Add target_id to elf_backend_data.

2010-08-25  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/11944
* elf-bfd.h (elf_backend_data): Add target_id.
(bfd_elf_make_generic_object): Renamed to ...
(bfd_elf_make_object): This.

* elf.c (bfd_elf_make_generic_object): Removed.
(bfd_elf_make_object): New.
(bfd_elf_mkcorefile): Really treat it as an object file.

* elf-m10300.c (ELF_TARGET_ID): New.
* elf32-arm.c (ELF_TARGET_ID): Likewise.
* elf32-bfin.c (ELF_TARGET_ID): Likewise.
* elf32-cris.c (ELF_TARGET_ID): Likewise.
* elf32-frv.c (ELF_TARGET_ID): Likewise.
* elf32-i386.c (ELF_TARGET_ID): Likewise.
* elf32-lm32.c (ELF_TARGET_ID): Likewise.
* elf32-m32r.c (ELF_TARGET_ID): Likewise.
* elf32-m68hc11.c (ELF_TARGET_ID): Likewise.
* elf32-m68hc12.c (ELF_TARGET_ID): Likewise.
* elf32-m68k.c (ELF_TARGET_ID): Likewise.
* elf32-microblaze.c (ELF_TARGET_ID): Likewise.
* elf32-ppc.c (ELF_TARGET_ID): Likewise.
* elf32-s390.c (ELF_TARGET_ID): Likewise.
* elf32-sh.c (ELF_TARGET_ID): Likewise.
* elf32-sparc.c (ELF_TARGET_ID): Likewise.
* elf32-spu.c (ELF_TARGET_ID): Likewise.
* elf32-tic6x.c (ELF_TARGET_ID): Likewise.
* elf32-xtensa.c (ELF_TARGET_ID): Likewise.
* elf64-alpha.c (ELF_TARGET_ID): Likewise.
* elf64-hppa.c (ELF_TARGET_ID): Likewise.
* elf64-ppc.c (ELF_TARGET_ID): Likewise.
* elf64-s390.c (ELF_TARGET_ID): Likewise.
* elf64-x86-64.c (ELF_TARGET_ID): Likewise.
* elfxx-ia64.c (ELF_TARGET_ID): Likewise.

* elf32-hppa.c (elf32_hppa_mkobject): Removed.
(bfd_elf32_mkobject): Likewise.
(ELF_TARGET_ID): New.

* elf32-mips.c (ELF_TARGET_ID): New.
(bfd_elf32_mkobject): Removed.

* elf64-mips.c (ELF_TARGET_ID): New.
(bfd_elf64_mkobject): Removed.

* elfn32-mips.c (ELF_TARGET_ID): New.
(bfd_elf32_mkobject): Removed.

* elfxx-mips.c (_bfd_mips_elf_mkobject): Removed.
* elfxx-mips.h (_bfd_mips_elf_mkobject): Likewise.

* elfxx-target.h (bfd_elfNN_mkobject): Default to
bfd_elf_make_object.
(ELF_TARGET_ID): New.  Default to GENERIC_ELF_DATA.
(elfNN_bed): Initialize target_id.

PR threads/10729

* linux-x86-low.c (update_debug_registers_callback): New.
(i386_dr_low_set_addr): Use it.
(i386_dr_low_get_addr): New.
(i386_dr_low_set_control): Use update_debug_registers_callback.
(i386_dr_low_get_control): New.
(i386_dr_low_get_status): Adjust.
* linux-low.c (linux_stop_lwp): New.
* linux-low.h (linux_stop_lwp): Declare.

* i386-low.c (I386_DR_GET_RW_LEN): Take the dr7 contents as
argument instead of a i386_debug_reg_state.
(I386_DR_WATCH_HIT): Take the dr6 contents as argument instead of
a i386_debug_reg_state.
(i386_insert_aligned_watchpoint): Adjust.
(i386_remove_aligned_watchpoint): Adjust.
(i386_low_stopped_data_address): Read the debug registers from the
inferior instead of from the mirrors.
* i386-low.h (struct i386_debug_reg_state): Extend comment.
(i386_dr_low_get_addr): Declare.
(i386_dr_low_get_control): Declare.
(i386_dr_low_get_status): Change prototype.

* win32-i386-low.c (dr_status_mirror, dr_control_mirror): New globals.
(i386_dr_low_get_addr): New.
(i386_dr_low_get_control): New.
(i386_dr_low_get_status): Adjust prototype.  Return
dr_status_mirror.
(i386_initial_stuff): Clear dr_status_mirror and
dr_control_mirror.
(i386_get_thread_context): Adjust.
(i386_set_thread_context): Adjust.
(i386_thread_added): Adjust.

bfd/
    * elf32-arm.c (arm_stub_required_alignment): New.
    (arm_build_one_stub): Use above to partition stubs.
    (make_branch_to_a8_stub): Use arm_stub_a8_veneer_lwm not
    arm_stub_a8_veneer_b_cond.

    ld/testsuite/
    * ld-arm/arm-elf.exp (armelftests): Add cortex-a8-fix-blx-bcond.s.
    * ld-arm/cortex-a8-fix-blx-bcond.s: New.
    * ld-arm/cortex-a8-fix-blx-bcond.d: New.

* configure.tgt (m68k-*-aout): Change to bfd_gas=yes.

* config/tc-d10v.c (do_assemble): Correctly detect overflow of
"name" buffer.
* config/tc-m68hc11.c (md_assemble): Likewise.
* config/tc-microblaze.c (md_assemble): Likewise.  Correct cast
of is_end_of_line index.

    * archive.cc: Formatting fixes: Remove whitespace between
        typename and following asterisk.  Remove whitespace between
        function name and opening parenthesis.
        * archive.h: Likewise.
        * arm.cc: Likewise.
        * attributes.cc: Likewise.
        * attributes.h: Likewise.
        * common.cc: Likewise.
        * copy-relocs.cc: Likewise.
        * dirsearch.h: Likewise.
        * dynobj.cc: Likewise.
        * ehframe.cc: Likewise.
        * ehframe.h: Likewise.
        * expression.cc: Likewise.
        * fileread.cc: Likewise.
        * fileread.h: Likewise.
        * gc.h: Likewise.
        * gold-threads.cc: Likewise.
        * gold.cc: Likewise.
        * i386.cc: Likewise.
        * icf.h: Likewise.
        * incremental-dump.cc: Likewise.
        * incremental.cc: Likewise.
        * layout.cc: Likewise.
        * layout.h: Likewise.
        * main.cc: Likewise.
        * merge.cc: Likewise.
        * merge.h: Likewise.
        * object.cc: Likewise.
        * object.h: Likewise.
        * options.cc: Likewise.
        * options.h: Likewise.
        * output.cc: Likewise.
        * output.h: Likewise.
        * plugin.cc: Likewise.
        * plugin.h: Likewise.
        * powerpc.cc: Likewise.
        * reloc.cc: Likewise.
        * script-c.h: Likewise.
        * script-sections.cc: Likewise.
        * script.cc: Likewise.
        * stringpool.cc: Likewise.
        * symtab.cc: Likewise.
        * symtab.h: Likewise.
        * target.cc: Likewise.
        * timer.cc: Likewise.
        * timer.h: Likewise.
        * version.cc: Likewise.
        * x86_64.cc: Likewise.

* NEWS: Mention ampersand in macro change.

* aout-arm.c (MY(bfd_reloc_type_lookup)): Use bfd_arch_bits_per_address.
* aout-ns32k.c (MY(bfd_reloc_type_lookup)): Likewise.
* aoutx.h (NAME(aout,reloc_type_lookup)): Likewise.
* coff-arm.c (coff_arm_reloc_type_lookup): Likewise.
* elf-hppa.h (elf_hppa_reloc_final_type): Likewise.
* reloc.c (bfd_default_reloc_type_lookup): Likewise.
* riscix.c (riscix_reloc_type_lookup): Likewise.

* elf.c (_bfd_elf_map_sections_to_segments): Don't load program
headers if any loaded section wraps the address space.  Simplify
~(m-1) to -m.  Use lma rather than vma when determining whether
note sections are adjacent.

* config/tc-arm.c (encode_arm_addr_mode_2): Fix comment.

* config/tc-arm.c (encode_arm_addr_mode_2): Fix
BAD_PC_ADDRESSING condition.

testsuite/
* gas/arm/ldst-pc.d: New test.
* gas/arm/ldst-pc.s: New test.
* gas/arm/sp-pc-validations-bad.s: `str r0,[pc,#4]' is valid.
* gas/arm/sp-pc-validations-bad.l: Adjust accordingly.

PR symtab/11942
* dwarf2read.c (dwarf2_per_objfile): New members debug_info_type_hash,
debug_types_type_hash.
(dwarf2_cu, dwarf2_per_cu_data): Delete member type_hash.
All uses updated.
(lookup_die_type): Renamed from tag_type_to_tag.  First look in
appropriate type_hash table.  All callers updated.
(allocate_signatured_type_table): Renamed from
allocate_signatured_type_hash_table.  All callers updated.
(create_signatured_type_table_from_index): Renamed from
create_signatured_type_hash_from_index.  All callers updated.
(read_die_type): Add comment.  Move actual reading to ...
(read_die_type_1): ... here.  New function.
(follow_die_ref_or_sig): Tweak comment.
(set_die_type): Rewrite to use appropriate choice of
debug_info_type_hash or debug_types_type_hash.
(get_die_type_at_offset): New function.
(get_die_type): Call it.

testsuite/
PR symtab/11942
* gdb.dwarf2/dw4-sig-types.cc: New file.
* gdb.dwarf2/dw4-sig-types.h: New file.
* gdb.dwarf2/dw4-sig-types-b.cc: New file.
* gdb.dwarf2/dw4-sig-types.exp: New file.

*** empty log message ***

daily update

* linux-low.h (linux_thread_area): Delete declaration.

gdb/
* python/py-type.c (typy_richcompare): Initialize worklist.

Fix typos in ld-x86-64/x86-64.exp.

2010-08-24  H.J. Lu  <hongjiu.lu@intel.com>

* ld-x86-64/x86-64.exp: Fix typos.

* arm-tdep.c: Include features/arm-with-m.c.
(arm_psr_thumb_bit): New.  Update all uses of CPSR_T to
call this function.
(arm_pc_is_thumb): Add a gdbarch argument.  Update all callers.
Check is_m after force-mode.
(arm_gdbarch_init): Check the binary before the target description.
Add check for M profile attribute.  If we have an M-profile device,
but no target register description, use arm-with-m.  Recognize the
new org.gnu.gdb.arm.m-profile feature and its xpsr register.
(_initialize_arm_tdep): Call initialize_tdesc_arm_with_m.
* arm-tdep.h (XPSR_T): Define.
(struct gdbarch_tdep): Add is_m member.
* features/arm-m-profile.xml, features/arm-with-m.c,
features/arm-with-m.xml: New files.

doc/
* gdb.texinfo (ARM Features): Document
org.gnu.gdb.arm.m-profile.

    PR 11899
        * layout.cc (segment_precedes): Sort segments by their physical
        addresses, if they have been set.

* sim/cris/asm/nonvcv32.ms: Neutralize changed &&-in-macro gas syntax.

daily update

*** empty log message ***

* dwarf2read.c (set_die_type): Fix typo in comment.

* dwarf2read.c (read_structure_type): Add comment.
(read_enumeration_type): Add comment.
(process_enumeration_scope): Move definition of some locals
closer to their use.
(read_namespace_type): Add comment.

gdb
PR python/11145:
* python/py-value.c: Include expression.h.
(valpy_do_cast): New function.
(valpy_cast): Use it.
(valpy_dynamic_cast): New function.
(valpy_reinterpret_cast): Likewise.
(value_object_methods): Add dynamic_cast, reinterpret_cast.
gdb/doc
PR python/11145:
* gdb.texinfo (Values From Inferior): Document dynamic_cast and
reinterpret_cast methods.
gdb/testsuite
PR python/11145:
* gdb.python/py-value.c (Base, Derived): New types.
(base): New global.
* gdb.python/py-value.exp (test_subscript_regression): Add
dynamic_cast test.

* archive.cc (Lib_group::add_symbols): Lock object before deleting its
symbols data.
(Lib_group::include_member): Unlock object after deleting its
symbols data.
* testsuite/start_lib_test_3.c: Remove all global symbols to trigger
the bug fixed here.

PR python/11391:
* python/py-value.c (valpy_nonzero): Don't throw error for other
Value types.

gdb
PR python/10676:
* python/py-type.c: Include bcache.h, vec.h.
(struct type_equality_entry): New.
(compare_strings): New function.
(check_types_equal): Likewise.
(check_types_worklist): Likewise.
(typy_richcompare): Likewise.
(type_object_type): Set tp_richcompare field.
gdb/testsuite
PR python/10676:
* gdb.python/py-type.exp (test_fields): Add tests for type
equality.

PR python/10953:
* python/py-type.c (typy_fields): Call check_typedef.
(typy_template_argument): Add TRY_CATCH.

gdb
PR python/11915:
* python/py-type.c (typy_array): New function.
(type_object_methods): Add "array".
gdb/doc
PR python/11915:
* gdb.texinfo (Types In Python): Document array method.
gdb/testsuite
PR python/11915:
* gdb.python/py-type.exp (test_fields): Add tests for array.

* gdb.dwarf2/dw2-double-set-die-type.S: DIE 0x51f does not
have any children. Create a new abbrev for it.
Add missing terminal sequence to .debug_abbrev.

Add --input-osabi and --output-osabi to elfedit.

binutils/

2010-08-23  H.J. Lu  <hongjiu.lu@intel.com>

* elfedit.c (input_elf_osabi): New.
(output_elf_osbi): Likewise.
(osabis): Likewise.
(elf_osabi): Likewise.
(update_elf_header): Support updating ELF OSABI.
(make_qualified_name): Break long line.
(command_line_switch): Add OPTION_INPUT_OSABI and
OPTION_OUTPUT_OSABI.
(options): Likewise.
(usage): Add --input-osabi and --output-osabi.
(main): Handle OPTION_INPUT_OSABI and OPTION_OUTPUT_OSABI.

* doc/binutils.texi: Document --input-osabi and --output-osabi
for elfedit.

binutils/testsuite/

2010-08-23  H.J. Lu  <hongjiu.lu@intel.com>

* binutils-all/elfedit-3.d: New.

* binutils-all/elfedit.exp: Run elfedit-3.

binutils/
* readelf.c (display_mips_gnu_attribute): Replace GCC options
with a textual description of a MIPS32r2 64-bit FPU.

ld/testsuite/
* ld-mips-elf/attr-gnu-4-04.d: Adjust Tag_GNU_MIPS_ABI_FP for
the new MIPS32r2 64-bit FPU description.
* ld-mips-elf/attr-gnu-4-40.d: Likewise.
* ld-mips-elf/attr-gnu-4-44.d: Likewise.

daily update

*** empty log message ***

Find one .eh_frame section for --eh-frame-hdr.

2010-08-22  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/11937
* emultempl/elf32.em (_after_open): Find one .eh_frame section
for --eh-frame-hdr.

Don't check relocation if input ELF object ID doesn't match output.

bfd/

2010-08-22  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/11933
* elflink.c (elf_link_add_object_symbols): Don't check relocation
if input ELF object ID doesn't match output.

ld/testsuite/

2010-08-22  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/11933
* ld-x86-64/mixed1a.s: New.
* ld-x86-64/mixed1b.s: Likewise.
* ld-x86-64/mixed2a.s: Likewise.
* ld-x86-64/mixed2b.s: Likewise.

* ld-x86-64/x86-64.exp: Text mixed x86_64 and i386 inputs.

Find an ELF input for --build-id and --eh-frame-hdr.

2010-08-22  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/11937
* emultempl/elf32.em (_after_open): Find an ELF input for
--build-id and --eh-frame-hdr.

*** empty log message ***

daily update

Fix misquoting in stdint.m4.

config/:
PR target/45084
* stdint.m4 (GCC_HEADER_STDINT): Use m4 quotes for arguments
of AC_MSG_ERROR.

libdecnumber/:
* configure: Regenerate.

bfd/:
* configure: Regenerate.

daily update

*** empty log message ***

merge from gcc

Revert the last checkin on sec_merge_emit.

2010-08-20  H.J. Lu  <hongjiu.lu@intel.com>

* merge.c (sec_merge_emit): Revert the last checkin.

* Makefile.am (eshlelf_fd.c): Correct the name of the tool
directory.
* Makefile.in: Regenerate.

* elf32-sh.c (sh_elf_relocate_section): Handle non-ELF output
BFD.

* ld-sh/sh.exp: Add missing newline.

* elf32-m68k.c (bfd_elf_m68k_set_target_options): Don't set GOT
options unless an m68k hash table has been found.

binutils/
* ChangeLog: Add a missing entry for 2010-04-20's commit.

ld/testsuite/
* ld-mips-elf/attr-gnu-4-01.d: Correct Tag_GNU_MIPS_ABI_FP.
* ld-mips-elf/attr-gnu-4-02.d: Likewise.
* ld-mips-elf/attr-gnu-4-10.d: Likewise.
* ld-mips-elf/attr-gnu-4-11.d: Likewise.
* ld-mips-elf/attr-gnu-4-15.d: Likewise.
* ld-mips-elf/attr-gnu-4-20.d: Likewise.
* ld-mips-elf/attr-gnu-4-22.d: Likewise.
* ld-mips-elf/attr-gnu-4-25.d: Likewise.

* doc/c-arm.texi (ARM Options): Document -mfpu=fp4-sp-d16.

* python/python.c: Include "serial.h".
(gdbpy_event_fds): Change type to `struct serial *' a array from
int array.
(gdbpy_run_events): Change parameters.  Use serial_readchar in
place of read.
(gdbpy_post_event): Use serial_write in place of write.
(gdbpy_initialize_events): Use serial_pipe instead of pipe, and
serial_async in place of add_file_handler.

* serial.h (gdb_pipe, serial_pipe): Declare.
* serial.c (serial_interface_lookup): Take a const char pointer.
(serial_fdopen): Rename to ...
(serial_fdopen_ops): ... this.  Add an OPS parameter and use it.
Call the OPS' fdopen function if there is one.
(serial_fdopen): Rewrite as wrapper to serial_fdopen_ops.
(serial_pipe): New.
(struct serial_ops) <fdopen>: New field.

* ser-mingw.c (free_pipe_state):
(free_pipe_state): Close output on non-pex pipes.
(pipe_windows_fdopen): New.
(gdb_pipe): New.
(_initialize_ser_windows): Register pipe_windows_fdopen.
* ser-go32.c (gdb_pipe): New.
* ser-pipe.c (pipe_close): Close file descriptor even if there's
no state pointer.
(pipe_ops): Delete.
(gdb_pipe): New.

PR symtab/11465:
  * dwarf2read.c (struct delayed_method_info): New struct.
(struct dwarf2_cu): Add vector method_list.
(scan_partial_symbols): Count methods for union, class, structure,
and interface types.
(add_to_method_list): New function.
(free_delayed_list): New function.
(compute_delayed_physnames): New function.
(process_full_comp_unit): Make a cleanup for the CU's delayed
physname list, compute the delayed physnames, and free the
the list.
(dwarf2_add_member_fn): For C++ and Java, delay the computation
of the physname until after the CU is read.

* dwarf2read.c (read_structure_type): Check if the current
DIE's type was already completed after dwarf2_full_name
was called.

PR symtab/11465:
* gdb.dwarf2/pr11465.exp: New test.
* gdb.dwarf2/pr11465.S: New file.
* gdb.dwarf2/dw2-double-set-die-type.S: New file.
* gdb.dwarf2/dw2-double-set-die-type.exp: New test.

    * emultempl/elf32.em (_after_open): Check for get_elf_backend_data
        returning NULL.

    * merge.c (sec_merge_emit): Do not zero pad sections that do not
        have an alignment.

* gold/archive.h (Add_lib_group_symbols): Add readsyms_blocker_, adjust
constructor, and set_blocker.
* gold/archive.cc (Add_lib_group_symbols::is_runnable): Also check
readsyms_blocker_.
* gold/readsyms.cc (Read_symbols::do_lib_group): Also pass
this->this_blocker_ to Add_lib_group_symbols::set_blocker.
* testsuite/Makefile.am (start_lib_test): New test case.
* testsuite/Makefile.in: Regenerate.
* testsuite/start_lib_test_main.c: New file.
* testsuite/start_lib_test_1.c: New file.
* testsuite/start_lib_test_2.c: New file.
* testsuite/start_lib_test_3.c: New file.

daily update

*** empty log message ***

* NEWS: Mention some additional changes.

* Makefile.in: Rebuild with automake 1.11.1.
* aclocal.m4: Likewise.
* testsuite/Makefile.in: Likewise.

PR 10893
* i386.cc (class Output_data_plt_i386): Update declarations.
Define Global_ifunc and Local_ifunc types.  Add global_ifuncs_ and
local_ifuncs_ fields.
(Target_i386::do_plt_section_for_global): New function.
(Target_i386::do_plt_section_for_local): New function.
(Output_data_plt_i386::Output_data_plt_i386): Add symtab
parameter; change all callers.  Initialize global_ifuncs_ and
local_ifuncs_.  If doing a static link define __rel_iplt_start and
__rel_iplt_end.
(Output_data_plt_i386::add_entry): Handle IFUNC symbols.
(Output_data_plt_i386::add_local_ifunc_entry): New function.
(Output_data_plt_i386::do_write): Fix GOT entries for IFUNC
symbols.
(Target_i386::make_plt_section): New function, broken out of
make_plt_entry.  Set sh_info field of .rel.plt to point to .plt.
(Target_i386::make_plt_entry): Call make_plt_section.
(Target_i386::make_local_ifunc_plt_entry): New function.
(Target_i386::Scan::reloc_needs_iplt_for_ifunc): New function.
(Target_i386::Scan::local): Handle IFUNC symbols.  Add
R_386_IRELATIVE to switch.
(Target_i386::Scan::global): Likewise.
(Target_i386::Relocate::relocate): Likewise.
(Target_i386::Relocatable_size_for_reloc): Add R_386_IRELATIVE to
switch.
* x86_64.cc (class Output_data_plt_x86_64): Update declarations.
(Target_x86_64::do_plt_section_for_global): New function.
(Target_x86_64::do_plt_section_for_local): New function.
(Output_data_plt_x86_64::Output_data_plt_x86_64): Add symtab
parameter; change all callers.  If doing a static link define
__rela_iplt_start and __rela_iplt_end.
(Output_data_plt_x86_64::add_entry): Handle IFUNC symbols.
(Output_data_plt_x86_64::add_local_ifunc_entry): New function.
(Target_x86_64::make_plt_section): Set sh_info field of .rel.plt
to point to .plt.
(Target_x86_64::make_local_ifunc_plt_entry): New function.
(Target_x86_64::Scan::check_non_pic): Add R_X86_64_IRELATIVE to
switch.
(Target_x86_64::Scan::reloc_needs_iplt_for_ifunc): New function.
(Target_x86_64::Scan::local): Handle IFUNC symbols.  Add
R_X86_64_IRELATIVE to switch.
(Target_x86_64::Scan::global): Likewise.
(Target_x86_64::Relocate::relocate): Likewise.
(Target_x86_64::Relocatable_size_for_reloc): Add R_X86_64_IRELATIVE to
switch.
* target.h (class Target): Add plt_section_for_global and
plt_section_for_local functions.  Add do_plt_section_for_global
and do_plt_section_for_local virtual functions.
* symtab.h (Symbol::needs_plt_entry): Handle IFUNC symbol.  Add
clarifying comments.
(Symbol::use_plt_offset): Handle IFUNC symbol.
* object.cc (Sized_relobj::Sized_relobj): Initialize
local_plt_offsets_.
(Sized_relobj::local_has_plt_offset): New function.
(Sized_relobj::local_plt_offset): New function.
(Sized_relobj::set_local_plt_offset): New function.
(Sized_relobj::do_count): Handle IFUNC symbol.
* object.h (class Symbol_value): Add is_ifunc_symbol_ field.  Take
a bit away from input_shndx_ field.  Add set_is_func_symbol and
is_ifunc_symbol functions.
(class Sized_relobj): Update declarations.  Remove Tls_got_entry
and Local_tls_got_offsets.  Define Local_plt_offsets.  Add
local_plt_offsets_ field.
(Sized_relobj::clear_local_symbols): Clear local_plt_offsets_.
* output.h (class Output_section_data): Add non-const
output_section function.
(class Output_data_got): Update declarations.
(class Output_data_got::Got_entry): Add use_plt_offset_ field.
Add use_plt_offset parameter to global and local constructors.
Change all callers.  Change local_sym_index_ field to 31 bits.
Change GSYM_CODE and CONSTANT_CODE accordingly.
* output.cc (Output_data_reloc_base::do_adjust_output_section): If
doing a static link don't set sh_link field.
(Output_data_got::Got_entry::write): Use PLT offset if
appropriate.
(Output_data_got::add_global_plt): New function.
(Output_data_got::add_local_plt): New function.
* target-reloc.h (relocate_section): Handle IFUNC symbol.
* defstd.cc (in_section): Remove entries for __rel_iplt_start,
__rel_iplt_end, __rela_iplt_start, and __rela_iplt_end.
* configure.ac: Set IFUNC automake conditional for glibc >= 2.11.
* testsuite/Makefile.am: Add a bunch of IFUNC tests, all within
IFUNC conditional.
* testsuite/ifunc-sel.h: New file.
* testsuite/ifuncmain1.c: New file.
* testsuite/ifuncmain1vis.c: New file.
* testsuite/ifuncmod1.c: New file.
* testsuite/ifuncdep2.c: New file.
* testsuite/ifuncmain2.c: New file.
* testsuite/ifuncmain3.c: New file.
* testsuite/ifuncmod3.c: New file.
* testsuite/ifuncmain4.c: New file.
* testsuite/ifuncmain5.c: New file.
* testsuite/ifuncmod5.c: New file.
* testsuite/ifuncmain6pie.c: New file.
* testsuite/ifuncmod6.c: New file.
* testsuite/ifuncmain7.c: New file.
* configure, testsuite/Makefile.in: Rebuild.

Fix a typo in comments.

2010-08-19  H.J. Lu  <hongjiu.lu@intel.com>

* config/tc-i386.c (VEX_check_operands): Fix a typo in comments.

* Makefile.in (install-python): Add DESTDIR.

* gdb.arch/vsx-regs.exp: Remove wrong comment about testing AltiVec
registers.  Update data sets with the new v2_double element in the
VSX register union.  Add vector_register3_vr data set for the AltiVec
registers.  Use gdb_test_no_output instead of send_gdb.

* gdb.python/py-shared.exp: New file, factored out from
python.exp.
* gdb.python/py-shared.c: New file.
* gdb.python/py-shared-sl.c: New file.
* gdb.python/python-1.c: New file.
* gdb.python/python-sl.c: Delete.
* gdb.python/python.c: Mention python-1.c.
* gdb.python/python.exp: Move shared library tests to
py-shared.exp.
* gdb.python/Makefile.in (EXECUTABLES): Add py-shared and python.
(MISCELLANEOUS): New.
(clean mostlyclean): Also remove $MISCELLANEOUS.

PR exp/11926
* parser-defs.h (parse_float, parse_c_float): Declare.
* parse.c (parse_float, parse_c_float): New function.
* c-exp.y (parse_number): Call parse_c_float.
* objc-exp.y (parse_number): Ditto.
* p-exp.y (parse_number): Ditto.  Use ANSI/ISO-style definition.
* jv-exp.y (parse_number): Call parse_float, fix suffix handling.

testsuite/
* gdb.base/printcmds.exp (test_float_accepted): New function.
Move existing float tests there.  Add tests for floats with suffixes.
(test_float_rejected): New function.
* gdb.java/jv-print.exp (test_float_accepted): New function.
(test_float_rejected): New function.
* gdb.objc/print.exp: New file.
* gdb.pascal/print.exp: New file.
* lib/objc.exp: New file.

Fix regression in -file-list-exec-source-files command.

See http://sourceware.org/ml/gdb/2010-07/msg00118.html for
a description of the problem. Namely, the file and fullname
fields are inverted in the output of the -file-list-exec-source-files
GDB/MI command:

    (gdb) interpreter-exec mi -file-list-exec-source-files
    ^done,files=[{file="/takamaka.a/brobecke/ex/list-exec-source-files/foo.c",fullname="foo.c"},{file="/takamaka.a/brobecke/ex/list-exec-source-files/foo.c",fullname="foo.c"},{file="",fullname="init.c"},{file="",fullname="../sysdeps/x86_64/elf/start.S"},{file="",fullname="../sysdeps/x86_64/elf/start.S"}]

It turns out to be a silly thinko: The map_symbol_filenames function
calls the psymtab version of map_symbol_filenames routine, and this
version called the callback function with filename and fullname
in the wrong order (fullname/filename instead of filename/fullname).

The routine description in symfile.h confirst that expected order for
the FUN callback parameters:

    /* Call a callback for every file defined in OBJFILE.  FUN is the
       callback.  It is passed the file's name, the file's full name,
       and the DATA passed to this function.  */
    void (*map_symbol_filenames) (struct objfile *objfile,
                                  void (*fun) (const char *, const char *,
                                               void *),
                                  void *data);

Fixing this error uncovered another location where the arguments
were reversed: maybe_add_partial_symtab_filename.  Once the first
error was fixed, the debugger would crash while attempting to do
completion, because it was given a NULL fullname instead of the
non-NULL filename.

gdb/ChangeLog:

        * psymtab.c (map_symbol_filenames_psymtab): Call FUN with
        the arguments in the correct order.
        * symtab.c (maybe_add_partial_symtab_filename): Declare
        the arguments in the correct order.

gdb/
* varobj.c (varobj_create): Replace variable old_fi with old_id,
initialize it by null_frame_id, wrap its usage by get_frame_id,
frame_id_p and frame_find_by_id.

* gdb.base/default.exp (info set): Avoid full buffer error and force
reading up to final gdb prompt.
(show): Likewise.

binutils/
* NEWS: Mention change in linker script expression evaluation.
ld/
* ld.texinfo (Expression Section): Detail expression evaluation.
(Builtin Functions <ADDR>): Correct.
(Builtin Functions <LOADADDR>): Don't mention LOADADDR normally
the same as ADDR.
(Builtin Functions <SEGMENT_START>): Typo fix.
* ldexp.c (new_number): New function.
(make_abs, exp_get_abs_int): Cope with NULL expld.result.section.
(fold_unary <'~', '!', '-'>): Don't make_abs.
(fold_binary): Simplify result section logic.  Return NULL section
for logical ops.
(fold_binary <SEGMENT_START>): Use new_rel_from_abs to set value to
a consistent result.
(fold_name <SIZEOF_HEADERS>): Return new_number, not new_abs.
(fold_name <DEFINED, SIZEOF, ALIGNOF, LENGTH, CONSTANT>): Likewise.
(fold_name <NAME>): No need to handle absolute symbols differently
from relative ones.
(fold_name <ORIGIN>): Don't return valid result when
lang_first_phase_enum.  Return new_rel_from_abs, not new_abs.
(exp_fold_tree_1 <etree_value>): Return new_number, not new_rel.
(exp_fold_tree_1): Ajust for NULL expld.result.section.  When assigning
a plain number to dot, assume the value is relative to expld.section.
Make terms not in an output section, absolute.
* ldlang.c (print_assignment): Fix style nit.
(lang_size_sections_1): Cope with NULL expld.result.section.
(lang_do_assignments_1): Likewise.
ld/testsuite/
* ld-scripts/memory.t: Remove ORIGIN fudge.

* incremental.cc
(Output_section_incremental_inputs::write_input_files): Add cast
to avoid signed/unsigned comparison warning.
(Output_section_incremental_inputs::write_info_blocks): Likewise.

*** empty log message ***

daily update

PR python/11900:
* configure: Rebuild.
* configure.ac: Add install-python to CONFIG_INSTALL.
* Makefile.in (install-python): New.

Check i.imm_operands VEXXDS.

2010-08-18  H.J. Lu  <hongjiu.lu@intel.com>

* config/tc-i386.c (build_modrm_byte): Check i.imm_operands
instead of VEXXDS.

* gdb_assert.h (gdb_assert_not_reached): New macro.
(gdb_assert_fail): Fix typo in comment.
* avr-tdep.c (avr_return_value): Use gdb_assert_not_reached instead of
gdb_assert (0).
* darwin-nat.c (darwin_check_new_threads): Ditto.
* dwarf2read.c (dwarf2_get_section_info): Ditto.
(munmap_section_buffer): Ditto.
* m32c-tdep.c (make_types): Ditto.
(m32c_decode_srcdest4, m32c_decode_sd23, m32c_frame_base): Ditto.
* macroexp.c (get_character_constant): Ditto.
(get_string_literal): Ditto.
* mep-tdep.c (mep_pseudo_cr_size): Ditto.
(mep_pseudo_cr_index, mep_register_type): Ditto.
(mep_pseudo_register_read, mep_pseudo_register_write): Ditto.
(mep_get_insn, mep_analyze_prologue): Ditto.
* objfiles.c (qsort_cmp): Ditto.
* prologue-value.c (pv_is_identical): Ditto.
* record.c (record_get_loc): Ditto.
* value.c (value_static_field): Ditto.
* xtensa-tdep.c (call0_track_op): Ditto.

gdb
PR symtab/11919:
* gdbtypes.c (lookup_struct_elt_type): Clean up error emission.
* parse.c (parse_field_expression): Use RETURN_MASK_ERROR.  Move
name-copying lower.  Document exception behavior.
* completer.c (expression_completer): Catch exceptions from
parse_field_expression.
gdb/testsuite
PR symtab/11919:
* gdb.base/completion.exp: Add test.

* common.cc (Sort_commons::operator()): Remove unnecessary code.

* gdb.texinfo (PowerPC Embedded): Mention support for the DVC register.

* gdb.base/call-ar-st.exp (set_lang_c): Delete, unused.
* gdb.base/callfuncs.exp (set_lang_c): Delete, set c language directly.
* gdb.base/printcmds.exp (set_lang_c): Delete, set c language directly.
Simplify, early exit if runto_main fails.
* gdb.fortran/exprs.exp: load_lib fortran.exp.
(set_lang_fortran): Moved to lib/fortran.exp.
* gdb.fortran/types.exp: load_lib fortran.exp.
(set_lang_fortran): Moved to lib/fortran.exp.
* gdb.java/jmisc.exp (set_lang_java): Moved to lib/java.exp.
* gdb.java/jprint.exp (set_lang_java): Ditto.
* gdb.java/jv-exp.exp: load_lib java.exp.
If set_lang_java fails, issue a warning instead of failure to be
consistent with other set_lang_foo uses.
(set_lang_java): Moved to lib/java.exp.
* gdb.java/jv-print.exp (set_lang_java): Moved to lib/java.exp.
If set_lang_java fails, issue a warning instead of failure to be
consistent with other set_lang_foo uses.
* gdb.pascal/types.exp: load_lib pascal.exp.
(set_lang_pascal): Moved to lib/pascal.exp.
* lib/fortran.exp: New file.
* lib/java.exp (set_lang_java): New function.
* lib/pascal.exp (set_lang_pascal): New function.

PR corefile/8210

* linux-thread-db.c (add_thread_db_info): Skip glibc/BZ5983
workaround on core files.
(try_thread_db_load_1): Don't try enabling thread event reporting
on core files.
(thread_db_load): Allow thread_db on core files.
(attach_thread): Don't check thread signals on core files, nor try
really attaching to the thread, nor enabling thread event event
reporting.
(thread_db_detach): Don't try disabing thread event reporting or
removing thread event breakpoints when debugging a core file.
(find_new_threads_callback): Don't try enabling thread event
reporting on core files.
(thread_db_find_new_threads_2): Don't look for a stopped lwp when
debugging a core file.
(thread_db_find_new_threads): Don't update thread
cores (processors) when debugging a core (dump).

2010-08-18  Pedro Alves  <pedro@codesourcery.com>

PR corefile/8210

bfd/
* bfd-in2.h: Regenerate.
* corefile.c (bfd_core_file_pid): New.

* targets.c (BFD_JUMP_TABLE_CORE): Add NAME##_core_file_pid.
(struct bfd_target) <_core_file_pid>: New.

* libbfd-in.h (_bfd_nocore_core_file_pid): Declare.
* libbfd.c (_bfd_nocore_core_file_pid): New.

* elf-bfd.h (bfd_elf32_core_file_pid, bfd_elf64_core_file_pid):
Declare.
* elfcode.h (elf_core_file_pid): New define.
* elfcore.h (elf_core_file_pid): New function.

* elf.c (elfcore_make_pid): Rewrite.
(elfcore_grok_prstatus): Only set core_pid if not set yet.
(elfcore_grok_prstatus) [!HAVE_PRSTATUS_T_PR_WHO]: Fallback to
getting the lwpid from prstat.pr_pid.

* elf64-x86-64.c (elf64_x86_64_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
(elf64_x86_64_grok_psinfo): Extract the the main process's PID,
and store it in elf_tdata's core_pid field.
* elf32-i386.c (elf_i386_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
(elf_i386_grok_psinfo): Extract the the main process's PID, and
store it in elf_tdata's core_pid field.

* elf32-am33lin.c (elf32_am33lin_grok_prstatus): Write the LWPID
to elf_tdata's core_lwpid instead of to core_pid.
* elf32-arm.c (elf32_arm_nabi_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-cris.c (cris_elf_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-frv.c (elf32_frv_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-hppa.c (elf32_hppa_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-mips.c (elf32_mips_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-ppc.c (ppc_elf_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-s390.c (elf_s390_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-score.c (s3_bfd_score_elf_grok_prstatus): Write the LWPID
to elf_tdata's core_lwpid instead of to core_pid.
* elf32-score7.c (s7_bfd_score_elf_grok_prstatus): Write the LWPID
to elf_tdata's core_lwpid instead of to core_pid.
* elf32-sh.c (elf32_shlin_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf32-xtensa.c (elf_xtensa_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf64-hppa.c (elf64_hppa_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf64-mips.c (elf64_mips_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elf64-ppc.c (ppc64_elf_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.
* elfn32-mips.c (elf32_mips_grok_prstatus): Write the LWPID to
elf_tdata's core_lwpid instead of to core_pid.

* plugin.c (bfd_plugin_core_file_pid): New function.
* aout-target.h (MY_core_file_pid): Define.
* aout-tic30.c (MY_core_file_pid, MY_core_file_p): New defines.
* coff-rs6000.c (coff_core_file_pid): New define.
(rs6000coff_vec, pmac_xcoff_vec): Use BFD_JUMP_TABLE_CORE.
* coff64-rs6000.c (coff_core_file_pid): New define.
(rs6000coff64_vec): Use BFD_JUMP_TABLE_CORE.
(xcoff64_core_file_pid): New define.
(aix5coff64_vec): Use BFD_JUMP_TABLE_CORE.
* mach-o-target.c (bfd_mach_o_core_file_pid): New define.
* aix386-core.c (aix386_core_file_pid): New define.
* hppabsd-core.c (hppabsd_core_core_file_pid): New define.
* hpux-core.c (hpux_core_core_file_pid): New define.
* irix-core.c (irix_core_core_file_pid): New define.
* lynx-core.c (lynx_core_file_pid): New define.
* osf-core.c (osf_core_core_file_pid): New define.
* ptrace-core.c (ptrace_unix_core_file_pid): New define.
* sco5-core.c (sco5_core_file_pid): New define.
* xcoff-target.h (coff_core_file_pid): New define.
* netbsd-core.c (netbsd_core_core_file_pid): New define.

gdb/
2010-08-18  Pedro Alves  <pedro@codesourcery.com>

PR corefile/8210

gdb/
* corelow.c (add_to_thread_list): Don't use
gdbarch_core_reg_section_encodes_pid.  Use bfd_core_file_pid.
(get_core_register_section): Don't use
gdbarch_core_reg_section_encodes_pid.

* gdbarch.sh (core_reg_section_encodes_pid): Delete.
* gdbarch.h, gdbarch.c: Regenerate.
* amd64-sol2-tdep.c (amd64_sol2_init_abi): Don't set
gdbarch_core_reg_section_encodes_pid.
* i386-sol2-tdep.c (i386_sol2_init_abi): Ditto.
* sparc-sol2-tdep.c (sparc32_sol2_init_abi): Ditto.
* sparc64-sol2-tdep.c (sparc64_sol2_init_abi): Ditto.

* macro.c (sub_actual): Add back ampersand suffix when no
substitution.
(macro_expand_body): Correct comment.

Forget to check in ChangeLog in my previous commit.
http://sourceware.org/ml/gdb-cvs/2010-08/msg00098.html

In order to fix a checkin mistake in dw2-ref-missing-frame-func.c:r1.2,
revert it back, and re-apply the correct patch again.

2010-08-18  Yao Qi  <yao@codesourcery.com

* gdb.dwarf2/dw2-ref-missing-frame-func.c: Add .p2align 4 for labels
func_nofb_start and func_loopfb_start, so that address of functions
is equal to these labels on Thumb.

revert back to previous revision

2010-08-18  Yao Qi  <yao@codesourcery.com

* gdb.dwarf2/dw2-ref-missing-frame-func.c: Add .p2align 4 for labels
func_nofb_start and func_loopfb_start, so that address of functions
is equal to these labels on Thumb.

2010-08-18  Yao Qi  <yao@codesourcery.com>

* gdb.gdb/selftest.exp (do_steps_and_nexts): Update test case
to reflect latest c source file.

2010-08-18  Yao Qi  <yao@codesourcery.com>

* MAINTAINERS : Add myself under Write After Approval.

Add missing files.

*** empty log message ***

daily update

* NEWS: Mention template parameter support.

gdb/testsuite/
* gdb.cp/infcall-dlopen.exp: New file.
* gdb.cp/infcall-dlopen.cc: New file.
* gdb.cp/infcall-dlopen-lib.cc: New file.

* breakpoint.c (decref_bp_location): Assert the reference count is
sane.

gdb/
2010-08-17  Pedro Alves  <pedro@codesourcery.com>

PR breakpoints/11371

* breakpoint.c (breakpoint_init_inferior): Decrement the
location's reference count instead of deleting right away.
(bpstat_free): Decrement the location's reference count.  Make
static.
(bpstat_copy): Increment the location's reference count.
(bpstat_find_breakpoint): Adjust.
(bpstat_num): Adjust.
(print_it_typical): Adjust.  Use the breakpoint pointer in the
bpstat instead of the location's owner.
(bpstat_alloc): Remove const qualifier from the 'bl' parameter.
Adjust to record the location's owner in the bpstat.
(watchpoint_check): Use the breakpoint pointer in the bpstat
instead of the location's owner.
(bpstat_check_breakpoint_conditions): Don't handle
bp_watchpoint_scope here.  Use the breakpoint pointer in the
bpstat instead of the location's owner.
(bpstat_stop_status): Defer inferior function calls to after
building the bpstat list.  Handle bp_watchpoint_scope here.  Use
the breakpoint pointer in the bpstat instead of the location's
owner.
(bpstat_what): Use the breakpoint pointer in the bpstat instead of
the location's owner.
(free_bp_location): Don't walk bpstats clearing locations.
(incref_bp_location): New.
(decref_bp_location): New.
(breakpoint_auto_delete): Use the breakpoint pointer in the bpstat
instead of the location's owner.
(update_global_location_list): Clear the location's owner, and
decrement the location's reference count instead of deleting it
right away.
(breakpoint_retire_moribund): Decrement the location's reference
count instead of deleting it right away.
(bpstat_remove_bp_location): Delete.
(bpstat_remove_breakpoint): New.
(bpstat_remove_bp_location_callback): Delete.
(bpstat_remove_breakpoint_callback): New.
(delete_breakpoint): Iterate over all threads' stop_bpstat's
clearing references to the breakpoint that is being deleted.

* breakpoint.h (struct bp_location) <refc>: New field.
<owner>: Update comments.
(bpstat_free): Delete declaration.
(struct bpstats): Change the type of the breakpoint_at field to
struct breakpoint point, from struct bp_location pointer.  Add new
field bp_location_at.

gdb/testsuite/
2010-08-17  Jan Kratochvil  <jan.kratochvil@redhat.com>
    Pedro Alves  <pedro@codesourcery.com>

PR breakpoints/11371

* gdb.base/watch-cond-infcall.exp: New file.
* gdb.base/watch-cond-infcall.c: New file.

Replace Eb with Mb on prefetch and prefetchw.

2010-08-17  H.J. Lu  <hongjiu.lu@intel.com>

* i386-dis.c (reg_table): Replace Eb with Mb on prefetch and
prefetchw.

Add "?" pseudo-flag to ELF .section/.pushsection directives.