2013-08-12  Jan-Benedict Glaw  <jbglaw@lug-owl.de>

* configure.ac: Sync with GCC repo.
* Makefile.def: Ditto.
* configure: Regenerate.
* Makefile.in: Ditto.

PR binutils/15818
* objdump.c (disassemble_section): Return early if nothing from
this section needs to be disassembled.

Because of:

<http://sourceware.org/ml/gdb-patches/2013-08/msg00289.html>

I have chosen to revert the patch applied to the AVR target-dependent code.
Therefore, this patch does just that.  It is better to keep the tree
buildable than to keep this patch in, for now.

2013-08-12  Sergio Durigan Junior  <sergiodj@redhat.com>

Revert implementation of gdbarch_gdb_signal_{to,from}_target for
AVR.
* avr-tdep.c: Remove include of "linux-tdep.h".  Remove enum with
different signals between the generic Linux kernel implementation
and AVR's.
(avr_linux_gdb_signal_from_target): Delete.
(avr_linux_gdb_signal_to_target): Delete.
(avr_gdbarch_init): Don't set gdbarch_gdb_signal_{to,from}_target.

daily update

daily update

daily update

* LRS: Remove file, describes a long-abandoned live-range
splitting extension to stabs.

* lib/future.exp (gdb_find_ldd): New proc.
* lib/prelink-support.exp (build_executable_own_libs): Call it.
Make "/usr/sbin/" in prelink path optional.

(create_addrmap_from_index): Use hex_string instead of pulongest in previous
patch.

* dwarf2read.c (create_addrmap_from_index): Ignore bad address table
entries.

This patch implements the new gdbarch method gdbarch_gdb_signal_to_target.
It will be used when one wants to convert between the internal GDB signal
representation (enum gdb_signal) and the target's representation.

The idea of this patch came from a chat between Pedro and I on IRC, plus
the discussion of my patches to add the new $_exitsignal convenience
variable:

<http://sourceware.org/ml/gdb-patches/2013-06/msg00452.html>
<http://sourceware.org/ml/gdb-patches/2013-06/msg00352.html>

What I did was to investigate, on the Linux kernel, which targets shared
the signal numbers definition with the generic definition, present at
<include/uapi/asm-generic/signal.h>.  For the record, I used linux-3.10-rc7
as the main source of information, always looking at
<arch/<ARCH_NAME>/include/uapi/asm/signal.h>.  For SIGRTMAX (which defaults
to _NSIG in most cases), I had to look at different signal-related
files, but most of them (except MIPS) were defined to 64 anyway.

Then, with all the differences in hand, I implemented the bits on each
target.

2013-08-09  Sergio Durigan Junior  <sergiodj@redhat.com>

* linux-tdep.c: Define enum with generic signal numbers.
(linux_gdb_signal_from_target): New function.
(linux_gdb_signal_to_target): Likewise.
(linux_init_abi): Set gdbarch_gdb_signal_{to,from}_target
methods to the functions above.
* linux-tdep.h (linux_gdb_signal_from_target): New prototype.
(linux_gdb_signal_to_target): Likewise.
* alpha-linux-tdep.c: Define new enum with signals different
from generic Linux kernel.
(alpha_linux_gdb_signal_from_target): New function.
(alpha_linux_gdb_signal_to_target): Likewise.
(alpha_linux_init_abi): Set gdbarch_gdb_signal_{to,from}_target
with the functions mentioned above.
* avr-tdep.c: Define enum with differences between Linux kernel
and AVR signals.
(avr_linux_gdb_signal_from_target): New function.
(avr_linux_gdb_signal_to_target): Likewise.
(avr_gdbarch_init): Set gdbarch_gdb_signal_{to,from}_target to
the functions mentioned above.
* sparc-linux-tdep.c: Define enum with differences between SPARC
and generic Linux kernel signal numbers.
(sparc32_linux_gdb_signal_from_target): New function.
(sparc32_linux_gdb_signal_to_target): Likewise.
(sparc32_linux_init_abi): Set gdbarch_gdb_signal_{to,from}_target
to the functions defined above.
* xtensa-linux-tdep.c: Define enum with differences between
Xtensa and Linux kernel generic signals.
(xtensa_linux_gdb_signal_from_target): New function.
(xtensa_linux_gdb_signal_to_target): Likewise.
(xtensa_linux_init_abi): Set gdbarch_gdb_signal_to_target
to the functions defined above.
* mips-linux-tdep.c: Define enum with differences between
signals in MIPS and Linux kernel generic ones.
(mips_gdb_signal_to_target): New function.
(mips_gdb_signal_from_target): Redefine to use new enum, handle
only different signals from the Linux kernel generic.
(mips_linux_init_abi): Set gdbarch_gdb_signal_{to,from}_target
the functions defined above.
* mips-linux-tdep.h (enum mips_signals): Remove.

Remove unnecessary XMALLOC definitions.

XMALLOC is defined in defs.h.

Tested by building with --enable-targets=all.

gdb/
2013-08-09  Pedro Alves  <palves@redhat.com>

* avr-tdep.c (XMALLOC): Delete macro.
* cli/cli-dump.c (XMALLOC): Delete macro.

Make functions in cli-dump.c static and remove cli-dump.h.

I noticed the functions declared in cli-dump.h aren't used anywhere
outside cli-dump.c.

The original patch that introduced cli-dump.c didn't include this header:

 http://sourceware.org/ml/gdb-patches/2002-03/msg00518.html

But for some reason that I couldn't find from reading the archives around
that patch's discussion, cli-dump.h was introduced in the final checkin,
at:

 http://sourceware.org/ml/gdb-patches/2002-03/msg00596.html

There seems to be no point in keeping this around nowadays.

gdb/
2013-08-09  Pedro Alves  <palves@redhat.com>

* cli/cli-dump.c: Don't include cli/cli-dump.h.
(scan_expression_with_cleanup, scan_filename_with_cleanup)
(fopen_with_cleanup, add_dump_command): Make static.
* cli/cli-dump.h: Delete file.
* Makefile.in (HFILES_NO_SRCDIR): Remove reference to
cli/cli-dump.h.

tsave: Make tilde-expanded filename visible.

Before:

  (gdb) tsave ~/a/b
  Unable to open file '~/a/b' for saving trace data (No such file or directory)

After:

  (gdb) tsave ~/a/b
  Unable to open file '/home/pedro/a/b' for saving trace data (No such file or directory)

Tested on x86_64 Fedora 17.

gdb/
2013-08-09  Pedro Alves  <palves@redhat.com>

* tracepoint.c (tfile_start): Show tilde-expanded filename in
error message.

save breakpoints: Make tilde-expanded filename visible.

Most commands in GDB show the tilde-expanded filename in user visible
output.  This makes "save breakpoints" behave the same.

Before:

  (gdb) save breakpoints ~/a/b
  Unable to open file '~/a/b' for saving (No such file or directory)

After:

  (gdb) save breakpoints ~/a/b
  Unable to open file '/home/pedro/a/b' for saving (No such file or directory)

Tested on x86_64 Fedora 17.

gdb/
2013-08-09  Pedro Alves  <palves@redhat.com>

        * breakpoint.c (save_breakpoints): Show tilde-expanded filename in
error message.

gcore: Make tilde-expanded filename visible.

Most commands in GDB show the tilde-expanded filename in user visible
output.  This makes gcore behave the same.

Before:

  (gdb) generate-core-file ~/a/b
  Failed to open '~/a/b' for output.
  (gdb) generate-core-file ~/core
  Saved corefile ~/core

After:

  (gdb) generate-core-file ~/a/b
  Failed to open '/home/pedro/a/b' for output.
  (gdb) generate-core-file ~/core
  Saved corefile /home/pedro/core

Tested on x86_64 Fedora 17.

gdb/
2013-08-09  Pedro Alves  <palves@redhat.com>

* gcore.c (create_gcore_bfd): Don't use tilde_expand here.
(gcore_command): Use tilde_expand here, and when showing the
filename to the user, show the expanded version.

* elf32-rl78.c (rl78_elf_merge_private_bfd_data): Complain if G10
flag bits do not match.
(rl78_elf_print_private_bfd_data): Describe G10 flag.

* readelf.c (get_machine_flags): Handle RL78 G10 flag.

* config/tc-rl78.c (elf_flags): New variable.
(enum options): Add OPTION_G10.
(md_longopts): Add mg10.
(md_parse_option): Parse -mg10.
(rl78_elf_final_processing): New function.
* config/tc-rl78.c (tc_final_processing): Define.
* doc/c-rl78.texi: Document -mg10 option.

* rl78.c (E_FLAG_RL78_G10): Define.

* lib/ld-lib.exp (check_shared_lib_support): Note that the RL78
does not support shared library generation.

gdb/

* stack.c (read_frame_arg): Set 'entryval_error' to NULL if
'entryval' is set.

gdb/testsuite/

* gdb.trace/collection.exp (gdb_collect_args_test): Set
"only" and "both" to 'print entry-values' before selecting
trace frame.

daily update

gcore: expand tilde in filename.

Before this patch, this fails:

 (gdb) generate-core-file ~/core
 Failed to open '~/core' for output.

After the patch:

 (gdb) generate-core-file ~/core
 Saved corefile ~/core

gdb/
2013-08-08  Azat Khuzhin  <a3at.mail@gmail.com>  (tiny change)

* gcore.c (create_gcore_bfd): Use tilde_expand.

gdb/
* frame.h (read_frame_local): Declare.
* mi/mi-cmd-stack.c (list_args_or_locals): Call
read_frame_local.
* stack.c (read_frame_local): New.

gdb/testsuite/

* gdb.trace/mi-trace-unavailable.exp: Don't set
"print entry-values" to "no".
(test_trace_unavailable): Set various values to
"print entry-values" to test that the output of
'-stack-list-locals' is not affected, and then set
set "print entry-values" to "no".

gdb/
* mi/mi-cmd-stack.c: Update comments to function
list_args_or_locals.

* gdb-if.c (hw_breakpoints): Remove.
(sim_store_register): Add an assert() to make sure PC is in range.
Delete code which referenced hw_breakpoints[].

daily update

* gdb.python/py-value-cc.cc: Renamed from py-value.cc.
* gdb.python/py-value-cc.exp: Update.
* gdb.python/py-value.exp: Use different names for .o files for
C and C++.  Only perform C++ tests if !skip_cplus_tests.

fix PR symtab/15028

This fixes some derivation.exp regressions with "dwz -m".

The bug here is that the imported PU is given language_minimal.
However, it ought to be C++.

The "pretend language" machinery exists to solve this problem, but it
wasn't handled in process_psymtab_comp_unit.  So, this patch adds it
there.

Built and regtested, both normally and using "dwz -m", on x86-64
Fedora 18.

PR symtab/15028:
* dwarf2read.c (struct process_psymtab_comp_unit_data): New.
(process_psymtab_comp_unit_reader): Use it.
(process_psymtab_comp_unit): Update.  Add "pretend_language"
argument.
(dwarf2_build_psymtabs_hard): Update.
(scan_partial_symbols): Pass CU's language to
process_psymtab_comp_unit.

remove unused qf method

After the previous patch in the series, nothing uses the "quick"
method find_symbol_file.

This patch removes it.

Tested by rebuilding.

* dwarf2read.c (dw2_get_primary_filename_reader): Remove.
(dwarf2_gdb_index_functions): Update.
* psymtab.c (find_symbol_file_from_partial): Remove.
(psym_functions): Update.
* symfile.h (struct quick_symbol_functions) <find_symbol_file>:
Remove.

use language of the main symbol

With "dwz -m", "main" appears in both the PU and the importing CU when
running anon-struct.exp.  However, the PU does not have a file name.
So, find_main_filename returns the empty string, making
deduce_language_from_filename return language_unknown.

This patch fixes this problem by changing gdb to use the ordinary
symbol-lookup functions to find "main"'s symbol.  Then, it examines the
symbol's language.

I think this is cleaner than the current approach.  For one thing it
avoids trying to guess the language based on the source file name,
instead deferring to the presumably more reliable debuginfo.

Another possible fix would have been to change how the file name is
found via the "qf" methods.  However, I think the approach given is
preferable for the reason outlined above.

This required a minor test suite change, as now a symtab is expanded
during the search for "main".

Built and regtested (both ways) on x86-64 Fedora 18.

* symfile.c (set_initial_language): Look up "main" symbol
and use its language.
* symtab.c (find_main_filename): Remove.
* symtab.h (find_main_filename): Remove.

* gdb.base/maint.exp: Allow zero symtabs to be expanded.

fix recursively_compute_inclusions and add dwz test case

Doug pointed out a while ago that in the final dwz -m patch, nothing
ever set symtab::user.

This patch fixes this oversight and adds a test case showing why it is
important.

Built and regtested (both ways) on x86-64 Fedora 18.
The new test unconditionally tests the partial unit machinery, which I
think is an added plus.

* dwarf2read.c (recursively_compute_inclusions): Add
"immediate_parent" argument.  Set symtab's "user" field
if not set.
(compute_symtab_includes): Update.

* gdb.dwarf2/dwz.exp: New file.

also filter label symbols

The bug here is that, with dwz -m, a function (and a label) appear in
both a PU and a CU when running cplabel.exp.  So, a breakpoint gets
two locations:

    (gdb) break foo::bar:to_the_top
    Breakpoint 2 at 0x400503: foo::bar:to_the_top. (2 locations)

What is especially wacky is that both locations are at the same place:

    (gdb) info b
    Num     Type           Disp Enb Address            What
    1       breakpoint     keep y   <MULTIPLE>
    1.1                         y     0x000000000040051c foo::bar:get_out_of_here
    1.2                         y     0x000000000040051c foo::bar:get_out_of_here

This happens due to the weird way we run "dwz -m".
It's unclear to me that this would ever happen for real code.

While I think this borders on "diminishing returns" territory, the fix
is pretty straightforward: use the existing address-filtering function
in linespec to also filter when looking at labels.

Built and regtested (both ways) on x86-64 Fedora 18.

* linespec.c (convert_linespec_to_sals): Use maybe_add_address
when adding label symbols.

gold/
Revert support for v2 DWP files:

2013-03-01  Cary Coutant  <ccoutant@google.com>

    Add dwp support for v2 DWARF package file format.
    * dwarf_reader.cc (Dwarf_info_reader::visit_type_unit): Add
    tu_length parameter.  Adjust all callers.
    * dwarf_reader.h (Dwarf_info_reader::visit_type_unit): Likewise.
    * dwp.cc: Include dwarf.h.
    (Section_bounds): New struct type.
    (Unit_set): New struct type.
    (Dwo_file::Dwo_file): Initialize new data member.
    (Dwo_file::read_compunit_index, Dwo_file::read_typeunit_index):
    Combine and rename to...
    (Dwo_file::read_unit_index): ...this.
    (Dwo_file::sized_read_compunit_index)
    (Dwo_file::sized_read_typeunit_index): Combine and rename to...
    (Dwo_file::sized_read_unit_index): ...this.
    (Dwo_file::copy_section): Remove section_name, is_str_offsets
    parameters; add section_id parameter.
    (Dwo_file::add_cu_set, Dwo_file::add_tu_set): Combine and rename to...
    (Dwo_file::add_unit_set): ...this.
    (Dwo_file::shndx_map_): Remove.
    (Dwo_file::sect_offsets_): New data member.
    (Dwp_output_file::Dwp_output_file): Initialize new data members.
    (Dwp_output_file::add_section): Rename to...
    (Dwp_output_file::add_contribution): ...this.
    (Dwp_output_file::add_cu_set): Combine parameters into a struct.
    (Dwp_output_file::add_tu_set): Likewise.
    (Dwp_output_file::Contribution): New type.
    (Dwp_output_file::Section::contributions): New data member.
    (Dwp_output_file::Cu_or_tu_set): Remove.
    (Dwp_output_file::Section::Section): New ctor.
    (Dwp_output_file::Dwp_index::Shndx_pool): Remove.
    (Dwp_output_file::Dwp_index::Section_table): New type.
    (Dwp_output_file::Dwp_index::Dwp_index): Initialize new data members.
    (Dwp_output_file::Dwp_index::enter_set): Change type of "set"
    parameter.
    (Dwp_output_file::Dwp_index::shndx_pool): Remove.
    (Dwp_output_file::Dwp_index::shndx_pool_end): Remove.
    (Dwp_output_file::Dwp_index::section_table): New member function.
    (Dwp_output_file::Dwp_index::section_table_end): New member function.
    (Dwp_output_file::Dwp_index::shndx_pool_size): Remove.
    (Dwp_output_file::Dwp_index::section_table_rows): New member function.
    (Dwp_output_file::Dwp_index::section_table_cols): New member function.
    (Dwp_output_file::Dwp_index::shndx_pool_): Remove.
    (Dwp_output_file::Dwp_index::section_table_): New data member.
    (Dwp_output_file::Dwp_index::section_mask_): New data member.
    (Dwp_output_file::add_output_section): New member function.
    (Dwp_output_file::write_new_section): New member function.
    (Dwp_output_file::write_contributions): New member function.
    (Dwp_output_file::section_id_map_): New data member.
    (class Dwo_id_info_reader): Remove.
    (class Unit_reader): New class.
    (get_dwarf_section_name): New function.
    (Dwo_file::read_executable): Adjust initializations of class data.
    (Dwo_file::read): Add support for v2 package file format.
    (Dwo_file::read_unit_index): Likewise.
    (Dwo_file::sized_read_unit_index): Likewise.
    (Dwo_file::copy_section): Likewise.
    (Dwo_file::add_unit_set): Likewise.
    (Dwp_output_file::add_output_section): Likewise.
    (Dwp_output_file::add_contribution): Likewise.
    (Dwp_output_file::Dwp_index::find_or_add): Use row index to check
    for empty slot.
    (Dwp_output_file::Dwp_index::enter_set): Add support for v2 package
    file format.
    (Dwp_output_file::Dwp_index::grow): Use row index to check for empty
    slot.
    (Dwp_output_file::initialize): Remove unused function.
    (Dwp_output_file::finalize): Add support for v2 package file format.
    (Dwp_output_file::write_index): Likewise.
    * gdb-index.cc (Gdb_index_info_reader::visit_type_unit): Adjust
    function prototype.

merge from gcc

2013-08-07  Raunaq Bathija  <raunaq12@in.ibm.com>
    Ulrich Weigand  <uweigand@de.ibm.com>

* configure.tgt (powerpc64-*-aix*): Match powerpc64 running aix.
* configure.host (powerpc64-*-aix*): Likewise.

2013-08-07  Raunaq Bathija  <raunaq12@in.ibm.com>
    Ulrich Weigand  <uweigand@de.ibm.com>

* gdb_ptrace.h: Use ptrace64 instead of ptrace if HAVE_PTRACE64
is defined.
* rs6000-nat.c: Check for __ld_info64_ if compiling 64 BIT gdb.
(rs6000_ptrace32): Call ptrace64 instead of ptrace if present.
(rs6000_ptrace64): Call ptace64 instead of ptracex if present.
* configure.ac: Check for ptrace64.
* configure, config.in: Regenerate.

2013-08-07  Raunaq Bathija  <raunaq12@in.ibm.com>
    Ulrich Weigand  <uweigand@de.ibm.com>

* aixthread.c: Call ptrace64 instead of ptracex if defined.
Call ptrace64 instead of ptrace if defined.
Add macro addr_ptr to take care of ptrace address argument.
(pdc_read_regs): Likewise.
(pdc_write_regs): Likewise.
(aix_thread_resume): Likewise.
(fetch_regs_kernel_thread): Likewise.
(store_regs_kernel_thread): Likewise.

2013-08-07  Anton Blanchard  <anton@samba.org>

* MAINTAINERS: Add myself to Write After Approval.

daily update

opcodes/
2013-08-06  Jürgen Urban  <JuergenUrban@gmx.de>

* mips-opc.c (mips_builtin_opcodes): Add a suffixless version of
VCLIPW.

gas/
2013-08-06  Jürgen Urban  <JuergenUrban@gmx.de>

* config/tc-mips.c (match_vu0_suffix_operand): Allow single-channel
suffixes to be elided too.
(mips_lookup_insn): Don't reject INSN2_VU0_CHANNEL_SUFFIX here.
(mips_ip): Assume .xyzw if no VU0 suffix is specified.  Allow +N
to be omitted too.

gas/testsuite/
2013-08-06  Jürgen Urban  <JuergenUrban@gmx.de>

* gas/mips/r5900-error-vu0.s, gas/mips/r5900-error-vu0.l,
gas/mips/r5900-full-vu0.s, gas/mips/r5900-full-vu0.d: Allow
single-channel suffixes to be elided.

daily update

bfd/
2013-08-05  John Tytgat  <john@bass-software.com>

* po/BLD-POTFILES.in: Regenerate.
* po/SRC-POTFILES.in: Likewise.

gas/
2013-08-05  John Tytgat  <john@bass-software.com>

* po/POTFILES.in: Regenerate.

gprof/
2013-08-05  John Tytgat  <john@bass-software.com>

* po/POTFILES.in: Regenerate.

* sparc-opc.c (v9andleon): Fix thinko.

more add_target removals

This removes a few more erroneous calls to add_target.  These calls
end up installing the target in a user-visible way; but these targets
are all auto-activated and, I think, should never be explicitly
requested.

I have no way to test these.

* aix-thread.c (_initialize_aix_thread): Use
complete_target_initialization.
* bsd-uthread.c (_initialize_bsd_uthread): Use
complete_target_initialization.
* dec-thread.c (_initialize_dec_thread): Use
complete_target_initialization.
* ravenscar-thread.c (_initialize_ravenscar): Use
complete_target_initialization.
* sol-thread.c (_initialize_sol_thread): Use
complete_target_initialization.
* spu-multiarch.c (_initialize_spu_multiarch): Use
complete_target_initialization.

gas/
* config/tc-sparc.c (sparc_arch_types): Add leon.
(sparc_arch): Move sparc4 around and add leon.
(sparc_target_format): Document -Aleon.
* doc/c-sparc.texi: Likewise.
include/
* opcode/sparc.h (sparc_opcode_arch_val): Add SPARC_OPCODE_ARCH_LEON.
opcodes/
* sparc-dis.c (compute_arch_mask): Set SPARC_OPCODE_ARCH_LEON bit for
bfd_mach_sparc.
* sparc-opc.c (MASK_LEON): Define.
(v6, v6notlet, v7, v8, v6notv9): Add MASK_LEON.
(letandleon): New macro.
(v9andleon): Likewise.
(sparc_opc): Add leon.
(umac): Enable for letandleon.
(smac): Likewise.
(casa): Enable for v9andleon.
(cas): Likewise.
(casl): Likewise.

Remove:
2013-07-10  Jack Howarth  <howarth@bromo.med.uc.edu>

       PR target/57792
       * configure.ac: Use --with-sysroot=\"`xcrun --show-sdk-path`\" on darwin13 and later.
       * configure: Regenerated.

remove msymbol_objfile

This is another patch in my ongoing series to "split" objfile to share
more read-only data across inferiors.  See

    http://sourceware.org/gdb/wiki/ObjfileSplitting

When symbols are finally shared, there will be no back-link from the
symbol to its containing objfile, because there may be more than one
such objfile.  So, all such back-links must be removed.

One hidden back-link is the msymbol_objfile function.  Since
(eventually) a symbol may appear in more than one objfile, trying to
look up the objfile given just a symbol cannot work.

This patch removes msymbol_objfile in favor of using a bound minimal
symbol.  It introduces a new function to make this conversion simpler
in some spots.

The bonus of this patch is that using msymbol_objfile is slower than
simply looking up the owning objfile in the first place.

Built and regtested on x86-64 Fedora 18.

* ada-exp.y (write_var_or_type): Use bound_minimal_symbol.
* ada-lang.c (ada_lookup_simple_minsym): Return
bound_minimal_symbol.
* ada-lang.h (ada_lookup_simple_minsym): Update.
* c-exp.y (variable): Use lookup_bound_minimal_symbol.
* f-exp.y (variable): Use lookup_bound_minimal_symbol.
* go-exp.y (variable): Use lookup_bound_minimal_symbol.
* jv-exp.y (push_expression_name): Use lookup_bound_minimal_symbol.
* m2-exp.y (variable): Use lookup_bound_minimal_symbol.
* minsyms.c (msymbol_objfile): Remove.
(lookup_minimal_symbol_internal): New function, from
lookup_minimal_symbol.
(lookup_minimal_symbol): Rewrite using
lookup_minimal_symbol_internal.
(lookup_bound_minimal_symbol): New function.
* minsyms.h (msymbol_objfile): Remove.
(lookup_bound_minimal_symbol): Declare.
* p-exp.y (variable): Use lookup_bound_minimal_symbol.
* parse.c (write_exp_msymbol): Change parameter to a
bound_minimal_symbol.
(write_dollar_variable): Use lookup_bound_minimal_symbol.
* parser-defs.h (write_exp_msymbol): Update.
* printcmd.c (address_info): Use lookup_bound_minimal_symbol.
* symfile.c (simple_read_overlay_table): Use
lookup_bound_minimal_symbol.
* symtab.c (skip_prologue_sal): Don't use msymbol_objfile.
(search_symbols): Likewise.
(print_msymbol_info): Take a bound_minimal_symbol argument.
(symtab_symbol_info, rbreak_command): Update.
* symtab.h (struct symbol_search) <msymbol>: Change type
to bound_minimal_symbol.
* valops.c (find_function_in_inferior): Use
lookup_bound_minimal_symbol.
* value.c (value_fn_field): Use lookup_bound_minimal_symbol.

doc: gdbserver allows for subsequent and not for multiple simultaneous connections.

doc/

2013-08-05  Sanimir Agovic  <sanimir.agovic@intel.com>

* gdb.texinfo (TCP port allocation lifecycle): Gdbserver by default
allows for subsequent and not for additional (multiple simultaneous)
        connections.

gdb/
Code cleanup.
* remote.c (cleanup_sigint_signal_handler): Rename the declaration
to ...
(async_cleanup_sigint_signal_handler): ... this.
(initialize_sigint_signal_handler): Remove declaration.
(handle_remote_sigint): Rename the declaration to ...
(async_handle_remote_sigint): ... this.
(handle_remote_sigint_twice): Rename the declaration to ...
(async_handle_remote_sigint_twice): ... this.
(async_remote_interrupt, async_remote_interrupt_twice)
(remote_interrupt): Remove the declarations.
(remote_interrupt_twice): Rename the declaration ...
(sync_remote_interrupt_twice): ... this.
(sigint_remote_twice_token): Rename the variable to ...
(async_sigint_remote_twice_token): ... this.
(sigint_remote_token): Rename the variable to ...
(async_sigint_remote_token): ... this.
(initialize_sigint_signal_handler): Rename the function to ...
(async_initialize_sigint_signal_handler): ... this.  Update the name
inside.
(handle_remote_sigint): Rename the function to ...
(async_handle_remote_sigint): ... this.  Update the names inside.
(handle_remote_sigint_twice): Rename the function to ...
(async_handle_remote_sigint_twice): ... this.  Update the names inside.
(cleanup_sigint_signal_handler): Rename the function to ...
(async_cleanup_sigint_signal_handler): ... this.
(remote_interrupt): Rename the function to ...
(sync_remote_interrupt): this.  Update the names inside.
(remote_interrupt_twice): Rename the function to ...
(sync_remote_interrupt_twice): this.  Update the names inside.
(remote_terminal_inferior, remote_terminal_ours, remote_wait_as)
(_initialize_remote): Update the names inside.

gas/
* config/tc-mips.c (mips_lookup_insn): Make length and opend signed.

daily update

include/opcode/
2013-08-04  Jürgen Urban  <JuergenUrban@gmx.de>
    Richard Sandiford  <rdsandiford@googlemail.com>

* mips.h: Document new VU0 operand characters.
(OP_VU0_SUFFIX, OP_VU0_MATCH_SUFFIX): New mips_operand_types.
(OP_REG_VF, OP_REG_VI, OP_REG_R5900_I, OP_REG_R5900_Q, OP_REG_R5900_R)
(OP_REG_R5900_ACC): New mips_reg_operand_types.
(INSN2_VU0_CHANNEL_SUFFIX): New macro.
(mips_vu0_channel_mask): Declare.

opcodes/
2013-08-04  Jürgen Urban  <JuergenUrban@gmx.de>
    Richard Sandiford  <rdsandiford@googlemail.com>

* mips-dis.c (print_reg): Handle OP_REG_VI, OP_REG_VF, OP_REG_R5900_I,
OP_REG_R5900_Q, OP_REG_R5900_R and OP_REG_R5900_ACC.
(print_vu0_channel): New function.
(print_insn_arg): Handle OP_VU0_SUFFIX and OP_VU0_MATCH_SUFFIX.
(print_insn_args): Handle '#'.
(print_insn_mips): Handle INSN2_VU0_CHANNEL_SUFFIX.
* mips-opc.c (mips_vu0_channel_mask): New constant.
(decode_mips_operand): Handle new VU0 operand types.
(VU0, VU0CH): New macros.
(mips_builtin_opcodes): Add VU0 opcodes.  Use "+7" rather than "E"
for LQC2 and SQC2.  Use "+9" rather than "G" for EE CFC2 and CTC2.
Use "+6" rather than "G" for QMFC2 and QMTC2.

gas/
2013-08-04  Jürgen Urban  <JuergenUrban@gmx.de>
    Richard Sandiford  <rdsandiford@googlemail.com>

* config/tc-mips.c (MAX_OPERANDS): Bump to 6.
(RWARN): Bump to 0x8000000.
(RTYPE_VI, RTYPE_VF, RTYPE_R5900_I, RTYPE_R5900_Q, RTYPE_R5900_R)
(RTYPE_R5900_ACC): New register types.
(RTYPE_MASK): Include them.
(R5900_I_NAMES, R5900_Q_NAMES, R5900_R_NAMES, R5900_ACC_NAMES): New
macros.
(reg_names): Include them.
(mips_parse_register_1): New function, split out from...
(mips_parse_register): ...here.  Add a channels_ptr parameter.
Look for VU0 channel suffixes when nonnull.
(reg_lookup): Update the call to mips_parse_register.
(mips_parse_vu0_channels): New function.
(OT_CHANNELS, OT_DOUBLE_CHAR): New mips_operand_token_types.
(mips_operand_token): Add a "channels" field to the union.
Extend the comment above "ch" to OT_DOUBLE_CHAR.
(mips_parse_base_start): Match -- and ++.  Handle channel suffixes.
(mips_parse_argument_token): Handle channel suffixes here too.
(validate_mips_insn): Handle INSN2_VU0_CHANNEL_SUFFIX.
Ignore OP_VU0_MATCH_SUFFIX when calculating the used bits.
Handle '#' formats.
(md_begin): Register $vfN and $vfI registers.
(operand_reg_mask): Handle OP_VU0_SUFFIX and OP_VU0_MATCH_SUFFIX.
(convert_reg_type): Handle OP_REG_VI, OP_REG_VF, OP_REG_R5900_I,
OP_REG_R5900_Q, OP_REG_R5900_R and OP_REG_R5900_ACC.
(match_vu0_suffix_operand): New function.
(match_operand): Handle OP_VU0_SUFFIX and OP_VU0_MATCH_SUFFIX.
(macro): Use "+7" rather than "E" for LDQ2 and STQ2.
(mips_lookup_insn): New function.
(mips_ip): Use it.  Allow "+K" operands to be elided at the end
of an instruction.  Handle '#' sequences.

gas/testsuite/
2013-08-04  Jürgen Urban  <JuergenUrban@gmx.de>

* gas/mips/r5900-vu0.d: Expect $vfN and $viN instead of numeric
coprocessor registers.
* gas/mips/r5900-all-vu0.s, gas/mips/r5900-all-vu0.d,
gas/mips/r5900-full-vu0.s, gas/mips/r5900-full-vu0.d,
gas/mips/r5900-error-vu0.s, gas/mips/r5900-error-vu0.l: New tests.
* gas/mips/mips.exp: Run them.

daily update

gas/
* config/tc-mips.c (macro, mips16_macro): Create an array of operand
values and use it instead of sreg, treg, xreg, etc.

include/opcode/
* mips.h (mips_pcrel_operand): Inherit from mips_int_operand.
(mips_int_operand_min, mips_int_operand_max): New functions.
(mips_decode_pcrel_operand): Use mips_decode_int_operand.

opcodes/
* mips-formats.h (PCREL): Reorder parameters and update the definition
to match new mips_pcrel_operand layout.
(JUMP, JALX, BRANCH): Update accordingly.
* mips16-opc.c (decode_mips16_operand): Likewise.

gas/
* config/tc-mips.c (match_int_operand): Use mips_int_operand_min
and mips_int_operand_max.
(mips16_immed_operand, mips16_immed_operands, MIPS16_NUM_IMMED):
Delete.
(mips16_immed_operand, mips16_immed_in_range_p): New functions.
(mips16_immed, mips16_extended_frag): Use them.  Use mips_int_operand
instead of mips16_immed_operand.

gas/
* config/tc-mips.c (mips16_macro): Don't use move_register.
(mips16_ip): Allow macros to use 'p'.

gas/testsuite/
* gas/mips/mips16-macro.s, gas/mips/mips16-macro.d: New test.
* gas/mips/mips.exp: Run it.

daily update

fix PR symtab/15719

This patch fixes PR symtab/15719.

The bug is that "watch -location" crashes on a certain expression.

The problem is that fetch_subexp_value is catching an exception.
For ordinary watchpoints this is ok; but for location watchpoints,
it is better for the exception to propagate.

Built and regtested on x86-64 Fedora 18.
New test case included.

PR symtab/15719:
* breakpoint.c (update_watchpoint, watchpoint_check)
(watch_command_1): Update.
* eval.c (fetch_subexp_value): Add "preserve_errors"
parameter.
* ppc-linux-nat.c (check_condition): Update.
* value.h (fetch_subexp_value): Update.

* gdb.base/watchpoint.c (struct foo5): New.
(nullptr): New global.
* gdb.base/watchpoint.exp (test_watch_location): Add test.

Remove duplicate call to add_file_handler.

  http://sourceware.org/ml/gdb-patches/2013-08/msg00067.html

This call to add_file_handler is a duplicate of one that is in
gdb_setup_readline that is always executed.

gdb/ChangeLog

        * mi/mi-interp.c (mi_interpreter_resume): Remove call to
        add_file_handler.

PR symtab/15691
* dwarf2read.c (struct dwarf2_per_cu_data): New member tu_read.
(fill_in_sig_entry_from_dwo_entry): Reorganize asserts.
Add assert of sig_entry->dwo_unit == NULL.
(lookup_dwo_signatured_type): Don't assign TU to a DWO if the TU
had already been read.
(read_signatured_type): Set per_cu.tu_read.

testsuite/
* gdb.dwarf2/fission-mix.exp: New file.
* gdb.dwarf2/fission-mix.h: New file.
* gdb.dwarf2/fission-mix.c: New file.
* gdb.dwarf2/fission-mix2.c: New file.

daily update

PR symtab/15695
* valops.c (value_struct_elt): Add missing call to check_typedef.
(value_find_oload_method_list): Ditto.

testsuite/
* gdb.base/func-ptr.exp: New file.
* gdb.base/func-ptr.c: New file.

* symtab.c (do_free_search_symbols_cleanup): Change arg to,
effectively, struct symbol_search **.
(make_cleanup_free_search_symbols): Change arg to struct
symbol_search **.  All callers updated.
(compare_search_syms): Compare symtab file name and block as well.
(search_symbols_equal): New function.
(sort_search_symbols_remove_dups): Renamed from sort_search_symbols.
New args new_head, new_tail.  Result is now void.  Remove dups after
sorting the symbols.
(search_symbols): Sort all found symbols once, after all have been
found, and remove duplicates.  Simplify cleanup tracking of result.
* symtab.h (make_cleanup_free_search_symbols): Update prototype.

Further workarounds for binutils/15021.
* dwarf2read.c (recursively_compute_inclusions): Change type of result
parameter to VEC (symtab_ptr) **.  New parameter all_type_symtabs.
Watch for duplicate symtabs coming from type units.
(compute_symtab_includes): Update call to
recursively_compute_inclusions. Build vector of included symtabs
instead of per_cus.
* symtab.h (symtab_ptr): New typedef.
(DEF_VEC_P (symtab_ptr)): New VEC type.
* linespec.c (symtab_p): Delete.  All uses updated to use symtab_ptr
instead.

opcodes/
* micromips-opc.c (WR_s): Delete.

include/opcode/
* mips.h (mips_decode_reg_operand): New function.
(INSN_WRITE_SHIFT, INSN_WRITE_1, INSN_WRITE_2, INSN_WRITE_ALL)
(INSN_READ_SHIFT, INSN_READ_1, INSN_READ_2, INSN_READ_3, INSN_READ_4)
(INSN_READ_ALL, INSN_READ_GPR_24, INSN_WRITE_GPR_24, INSN_UDI):
New macros.
(INSN_WRITE_GPR_D, INSN_WRITE_GPR_T, INSN_WRITE_FPR_D)
(INSN_WRITE_FPR_S, INSN_WRITE_FPR_T, INSN_READ_GPR_S, INSN_READ_GPR_T)
(INSN_READ_FPR_S, INSN_READ_FPR_T, INSN_READ_FPR_R, INSN_WRITE_GPR_S)
(INSN2_WRITE_GPR_Z, INSN2_WRITE_FPR_Z, INSN2_READ_GPR_Z)
(INSN2_READ_FPR_Z, INSN2_READ_GPR_D, INSN2_READ_FPR_D)
(INSN2_WRITE_GPR_MB, INSN2_READ_GPR_MC, INSN2_MOD_GPR_MD)
(INSN2_READ_GPR_ME, INSN2_MOD_GPR_MF, INSN2_READ_GPR_MG)
(INSN2_READ_GPR_MJ, INSN2_WRITE_GPR_MJ, INSN2_READ_GPR_MP)
(INSN2_WRITE_GPR_MP, INSN2_READ_GPR_MQ, INSN2_READ_GP)
(INSN2_WRITE_GPR_MH, INSN2_READ_GPR_MMN): Delete.  Renumber other
macros to cover the gaps.
(INSN2_MOD_SP): Replace with...
(INSN2_WRITE_SP, INSN2_READ_SP): ...these new macros.
(MIPS16_INSN_WRITE_X, MIPS16_INSN_WRITE_Y, MIPS16_INSN_WRITE_Z)
(MIPS16_INSN_WRITE_T, MIPS16_INSN_WRITE_31, MIPS16_INSN_WRITE_GPR_Y)
(MIPS16_INSN_READ_X, MIPS16_INSN_READ_Y, MIPS16_INSN_READ_Z)
(MIPS16_INSN_READ_T, MIPS16_INSN_READ_SP, MIPS16_INSN_READ_GPR_X):
Delete.

opcodes/
* mips-opc.c (WR_1, WR_2, RD_1, RD_2, RD_3, RD_4, MOD_1, MOD_2, UDI):
New macros.
(WR_d, WR_t, WR_D, WR_T, WR_S, RD_s, RD_b, RD_t, RD_S, RD_T, RD_R)
(WR_z, WR_Z, RD_z, RD_Z, RD_d): Delete.
(mips_builtin_opcodes): Use the new position-based read-write flags
instead of field-based ones.  Use UDI for "udi..." instructions.
* mips16-opc.c (WR_1, WR_2, RD_1, RD_2, RD_3, RD_4, MOD_1, MOD_2):
New macros.
(WR_x, WR_y, WR_z, WR_Y, RD_x, RD_y, RD_Z, RD_X): Delete.
(RD_T, WR_T, WR_31): Redefine using generic INSN_* flags.
(WR_SP, RD_16): New macros.
(RD_SP): Redefine as an INSN2_* flag.
(MOD_SP): Redefine in terms of RD_SP and WR_SP.
(mips16_opcodes): Use the new position-based read-write flags
instead of field-based ones.  Use RD_16 for "nop".  Move RD_SP to
pinfo2 field.
* micromips-opc.c (WR_1, WR_2, RD_1, RD_2, RD_3, RD_4, MOD_1, MOD_2):
New macros.
(WR_mb, RD_mc, RD_md, WR_md, RD_me, RD_mf, WR_mf, RD_mg, WR_mh, RD_mj)
(WR_mj, RD_ml, RD_mmn, RD_mp, WR_mp, RD_mq, RD_gp, WR_d, WR_t, WR_D)
(WR_T, WR_S, RD_s, RD_b, RD_t, RD_T, RD_S, RD_R, RD_D): Delete.
(RD_sp, WR_sp): Redefine to INSN2_READ_SP and INSN2_WRITE_SP.
(micromips_opcodes): Use the new position-based read-write flags
instead of field-based ones.
* mips-dis.c (print_insn_arg): Use mips_decode_reg_operand.
(print_insn_mips, print_insn_micromips): Use INSN_WRITE_1 instead
of field-based flags.

gas/
* config/tc-mips.c (MAX_OPERANDS): New macro.
(mips_operand_array): New structure.
(mips_operands, mips16_operands, micromips_operands): New arrays.
(micromips_to_32_reg_b_map, micromips_to_32_reg_c_map)
(micromips_to_32_reg_e_map, micromips_to_32_reg_f_map)
(micromips_to_32_reg_g_map, micromips_to_32_reg_l_map)
(micromips_to_32_reg_q_map): Delete.
(insn_operands, insn_opno, insn_extract_operand): New functions.
(validate_mips_insn): Take a mips_operand_array as argument and
use it to build up a list of operands.  Extend to handle INSN_MACRO
and MIPS16.
(validate_mips16_insn): New function.
(validate_micromips_insn): Take a mips_operand_array as argument.
Handle INSN_MACRO.
(md_begin): Initialize mips_operands, mips16_operands and
micromips_operands.  Call validate_mips_insn and
validate_micromips_insn for macro instructions too.
Call validate_mips16_insn for MIPS16 instructions.
(insn_read_mask, insn_write_mask, operand_reg_mask, insn_reg_mask):
New functions.
(gpr_read_mask, gpr_write_mask, fpr_read_mask, fpr_write_mask): Use
them.  Handle INSN_UDI.
(get_append_method): Use gpr_read_mask.

include/opcode/
* mips.h (MIPS16_INSN_WRITE_SP, MIPS16_INSN_READ_31)
(MIPS16_INSN_READ_PC, MIPS16_INSN_UNCOND_BRANCH)
(MIPS16_INSN_COND_BRANCH): Delete.

opcodes/
* mips16-opc.c (UBR, CBR, RD_31, RD_PC): Redefine as INSN2_* flags.
(WR_SP): Replace with...
(MOD_SP): ...this.
(mips16_opcodes): Update accordingly.
* mips-dis.c (print_insn_mips16): Likewise.

gas/
* config/tc-mips.c (compact_branch_p, uncond_branch_p): Use the same
flags for MIPS16 and non-MIPS16 instructions.
(gpr_mod_mask): Move the INSN2_MOD_SP case outside the micromips block.
(gpr_read_mask): Use INSN2_READ_GPR_31 for MIPS16 instructions too.
(gpr_write_mask): Remove MIPS16_INSN_WRITE_SP handling.
(can_swap_branch_p, get_append_method): Use the same flags for MIPS16
and non-MIPS16 instructions.  Fix formatting.

opcodes/
* mips16-opc.c (mips16_opcodes): Reformat.

gas/
* config/tc-mips.c (reg_needs_delay): Move later in file.
Use gpr_write_mask.
(insns_between): Use gpr_read_mask instead of EXTRACT_OPERAND.

opcodes/
* mips-opc.c (mips_builtin_opcodes): Remove WR_* and RD_* flags
for operands that are hard-coded to $0.
* micromips-opc.c (micromips_opcodes): Likewise.

opcodes/
* mips-opc.c (mips_builtin_opcodes): Use WR_31 rather than WR_d
for the single-operand forms of JALR and JALR.HB.
* micromips-opc.c (micromips_opcodes): Likewise JALR, JALRS, JALR.HB
and JALRS.HB.

opcodes/
* mips-opc.c (mips_builtin_opcodes): Add FP_D to VR5400 vector
instructions.  Fix them to use WR_MACC instead of WR_CC and
add missing RD_MACCs.

Remove error_pre_print and quit_pre_print.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00844.html

gdb/ChangeLog

        * cli/cli-script.c (script_from_file): Remove use of
        error_pre_print.
        * main.c (captured_main): Remove use of error_pre_print and
        quit_pre_print.
        * utils.c (error_pre_print, quit_pre_print): Remove.
        * utils.h (error_pre_print, quit_pre_print): Likewise.

opcodes/
* mips-dis.c (print_mips16_insn_arg): Include ISA bit in base address.

gdb/
* mi/mi-cmd-stack.c (mi_cmd_stack_list_locals): Parse argv
with mi_getopt.
(mi_cmd_stack_list_variables): Likewise.

gdb/testsuite/

* gdb.python/py-sync-interp.c: New.
* gdb.python/py-sync-interp.exp: New.

daily update

gold/
* object.cc (Sized_relobj::do_output_section_address): New function.
(Sized_relobj): Instantiate explicitly.
* object.h (Object::output_section_address): New function.
(Object::do_output_section_address): New function.
(Sized_relobj::do_output_section_address): New function.
* powerpc.cc (Target_powerpc::symval_for_branch): Use it.

PR ld/15787
* elf32-arm.c (elf32_arm_final_link_relocate): Base SB on the
output section VMA.

* ld-arm/group-relocs-alu-bad-2.d; New.
* ld-arm/group-relocs-alu-bad-2.s: New.
* ld-arm/group-relocs-ldc-bad-2.d: New.
* ld-arm/group-relocs-ldc-bad-2.s: New.
* ld-arm/group-relocs-ldr-bad-2.d: New.
* ld-arm/group-relocs-ldr-bad-2.s: New.
* ld-arm/group-relocs-ldrs-bad-2.d: New.
* ld-arm/group-relocs-ldrs-bad-2: New.
* ld-arm/arm-elf.exp: Add the new tests.
* ld-arm/group-relocs-ldr-bad.d: Update expected output.
* ld-arm/group-relocs-ldr-bad.s: Likewise.
* ld-arm/group-relocs-ldrs-bad.d: Likewise.
* ld-arm/group-relocs-ldrs-bad.s: Likewise.
* ld-arm/group-relocs.d: Likewise.
* ld-arm/group-relocs.s: Likewise.

Remove deprecated_throw_reason.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00779.html

gdb/ChangeLog

        * exceptions.c (deprecated_throw_reason): Remove.
        * exceptions.h (deprecated_throw_reason): Remove.

Remove deprecated_throw_reason from mips_error.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00777.html

gdb/ChangeLog

        * remote-mips.c (mips_error): Replace use of
        deprecated_throw_reason with throw_verror.  Use the error message
        passed to mips_error as the error message for throw_verror.

Replace most uses of deprecated_throw_reason with quit.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00778.html

gdb/ChangeLog

        * monitor.c (monitor_interrupt_query): Replace use of
        deprecated_throw_reason with quit.
        * nto-procfs.c (interrupt_query): Likewise.
        * remote-fileio.c (remote_fileio_sig_exit): Likewise.
        * remote-mips.c (mips_kill): Likewise.
        * remote.c (interrupt_query): Likewise.

Remove deprecated_throw_reason from internal_verror.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00776.html

gdb/ChangeLog

        * utils.c (internal_verror): Replace use of deprecated_throw_reason
        with call to fatal.

gdb/

* tracepoint.c (trace_dump_command): Select the current frame.

gdb/testsuite/

* gdb.trace/backtrace.exp (gdb_backtrace_tdp_4): Test command
'tdump' on stack frame 0 and 1 respectively.

gdb/testsuite/

PR gdb/13443
* gdb.trace/backtrace.exp (gdb_backtrace_tdp_3): Add parameter
'traceframe'.  Wrap test with with_test_prefix.
(top level): Update.

gdb/testsuite/

* gdb.trace/backtrace.exp: Don't invoke 'get_exit' and
'gdb_start'.
Invoke 'prepare_for_testing' instead of 'gdb_compile'.
Use gdb_test_no_output to issue commands 'tstart' and 'tstop'.

* dwarf2read.c (process_queue): Add type signature to debug output.

daily update

Add missing author to last ChangeLog entry.

gold/
     * parameters.cc (Parameters::entry): Return target-specific entry
     symbol name.
     * target.h (Target::entry_symbol_name): New function.
     (Target_info::entry_symbol_name): New data member.

     * arm.cc (Target_arm::arm_info): Add entry_symbol_name.
     (Target_arm_nacl::arm_nacl_info): Likewise.
     * i386.cc (Target_i386::i386_info): Likewise.
     (Target_i386_nacl::i386_nacl_info): Likewise.
     * powerpc.cc (Target_powerpc::powerpc_info): Likewise.
     * sparc.cc (Target_sparc::sparc_info): Likewise.
     * tilegx.cc (Target_tilegx::tilegx_info): Likewise.
     * x86_64.cc: (Target_x86_64::x86_64_info) Likewise.
     (Target_x86_64_nacl::x86_64_nacl_info) Likewise.
     * testsuite/testfile.cc (Target_test::test_target_info): Likewise.

Don't raise an error for optimized out sub-fields.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00317.html

gdb/ChangeLog

        * value.c (value_fetch_lazy): Mark optimized out values as such
        rather than raising an error.

gdb/testsuite/ChangeLog

        * gdb.dwarf2/pieces-optimized-out.exp: Expect "<optimized out>"
        when printing an optimized out value.  Expect an error when using
        an optimized out value in an expression.

value_fetch_lazy, ensure parent is not lazy and check parent for validity.

  http://sourceware.org/ml/gdb-patches/2013-07/msg00316.html

gdb/ChangeLog

        * value.c (value_fetch_lazy): Ensure parent value is not lazy
        before checking which bits of the parent, not the child, value are
        valid.

gdb/testsuite/ChangeLog

        * gdb.dwarf2/pieces-optimized-out.exp: New file.
        * gdb.dwarf2/pieces-optimized-out.c: New file.
        * gdb.dwarf2/pieces-optimized-out.S: New file.

2013-07-30  Muhammad Bilal  <mbilal@codesorcery.com>

        PR gdb/15715
        * top.c: Include "filenames.h".
        (set_history_filename): New function.
        (init_main): Install it as set hook of the "set history filename"
        command.
2013-07-30  Muhammad Bilal  <mbilal@codesourcery.com>

        PR gdb/15715
        * gdb.base/setshow.exp: Test that relative paths passed to
        'set history filename' are converted to absolute paths.

cleanup: constify "struct attribute" function parameter

gdb/

2013-07-30  Sanimir Agovic  <sanimir.agovic@intel.com>

* dwarf2read.c (dwarf2_get_ref_die_offset): Constify struct
attribute parameter.
(dwarf2_const_value_data): Constify struct attribute parameter.
(dwarf2_const_value): Constify struct attribute parameter.
(dwarf2_const_value_attr): Constify struct attribute parameter.
(lookup_die_type): Constify struct attribute parameter.
(dwarf2_get_attr_constant_value): Constify struct attribute parameter.
(follow_die_ref_or_sig): Constify struct attribute parameter.
(follow_die_ref): Constify struct attribute parameter.
(follow_die_sig): Constify struct attribute parameter.
(get_DW_AT_signature_type): Constify struct attribute parameter.
(get_type_unit_group): Constify struct attribute parameter.
(fill_in_loclist_baton): Constify struct attribute parameter.
(dwarf2_symbol_mark_computed): Constify struct attribute parameter.
(type_unit_group): Constify struct attribute parameter.

cleanup: constify argument passed to dwarf form predicates

2013-07-30  Sanimir Agovic  <sanimir.agovic@intel.com>

gdb/
* dwarf2read.c (attr_form_is_block): Make argument const.
(attr_form_is_section_offset): Make argument const.
(attr_form_is_constant): Make argument const.
(attr_form_is_ref): Make argument const.

cleanup: rename is_ref_attr to attr_form_is_ref

2013-07-30  Sanimir Agovic  <sanimir.agovic@intel.com>

gdb/
* dwarf2read.c (attr_is_ref): Rename to attr_form_is_ref.
All uses updated.
(attr_form_is_ref): Moved below attr_form_is_constant.

opcodes/
* ppc-dis.c (powerpc_init_dialect): Use ppc_parse_cpu() to set dialect.

daily update

* main.c (captured_command_loop): Tweak comment.

* target.c (target_async_permitted_1): Fix comment.

* symtab.c (iterate_over_some_symtabs): Add comment.

* symtab.c (iterate_over_some_symtabs): Fix indentation.