_ext_reloc_out.
(aout_link_reloc_link_order): Use MY_put_ext_reloc if defined.
HOWTO references to bfd_elf_generic_reloc, that have
partial_inplace == true, now use the new function. The function
is based on the recent rewrite of m32r_elf_lo16_reloc(), and
extends its fixes to the R_M32R_{16,24,32} relocs.
The new logic in m32r_elf_lo16_reloc() has been removed, and
it instead calls the new routine to obtain that functionality.
complain_overflow_bitfield doesn't complain) from -2**(n-1)..2**n-1 to
-2**n..2**n. This might mean that some reloc overflows are no longer
caught, but it solves the address wrap problem for 16-bit relocs
nicely. In any case, ports that rely on complain_overflow_bitfield
for reloc overflow checking were not getting a very good check
previously. A bitfield range in a machine instruction is typically
either the signed or unsigned n bit numbers, not the overlap of these
two ranges.
Nick Clifton
Jason Eckhardt
H.J. Lu
Geoffrey Keating
Ulf Carlsson
Hans-Peter Nilsson
DJ Delorie
Alan Modra
Andrew Cagney
Alexandre Oliva
Jim Wilson
Mark Kettenis
Nicholas Duffek
Joern Rennecke
David O'Brien
Michael Snyder
J.T. Conklin
Ulrich Drepper
Phil Blundell
Jakub Jelinek
Richard Henderson
Jeff Law
Catherine Moore
Clinton Popetz
Ben Elliston
Timothy Wall
Kevin Buettner
Ian Lance Taylor
Doug Evans