gdb.base/ldbl_e308.exp: Do not allow "inf" in value output.

That was an attempt at handling the targets where sizeof(long double)
is less than 8, but the way it was implement allows the bug that
this testcase verifies to come back without being noticed.

gdb/testsuite/ChangeLog:

        * gdb.base/ldbl_e308.exp: Remove "inf" from possible expected
        output for "print inp" test.

* gdb.base/varargs.exp: PR gdb/12790 is now fixed.

* gdb.base/callfuncs.exp: PR gdb/12783 is now fixed.

Remove address check if HIDDEN_UNDEF_TEST is defined

* ld-elfvsb/main.c (main_visibility_checkcom): Remove address
check for visibility_def if HIDDEN_UNDEF_TEST is defined.
(main_visibility_checkweak): Remove address check for
visibility_func if HIDDEN_UNDEF_TEST is defined.

*** empty log message ***

* ld-scripts/empty-aligned.d: Revert last change.  Allow more
than one header, just not two PT_LOAD.
* ld-elf/stack.s: Use .long 0 rather than nops.

daily update

gas/testsuite/
* gas/i386/rex.s: Add test of REX prefix before fsave (i.e. fwait).
* gas/i386/rex.d: Update.

opcodes/
* i386-dis.c (ckprefix): When bailing out for fwait with prefixes,
set rex_used to rex.

Don't decompress the section in bfd_is_section_compressed

* compress.c (bfd_is_section_compressed): Don't decompress the
section.

Add comments in ravenscar_sparc_fetch_registers.

gdb/ChangeLog:

* ravenscar-sparc-thread.c (ravenscar_sparc_fetch_registers):
Add comments.

remote packet sent after Ravenscar inferior exited

When debugging a program using the Ravenscar profile, the debugger
sometimes tries to send the following packet to the remote after
the inferior exited.

     (gdb) c
     Continuing.
     [...]
     Sending packet: $vCont;c:1#13...Ack
     Packet received: W00
     Sending packet: $Hg1#e0...putpkt: write failed: Broken pipe.

As the inferior exited, the remote has already disconnected, and thus
the operation fails.

The reason why GDB sends the package is because the ravenscar-thread
module tries to updates the list of threads.  But this doesn't make
sense, since the program has exited.  This patch fixes it.

gdb/ChangeLog:

        * ravenscar-thread.c (ravenscar_wait): Only update the list
        of threads and inferior_ptid if the inferior is still alive.

[Ada] use lbasename when testing if file is part of Ada runtime.

We use a list of regular expressions to match a symtab filenames
against the names of the files in the Ada runtime.  These regular
expressions do assume that the filename is a basename, however.
So make sure to evaluate these regular expressions against
the symtab's filename.

Without this patch, we run into problems when the Ada runtime was built
using a project file (through gprbuild).

gdb/ChangeLog:

        * ada-lang.c (is_known_support_routine): Use lbasename when
        matching the symtab's filename against
        known_runtime_file_name_patterns.

[Ada] Allow assignment to wide string.

Given the following variable declaration...

   Www : Wide_String := "12345";

... this patch allows the following assignment to work:

   (gdb) set variable www := "qwert"

Without this patch, the debugger rejects the assignment because
the size of the array elements are different:

   (gdb) set www := "asdfg"
   Incompatible types in assignment

(on the lhs, we have an array of 2-bytes elements, and on the rhs,
we have a standard 1-byte string).

gdb/ChangeLog:

        * ada-lang.c (ada_same_array_size_p): New function.
        (ada_promote_array_of_integrals): New function.
        (coerce_for_assign): Add handling of arrays where the elements
        are integrals of a smaller size than the size of the target
        array element type.

gdb/testsuite/ChangeLog:

        * gdb.ada/set_wstr: New testcase.

off-by-one max exponent computation in convert_doublest_to_floatformat

Assuming the following variable definition:

    long double inp = 2.0;

On platforms where "long double" is a double precision IEEE flaoting
point, GDB currently behaves as follow:

    (gdb) set variable inp = 1.6e+308l
    (gdb) p inp
    $2 = inf   <<<<---- !!!!

Instead, the value of "inp" should be printed as:

    (gdb) p inp
    $1 = 1.6e+308

The problem is due to a small error in the comparison of the exponent
versus the maximum value this exponent can take, causing us to think
that the value was too big to fit. But it isn't.

gdb/ChangeLog:

        * doublest.c (convert_doublest_to_floatformat): Fix comparison
        against maximum exponent value.

gdb/testsuite/ChangeLog:

        * gdb.base/ldbl_e308.c, gdb.base/ldbl_e308.exp: New files.

[Ada] Skip unwind-seh.c when searching first "user" frame

On x86_64-windows with GCC 4.7 (using native SEH info), the debugger
behaves as follow:

    (gdb) catch exception unhandled
    Catchpoint 1: unhandled Ada exceptions
    (gdb) run
    Starting program: C:\[...]\b.exe

    Catchpoint 1, unhandled CONSTRAINT_ERROR at 0x000000000040cc57 in _GCC_specific_handler ([...]) at ../../../src/libgcc/unwind-seh.c:289
    [...]

This is after compiler the following code:

    procedure B is
    begin
       raise Constraint_Error;
    end B;

... using the following command:

    % gnatmake -g b

When hitting the exception catchpoint, it should have gone up the stack
all the way until finding the frame corresponding to procedure B.
But if stopped short because unwind-seh.c is compiled with debugging
information, and the debugger is also able to locate that source file.

To prevent this from happening, this patch adds unwind-seh.c to the list
of files that should be ignored, regardless of other factors.

gdb/ChangeLog:

        * ada-lang.h (ADA_KNOWN_RUNTIME_FILE_NAME_PATTERNS): Add entry for
        "unwind-seh.c".

[Ada] Pointers to unconstrained arrays inside variant record.

gdb/ChangeLog:

        * ada-lang.c (ada_template_to_fixed_record_type_1): Do not
        strip typedef layer when computing the fixed type's field type,
        only when computing its size.

gdb/testsuite/ChangeLog:

        * gdb.ada/unc_arr_ptr_in_var_rec: New testcase.

* gdb.base/callfuncs.exp: PR gdb/12783 is now fixed.

PR gdb/12783
* i386-tdep.c (i386_return_value): Handle complex double and long
double.

gdb/testsuite/
* gdb.base/callfuncs.exp (do_function_calls): Fix
setup_kfail_for_target for -m32 mode.

* configure.ac (FLAGS_FOR_TARGET,target=cygwin): Fix for building
against Mingw64 w32api.
* configure: Regenerate.

[Windows] run program with space in path to exe.

The following works...

    % gdb c:\path to exe\foo.exe
    (gdb) start

... unless a file or directory called "c:\path" or "c:\path to" exist.
This is what happens in the latter case:

    (gdb) start
    [...]
    Error creating process C:\path to exe\foo.exe (error 193).

This is because we are calling CreateProcess (et al) without specifying
the lpApplicationName, so Windows determines the name of the executable
using the second argument, which is the entire command line.  This
command line is a space-separated list of tokens, so the space in
the path to the executable which potentially creates an ambiguity.
The ambiguity is automatically resolved unless we're in the situation
above.

The solution, as suggested by the MSDN documentation for CreateProcess
is to quote the executable name.

gdb/ChangeLog:

        * windows-nat.c (windows_create_inferior) [!__CYGWIN__]:
        New local variable args_len.
        Quote the name of the executable when computing the command line.

Don't set bfdtest1 with findfile

* binutils-all/ar.exp (bfdtest1): Remove findfile.

Move disable_target_specific_optimizations to bfd_link_info

include/

* bfdlink.h (bfd_link_info): Add
disable_target_specific_optimizations.

ld/

* ld.h (command_line): Remove
disable_target_specific_optimizations.
(RELAXATION_DISABLED_BY_DEFAULT): Removed.
(RELAXATION_DISABLED_BY_USER): Likewise.
(RELAXATION_ENABLED): Likewise.
(DISABLE_RELAXATION): Likewise.
(ENABLE_RELAXATION): Likewise.

* ldmain.c (main): Updated.

* ldmain.h (RELAXATION_DISABLED_BY_DEFAULT): New macro.
(RELAXATION_DISABLED_BY_USER): Likewise.
(RELAXATION_ENABLED): Likewise.
(DISABLE_RELAXATION): Likewise.
(ENABLE_RELAXATION): Likewise.

gold/
* testsuite/Makefile.am (TEST_OBJCOPY): New macro.
* testsuite/Makefile.in: Regenerate.
* testsuite/dwp_test.h: New source file.
* testsuite/dwp_test_1.cc: New source file.
* testsuite/dwp_test_1.s: New source file.
* testsuite/dwp_test_1.sh: New source file.
* testsuite/dwp_test_1b.cc: New source file.
* testsuite/dwp_test_1b.s: New source file.
* testsuite/dwp_test_2.cc: New source file.
* testsuite/dwp_test_2.s: New source file.
* testsuite/dwp_test_2.sh: New source file.
* testsuite/dwp_test_main.cc: New source file.
* testsuite/dwp_test_main.s: New source file.

gold/
* dwp.h: New header file.
* dwp.cc: New source file.
* gold.h: Move shared declarations to system.h.
* system.h: New header file.
* Makefile.am: Add dwp.
* Makefile.in: Regenerate.

binutils/
* dwarf.c (do_debug_addr, do_debug_cu_index): New global flags.
(load_debug_info): Fix typo.
(cu_tu_indexes_read, shndx_pool, shndx_pool_size, shndx_pool_used):
New global variables.
(prealloc_cu_tu_list, add_shndx_to_cu_tu_entry, end_cu_tu_entry)
(process_cu_tu_index, load_cu_tu_indexes, find_cu_tu_set)
(display_cu_index): New functions.
(dwarf_select_sections_by_names): Add "debug_addr", "cu_index".
Sort entries alphabetically.
(dwarf_select_sections_all): Set do_debug_addr, do_debug_cu_index.
(debug_displays): Add .debug_cu_index, .debug_tu_index.
Clean up formatting.
* dwarf.h (dwarf_section_display_enum): Add dwp_cu_index,
dwp_tu_index.
(do_debug_addr, do_debug_cu_index): New global flags.
(find_cu_tu_set): New function declaration.
* objdump.c (usage): Add --dwarf=addr, --dwarf=cu_index.
* readelf.c (find_section_in_set): New function.
(usage): Add --debug-dump=addr, --debug_dump=cu_index.
(process_section_headers): Check do_debug_addr and do_debug_cu_index.
(section_subset): New global variable.
(load_debug_section): Restrict search to section subset.
(display_debug_section): Add section index as paramter.  Select subset
of sections when dumping a .dwp file.  Update caller.

*** empty log message ***

PR bootstrap/54820
* configure.ac (have_static_libs): Force 'no' for GCC version < 4.5.
* configure: Regenerate.

daily update

gold/
* dwarf_reader.cc (Dwarf_ranges_table::read_range_list): Call
Dwarf_info_reader::read_from_pointer.
(Dwarf_pubnames_table::read_header): Likewise.
(Dwarf_pubnames_table::next_name): Likewise.
(Dwarf_die::read_attributes): Likewise.
(Dwarf_die::skip_attributes): Likewise.
(Dwarf_info_reader::read_from_pointer): New function template.
* dwarf_reader.h (Dwarf_ranges_table): Add dwinfo_.
(Dwarf_pubnames_table): Likewise.
(Dwarf_info_reader::read_from_pointer): New function template.
* gdb-index.cc (Gdb_index_info_reader): Adjust call to
Dwarf_pubnames_table ctor.

gold/
* dwarf_reader.cc (Dwarf_info_reader::do_parse): Use stored
abbrev_shndx.
* dwarf_reader.h (Dwarf_info_reader::Dwarf_info_reader): Initialize
abbrev_shndx_.
(Dwarf_info_reader::set_abbrev_shndx): New method.
(Dwarf_info_reader::abbrev_shndx_): New data member.

gold/
* dwarf_reader.cc (make_elf_reloc_mapper): Check size and endianness
from object, not parameters.
(Dwarf_info_reader::parse): Likewise.
* object.h (Relobj::elfsize, Relobj::is_big_endian): New methods.
(Relobj::do_elfsize, Relobj::do_is_big_endian): New methods.
(Sized_relobj::do_elfsize, Sized_relobj::do_is_big_endian): New
methods.

gold/
* fileread.cc (Input_file::Input_file): New constructor.
* fileread.h (class Input_file): Add new constructor.

* gdb.base/callfuncs.exp: PR gdb/12796, gdb/12798 and gdb/12800
are now fixed.

PR gdb/12796
PR gdb/12798
PR gdb/12800
* amd64-tdep.h (enum amd64_regnum): Add AMD64_ST1_REGNUM and
AMD64_FTAG_REGNUM.
* amd64-tdep.c (amd64_classify): Classify complex types.
(amd64_return_value): Handle the COMPLEX_X87 class.

binutils
* readelf.c (get_note_type): Handle NT_SIGINFO, NT_FILE.
(print_core_note): New function.
(process_note): Call it.
include/common/elf
* common.h (NT_SIGINFO, NT_FILE): New defines.

gdb/testsuite:

* gdb.base/info-os.exp: Resume the inferior until it exits.

Change target-wide-charset to UTF-16 on ppc-aix.

On ppc-aix, type wchar_t is 2 bytes long, so override the default
target-wide-charset (UTF-32) with UTF-16. This allows us to print
wide characters correctly.

gdb/ChangeLog:

        * rs6000-aix-tdep.c (rs6000_aix_auto_wide_charset): New function.
        (rs6000_aix_init_osabi): Set auto_wide_charset gdbarch method.

Set default target-wide-charset setting to "UTF-16" for x86_64-windows.

Type wchar_t is only 2 bytes long on x86_64-windows.

gdb/ChangeLog:

        * amd64-windows-tdep.c (amd64_windows_auto_wide_charset): New
        function.
        (amd64_windows_init_abi): Set auto_wide_charset gdbarch method
        to amd64_windows_auto_wide_charset.

bfd/
* bfd-in.h (bfd_elf_stack_segment_size): Declare.
* bfd-in2.h: Rebuilt.
* elfxx-target.h (elf_backend_stack_align): New.
(elfNN_bed): Add it.
* elf-bfd.h (struct elf_backend_data): Add stack_align field.
* elf.c (bfd_elf_map_sections_to_segments): Pay attention to
stack_align and stacksize for PT_GNU_STACK segment.
(assign_file_positions_for_non_load_sections): Set p_memsz for
PT_GNU_STACK segment.
(copy_elf_program_header): Copy PT_GNU_STACK size.
* elflink.c (bfd_elf_stack_segment_size): New function, taken from
uclinux backends.
(bfd_elf_size_dynamic_sections): Determine
PT_GNU_STACK requirements after calling backend.  Pay attention to
stacksize.
* elf32-bfin.c (elf32_bfinfdpic_always_size_sections): Call
bfd_elf_stack_segment_size.
(elf32_bfinfdpic_modify_program_headers): Delete.
(elf32_bfingfdpic_copy_private_bfd_data): Don't copy PT_GNU_STACK
here.
(elf_backend_stack_align): Override.
(elf_backend_modify_program_headers): Don't override.
* elf32-frv.c (frvfdpic_always_size_sections): Call
bfd_elf_stack_segment_size.
(elf32_frvfdpic_modify_program_headers): Delete.
(elf32_frvfdpic_copy_private_bfd_data): Don't copy PT_GNU_STACK
here.
(elf_backend_stack_align): Override.
(elf_backend_modify_program_headers): Don't override.
* elf32-lm32.c (lm32_elf_always_size_sections): Leave
PT_GNU_STACK creation to underlying elf support.  Check
__stacksize here for backwards compatibility, and set it if
needed.
(lm32_elf_modify_segment_map): Delete.
(lm32_elf_modify_program_headers): Delete.
(elf_backend_stack_align): Override.
(elf_backend_modify_segment_map): Don't override.
(elf_backend_modify_program_headers): Don't override.
* elf32-sh.c (sh_elf_always_size_sections): Call
bfd_elf_stack_segment_size.
(sh_elf_modify_program_headers): Delete.
(sh_elf_copy_private_data): Don't copy PT_GNU_STACK
here.
(elf_backend_stack_align): Override.
(elf_backend_modify_program_headers): Don't override.
* elf32-tic6x.c (elf32_tic6x_always_size_sections): Call
bfd_elf_stack_segment_size.
(elf32_tic6x_modify_program_headers): Delete.
(elf32_tic6x_copy_private_data): Delete.
(elf_backend_stack_align): Override.
(bfd_elf32_bfd_copy_private_bfd_data): Don't override.
(elf_backend_modify_program_headers): Don't override.

include/
* bfdlink.h (struct bfd_link_info): Add stacksize option.

ld/
* ld.texinfo (stack-size): New option.
* emultempl/elf32.em: Add stack-size option.

ld/testsuite/
* ld-elf/binutils.exp: Add -z stack-size=0.
* ld-elf/elf.exp: Add stack-exec and stack-size tests.
* ld-elf/orphan-region.d: Add stack-size option. Remove xfail.
* ld-elf/stack-exec.rd: New.
* ld-elf/stack-size.rd: New.
* ld-elf/stack.s: New.
* ld-scripts/empty-aligned.d: Add stack-size option.
* ld-sh/fdpic-stack-set.d: New.
* ld-tic6x/shlib-1.rd: Remove __stacksize symbol.
* ld-tic6x/shlib-1b.rd: Likewise.
* ld-tic6x/shlib-1r.rd: Likewise.
* ld-tic6x/shlib-1rb.rd: Likewise.
* ld-tic6x/shlib-app-1.rd: Likewise.
* ld-tic6x/shlib-app-1b.rd: Likewise.
* ld-tic6x/shlib-app-1r.rd: Likewise.
* ld-tic6x/shlib-app-1rb.rd: Likewise.
* ld-tic6x/shlib-noindex.rd: Likewise.
* ld-tic6x/static-app-1.rd: Likewise.
* ld-tic6x/static-app-1b.rd: Likewise.
* ld-tic6x/static-app-1r.rd: Likewise.
* ld-tic6x/static-app-1rb.rd: Likewise.

gdb/
* event-top.c (mark_async_signal_handler_wrapper): Remove.
* event-top.h: Remove its declaration.
(async_request_quit): Call mark_async_signal_handler instead of
mark_async_signal_handler_wrapper.
(async_do_nothing, async_disconnect): Likewise.
(async_stop_sig): Likewise.
* remote.c (handle_remote_sigint): Likewise.
(handle_remote_sigint_twice): Likewise.

gdb/
* event-top.c (sigint_token, sighup_token): Replace 'void *'
with 'static struct async_signal_handler *'.
(sighup_token, sigquit_token, sigstp_token): Likewise.

*** empty log message ***

daily update

opcodes/
* ppc-opc.c (powerpc_opcodes) <vcfpsxws>: Fix opcode spelling.

gas/testsuite/
* gas/ppc/altivec.s <vcfpsxws>: Fix opcode spelling.

Ali Anwar  <ali_anwar@codesourcery.com>

* gdbarch.sh (function_list): Use 'pstring' when printing a
variable which could return NULL.
* gdbarch.c: Regenerate.

include/
PR ld/14426
* bfdlink.h (bfd_link_info): Add ignore_hash.
ld/
PR ld/14426
* ldlex.h (option_values): Add OPTION_IGNORE_UNRESOLVED_SYMBOL.
* lexsup.c (parse_args): Likewise.
(ld_options): Describe --ignore-unresolved-symbol.
* ldmain.h (add_ignoresym): Declare.
* ldmain.c (add_ignoresym): New function, extracted from..
(undefined_symbol): ..here.  Return if the symbol is in ignore_hash.
(constructor_callback): Don't use global link_info here.
(reloc_overflow): Likewise.

* as.c (dump_statistics): Compute data size as the delta between
current sbrk(0) and start_sbrk.
* (main): Set start_sbrk to sbrk(0) on entry.

PR bootstrap/54820
* Makefile.tpl (STAGE1_FLAGS_TO_PASS): New variable.
(all-[+prefix+][+module+]): Pass stage1_args to sub-makes.
(all-stage[+id+]-[+prefix+][+module+]): Likewise, if prev is false.
(clean-stage[+id+]-[+prefix+][+module+]): Likewise, if prev is false.
(host_modules): Set stage1_args to STAGE1_FLAGS_TO_PASS.
* Makefile.in: Regenerate.
* configure.ac (have_static_libs): New variable and associated check.
(stage1-ldflags): Move to after stage1_libs and set to -static-libstdc++
-static-libgcc if stage1_libs is empty and have_static_libs is yes.
* configure: Regenerate.

* plugin.c (plugin_load_plugins): Warning fix.

* cache.c (cache_bmmap): Don't use void* arithmetic.

*** empty log message ***

daily update

Update binutils-all/dw2-1.W

* binutils-all/dw2-1.W: Updated.

Handle x32 in dump_dwarf

* objdump.c (dump_dwarf): Handle bfd_mach_x64_32 and
bfd_mach_x64_32_intel_syntax.

bfd/
* compress.c (bfd_cache_section_contents): New function.
* bfd-in2.h: Regenerate.
binutils/
* objdump.c (load_specific_debug_section): Use
bfd_cache_section_contents.

Correct last entry

* linker.c (_bfd_generic_link_output_symbols): Handle a
no-longer-global symbol entered as a BFD_PLUGIN.

*** empty log message ***

daily update

Add an objdump test on compress debug sections

* binutils-all/compress.exp: Add an objdump test.
* binutils-all/dw2-1.W: New file.

* compress.c: Reinstate 2012-10-19 change.
(bfd_get_full_section_contents): Don't free unless we alloc.
Use proper decompress size.  Delete some vars, rename others.

*** empty log message ***

daily update

ppc-aix osabi sniffer: Turn test of bfd flavour into assertion

Due to the way this function is registers, we know that given bfd's
flavour should always be bfd_target_xcoff_flavour, thus making
the former test always true, which means that this function should
always return GDB_OSABI_AIX, and never return GDB_OSABI_UNKNOWN.

This patch also fixes a typo detected by Tom Tromey that caused
the test itself to be completely ineffective.

gdb/ChangeLog (by Tom Tromey and Joel Brobecker):

        * rs6000-aix-tdep.c (rs6000_aix_osabi_sniffer): Replace
        inneffective if condition by gdb assertion.  Add function
        description comment.

Document exp_descriptor.op_name should never return NULL.

This documents a constaint that struct exp_descriptor's "op_name"
method implementation should obey.  This might not have been part
of the initial design, but is currently true of all instantiations,
and already assumed by the current users.

gdb/ChangeLog:

        * parser-defs.h (struct exp_descriptor): Document constraint
        on return value for "op_name" callbacks.

* compress.c: Revert previous patch.

* compress.c (bfd_get_full_section_contents): Don't cache
decompressed contents.

Correct my ChangeLog entry

    * gdb.base/ctxobj.exp: Skip if skip_shlib_tests returns true.
        * gdb.base/print-file-var.exp: Likewise.
        * gdb.base/type-opaque.exp: Likewise.

*** empty log message ***

daily update

Add missed file on my last commit

* tracepoint.c (print_one_static_tracepoint_marker): Constify.
* symtab.c (iterate_over_some_symtabs): Constify.
* source.h (symtab_to_fullname): Return 'const char *'.
* source.c (symtab_to_fullname): Return 'const char *'.
* python/py-symtab.c (stpy_fullname): Constify.
* cli/cli-cmds.c (edit_command): Constify.
* breakpoint.c (print_breakpoint_location)
(update_static_tracepoint): Constify.

* breakpoint.c (compare_breakpoints): Fix comparison.

Remove some extraneous changes in readline/terminal.c

This patch removes some unnecessary differences between the official
version of readline, and GDB's copy of it.

readline/ChangeLog.gdb:

        * terminal.c: Remove duplicate includes of windows.h and
        wincon.h.
        (_rl_get_screen_size): Remove redundant code for MinGW getting
        the console size from the Windows API.

ChangeLog bfd

2012-10-18  Kai Tietz  <ktietz@redhat.com>

PR binutils/14067
* coff-i386.c (bfd_target): Add section flag SEC_EXCLUDE.
Allow BFD_COMPRESS and BFD_DECOMPRESS flags.
* coff-x86_64.c: Likewise.
* coffcode.h (DOT_ZDEBUG): New define.
(sec_to_styp_flags): Check for .zdebug.
(styp_to_sec_flags): Likewise.
* coffgen.c (make_a_section): Handle .debug_* section
compression/decompression flags.
* cofflink.c (mark_relocs): Ignore relocations
for a section, which isn't marked as used.
(_bfd_coff_link_input_bfd): Add support of compressed
debug sections.
* compress.c (decompress_contents): Loop as long
as there is input available and there is room for
output.
* bfd/pe-arm.c: Add .zdebug_ partial match entry.
* pe-i386.c: Likewise.
* pe-x86_64.c: Likewise.
* peXXigen.c (_bfd_XXi_swap_aouthdr_out): Don't clear all
data-directories as this might destroy content.

        * coff-i386.c (_bfd_generic_find_nearest_line_discriminator):
        define as coff_find_nearest_line_discriminator.
        * libcoff-in.h (coff_find_nearest_line_discriminator): New
        * libcoff.h: Regenerated.
        * coff-x86_64.c: Likewise.
        * coffgen.c (coff_find_nearest_line_discriminator): New function.
        prototype.

ChangeLog binutils

2012-10-18  Kai Tietz  <ktietz@redhat.com>

* objdump.c (dump_bfd): Call dump headers after
call of slurp_symtab.

ChangeLog ld

2012-10-18  Kai Tietz  <ktietz@redhat.com>

PR binutils/14067
* NEWS: Menition new feature.
* scripttempl/pep.sc: Add zdebug sections.
* scripttempl/pe.sc: L

    * objcopy.c (is_strip_section_1): Don't strip
        .reloc section by default.

    * config/obj-coff.c: Add include of struc-symbol.h header.
        (coff_frob_symbol): Check that function-aux entries are generated for
        defined symbols only.

* tic54x-dis.c (print_instruction): Don't use K&R style.
(print_parallel_instruction, sprint_dual_address)
(sprint_indirect_address, sprint_direct_address, sprint_mmr)
(sprint_cc2, sprint_condition): Likewise.

* valprint.c (generic_emit_char, generic_printstr): Pass size of
gdb_wchar_t to convert_between_encodings.

    * aarch64-asm.c (aarch64_ins_ldst_reglist): Initialize
        value with a default.
        (do_special_encoding): Likewise.
        (aarch64_ins_ldst_elemlist): Pre-initialize QSsize, and opcodeh2
        variables with default.
        * arc-dis.c (write_comments_): Don't use strncat due
        size of state->commentBuffer pointer isn't predictable.

    * doc/as.texinfo: Add missing documentation about section flag
        exclude.

PR gold/14727
* object.cc (Relobj::is_section_name_included): Also match
.sdata personality section.

* target-reloc.h (class Default_comdat_behavior): New, package up..
(get_comdat_behaviour): ..this.
(relocate_section): Add Relocate_comdat_behavior template arg,
adjust code to suit.
* arm.cc (Target_arm::relocate_section): Adjust to suit.
(Target_arm::scan_reloc_section): Likewise.
* i386.cc (Target_i386::relocate_section): Likewise.
* sparc.cc (Target_sparc::relocate_section): Likewise.
* tilegx.cc (Target_tilegx::relocate_section): Likewise.
* x86_64.cc (Target_x86_64::relocate_section): Likewise.
* powerpc.cc (class Relocate_comdat_behavior): New.
(Target_powerpc::relocate_section): Don't zap opd relocs.  Supply
gold::relocate_section with new template arg.

*** empty log message ***

daily update

* powerpc.cc (Target_powerpc::Scan::local, global): Always emit
dynamic relocs for GOT_TPREL got entries, without symbol if
resolving locally.
(Target_powerpc::do_gc_add_reference): Don't add for dynamic objects.
(Target_powerpc::scan_relocs): Define _GLOBAL_OFFSET_TABLE_ early.
(Target_powerpc::Relocate:relocate): REL32 reloc may be unaligned.

PR gold/14726
* gold.cc (queue_middle_tasks): Call gc_mark_symbol on _init and _fini.

gdb:
* breakpoint.c (invalidate_bp_value_on_memory_change): Add one
more parameter 'inferior'.
* corefile.c (write_memory_with_notification): Caller update.

* mi/mi-cmd-var.c: Include "mi-main.h".
(mi_cmd_var_assign): Set mi_suppress_notification.data_write_memory
to 1 and restore it later.
* mi/mi-cmds.c (mi_cmd mi_cmds): Update for "data-write-memory"
and "data-write-memory-bytes.
* mi/mi-interp.c: Include objfiles.h.
(mi_interpreter_init): Call observer_attach_memory_changed.
(mi_memory_changed): New.
* mi/mi-main.h (struct mi_suppress_notification) <memory>:
New field.

* NEWS: Mention new MI notification "memory-changed".

gdb/doc:

* observer.texi (GDB Observers): Update observer
'memory_changed'.
* gdb.texinfo (GDB/MI Async Records): Document for
"memory-changed" notification.

gdb/testsuite:

* gdb.mi/mi-memory-changed.exp: New.

*** empty log message ***

daily update

Add missing ChangeLog entry for previous commit:

2012-10-16  Joel Brobecker  <brobecker@adacore.com>

       * coff64-rs6000.c (rs6000coff64_vec): Set _close_and_cleanup
       field to _bfd_archive_close_and_cleanup.
       (aix5coff64_vec): Likewise.

dangling pointer in coff64-rs6000 archive cache.

bfd/ChangeLog:

        * coff64-rs6000.c (rs6000coff64_vec): Set _close_and_cleanup
        field to _bfd_archive_close_and_cleanup.
        (aix5coff64_vec): Likewise.

Dangling bfd pointer in archive cache.

This dandling pointer eventually leads to a crash when trying to run
on ppc-aix a program using threading...

bfd/ChangeLog:

        * coff-rs6000.c (rs6000coff_vec): Set _close_and_cleanup
        field to _bfd_archive_close_and_cleanup.
        (pmac_xcoff_vec): Likewise.

2012-10-16  Sriraman Tallam  <tmsriram@google.com>

* layout.cc (Layout::include_section): Keep sections marked
SHF_EXCLUDE when doing relocatable links.

Fixed AArch64 TLS relocation resolution.

bfd/

* elf64-aarch64.c (elf64_aarch64_tls_howto_table): Fix shift value for
R_AARCH64_TLSIE_LD_GOTTPREL_PREL19, R_AARCH64_TLSLE_MOVW_TPREL_G2,
R_AARCH64_TLSLE_MOVW_TPREL_G1, R_AARCH64_TLSLE_MOVW_TPREL_G1_NC,
R_AARCH64_TLSLE_ADD_TPREL_HI12.
(elf64_aarch64_tlsdesc_howto_table): Fix shift value for
R_AARCH64_TLSDESC_LD64_PREL19 and R_AARCH64_TLSDESC_OFF_G1.
(elf64_aarch64_final_link_relocate): Add signed_addend when resolving
AARCH64_TLSLE_*_TPREL_* relocations.

ld/testsuite/

* ld-aarch64/tlsle-symbol-offset.s: New file.
* ld-aarch64/tlsle-symbol-offset.d: New file.
* ld-aarch64/aarch64-elf.exp: Add tlsle-symbol-offset test.

gdb/testsuite/
* gdb.reverse/solib-precsave.exp: Skip if skip_shlib_test
returns true.
Call gdb_load_shlibs.
* gdb.reverse/solib-reverse.exp: Likewise.

gdb/
* dwarf2loc.c (dwarf_expr_reg_to_entry_parameter): Remove comment.

* elf32-xtensa.c (free_section_cache): Renamed from
clear_section_cache.  Don't zero cache.
(section_cache_section): Remove ineffectual zero of cache.
Call init_section_cache instead.