Define __STDC_CONSTANT_MACROS/__STDC_LIMIT_MACROS again.

Revert commit f6abaf7a4088 (gdb: no longer define
__STDC_CONSTANT_MACROS/__STDC_LIMIT_MACROS), with the tweak suggested
in that commit's log: the macros are now defined before any system
header is included.

This should fix AIX:
  https://sourceware.org/ml/gdb-patches/2016-10/msg00682.html

gdb/ChangeLog:
2016-10-25  Pedro Alves  <[email protected]>

       * common/common-defs.h (__STDC_CONSTANT_MACROS)
       (__STDC_LIMIT_MACROS): Define.

Don't override operator new if GDB is built with -fsanitize=address

Nowadays, if we build GDB with -fsanitize=address, we can get the asan
error below,

(gdb) quit
=================================================================
==9723==ERROR: AddressSanitizer: alloc-dealloc-mismatch (malloc vs operator delete) on 0x60200003bf70
    #0 0x7f88f3837527 in operator delete(void*) (/usr/lib/x86_64-linux-gnu/libasan.so.1+0x55527)
    #1 0xac8e13 in __gnu_cxx::new_allocator<void (*)()>::deallocate(void (**)(), unsigned long) /usr/include/c++/4.9/ext/new_allocator.h:110
    #2 0xac8cc2 in __gnu_cxx::__alloc_traits<std::allocator<void (*)()> >::deallocate(std::allocator<void (*)()>&, void (**)(), unsigned long) /usr/include/c++/4.9/ext/alloc_traits.h:185
....
0x60200003bf70 is located 0 bytes inside of 8-byte region [0x60200003bf70,0x60200003bf78)
allocated by thread T0 here:
    #0 0x7f88f38367ef in __interceptor_malloc (/usr/lib/x86_64-linux-gnu/libasan.so.1+0x547ef)
    #1 0xbd2762 in operator new(unsigned long) /home/yao/SourceCode/gnu/gdb/git/gdb/common/new-op.c:42
    #2 0xac8edc in __gnu_cxx::new_allocator<void (*)()>::allocate(unsigned long, void const*) /usr/include/c++/4.9/ext/new_allocator.h:104
    #3 0xac8d81 in __gnu_cxx::__alloc_traits<std::allocator<void (*)()> >::allocate(std::allocator<void (*)()>&, unsigned long) /usr/include/c++/4.9/ext/alloc_traits.h:182

The reason for this is that we override operator new but don't override
operator delete.  This patch does the override if the code is NOT
compiled with asan.

gdb:

2016-10-25  Yao Qi  <[email protected]>

PR gdb/20716
* common/new-op.c (__has_feature): New macro.
Don't override operator new if asan is used.

elfxx-mips: Correct STUB_JALR's description

bfd/
* elfxx-mips.c (STUB_JALR): Correct description.

Automatic date update in version.in

Fix potential NULL pointer dereference

This patch addresses a potential NULL pointer dereference when we try to
duplicate a string. The input pointer can be NULL and that may lead to
crashes. We simply add a check for that case.

gdb/ChangeLog:
2016-10-24  Luis Machado  <[email protected]>

* exec.c (exec_file_locate_attach): Prevent NULL pointer dereference
when duplicating a string.

Fix obvious gotcha in string comparison

This patch fixes a gotcha when comparing exception's messages in
exception_print_same. It should've used the statically-allocated
string versions msg1 and msg2 instead.

As is, it could lead to crashes.

gdb/ChangeLog:
2016-10-24  Luis Machado  <[email protected]>

* exec.c (exception_print_same): Fix string comparison to use
statically-allocated ones.

testsuite: Fix false FAIL for gdb.base/morestack.exp

Since
[commit] [testsuite patch] Fix gcc_compiled for gcc 6 & 7
https://sourceware.org/ml/gdb-patches/2016-10/msg00620.html

there has started running again
 Running gdb/testsuite/gdb.base/morestack.exp ...
+FAIL: gdb.base/morestack.exp: continue
+PASS: gdb.base/morestack.exp: up 3000

but as you can see it FAILs now - on Fedora 24 x86_64 (although for example it
still PASSes on CentOS-7.2 x86_64).

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff787c7bb in malloc_consolidate (av=av@entry=0x7ffff7bbcb00 <main_arena>) at malloc.c:4181
4181       unlink(av, nextchunk, bck, fwd);
(gdb) bt
[...]
[...]

This apparently is due to - man gcc - -fsplit-stack:
When code compiled with -fsplit-stack calls code compiled without
-fsplit-stack, there may not be much stack space available for the
latter code to run.  If compiling all code, including library code,
with -fsplit-stack is not an option, then the linker can fix up these
calls so that the code compiled without -fsplit-stack always has
a large stack.  Support for this is implemented in the gold linker in
GNU binutils release 2.21 and later.

gdb/testsuite/ChangeLog
2016-10-24  Jan Kratochvil  <[email protected]>

* gdb.base/morestack.exp: Try to build it using -fuse-ld=gold first.

[GDBserver] Fix conversion warning

I got the following warning if I build GDBserver for aarch64_be-linux-gnu,

git/gdb/gdbserver/linux-aarch64-low.c:1539:39: error: invalid conversion from 'void*' to 'uint32_t* {aka unsigned int*}' [-fpermissive]
   uint32_t *le_buf = xmalloc (byte_len);
                                       ^
The patch is to fix the warning.

gdb/gdbserver:

2016-10-24  Yao Qi  <[email protected]>

PR server/20733
* linux-aarch64-low.c (append_insns): Cast the return value to
'uint32_t *'.

Automatic date update in version.in

Automatic date update in version.in

Automatic date update in version.in

Make some dwarf_expr_context methods pure virtual

This patch changes some dwarf_expr_context to be pure virtual, as
mentioned during the discussion of an earlier patch in this series.

2016-10-21  Tom Tromey  <[email protected]>

* dwarf2expr.h (class dwarf_expr_context)
<get_frame_base, get_frame_cfa, get_tls_address, dwarf_call,
push_dwarf_block_entry_value, get_addr_index, get_object_address>:
Now pure virtual.
* dwarf2-frame.c (class dwarf_expr_executor)
<get_frame_base, get_frame_cfa, get_tls_address, dwarf_call,
push_dwarf_block_entry_value, get_addr_index, get_object_address>:
New methods.
<invalid>: New method.

Change minimal_symbol_reader::record_full to take a bool

This changes an "int" to a "bool" in the signature for
minimal_symbol_reader::record_full, and then fixes the callers.

2016-10-21  Tom Tromey  <[email protected]>

* minsyms.h (minimal_symbol_reader::record_full): "copy_name" now
a bool.
(record, record_with_info): Update.
* minsyms.c (record): Fix indentation.
(record_full): Fix indentation.  Update for type change.
* elfread.c (record_minimal_symbol): "copy_name" now a bool.
(elf_symtab_read): "copy_names" now a bool.
(elf_rel_plt_read, elf_read_minimal_symbols): Update.

Remove last cleanup from captured_main_1

An earlier patch split captured_main into two parts.  This patch
removes the last remaining cleanup from captured_main_1, and also
replaces a second hand-rolled VEC-like array with a std::vector.

2016-10-21  Tom Tromey  <[email protected]>

* main.c: Include <vector>.
(cmdarg_s): Remove typedef.  Don't define VEC.
(captured_main_1): Use vector, not VEC.  Remove cleanups.

Convert dwarf_expr_context_funcs to methods

This patch converts the function pointers in dwarf_expr_context_funcs
into methods on dwarf_expr_context, and then updates the various
implementations and callers to follow.

NB this patch uses "override" (which caught a couple of renaming bugs
during development) -- but this is C++11, so this patch at least has
to wait for Pedro's patch that adds the OVERRIDE macro.

After this patch it would be possible to do one more, that makes
various members of dwarf_expr_context "protected"; but I haven't done
this.

2016-10-21  Tom Tromey  <[email protected]>

* dwarf2loc.c (struct dwarf_expr_context_funcs): Don't declare.
(dwarf_expr_read_addr_from_reg, dwarf_expr_get_reg_value)
(dwarf_expr_read_mem, dwarf_expr_frame_base): Rename; turn into
methods.
(get_frame_pc_for_per_cu_dwarf_call): New function.
(dwarf_expr_frame_cfa, dwarf_expr_frame_pc)
(dwarf_expr_tls_address): Rename; turn into methods.
(per_cu_dwarf_call): Remove arguments.  Use
get_frame_pc_for_per_cu_dwarf_call.
(dwarf_evaluate_loc_desc): New class.
(dwarf_expr_dwarf_call, dwarf_expr_context)
(dwarf_expr_push_dwarf_reg_entry_value)
(dwarf_expr_get_addr_index, dwarf_expr_get_obj_addr): Rename; turn
into methods.
(dwarf_expr_ctx_funcs): Remove.
(dwarf2_evaluate_loc_desc_full): Update.
(dwarf2_locexpr_baton_eval): Update.
(symbol_needs_eval_context): New class.
(symbol_needs_read_addr_from_reg, symbol_needs_get_reg_value)
(symbol_needs_read_mem, symbol_needs_frame_base)
(symbol_needs_frame_cfa, symbol_needs_tls_address)
(symbol_needs_dwarf_call, needs_dwarf_reg_entry_value): Rename;
turn into methods.
(needs_get_addr_index, needs_get_obj_addr): Remove; turn into
methods.
(symbol_needs_ctx_funcs): Remove.
(dwarf2_loc_desc_get_symbol_read_needs): Update.
* dwarf2expr.h (struct dwarf_expr_context_funcs): Remove; turn
contents into methods.
(struct dwarf_expr_context) <baton, funcs>: Remove.
<read_addr_from_reg, get_reg_value, read_mem, get_frame_base,
get_frame_cfa, get_frame_pc, get_tls_address, dwarf_call,
impl_get_base_type, push_dwarf_block_entry_value, get_addr_index,
get_object_address>: Declare new methods.
(ctx_no_get_frame_base, ctx_no_get_frame_cfa)
(ctx_no_get_frame_pc, ctx_no_get_tls_address, ctx_no_dwarf_call)
(ctx_no_get_base_type, ctx_no_push_dwarf_reg_entry_value)
(ctx_no_get_addr_index): Don't declare.
* dwarf2expr.c (get_base_type): Use impl_get_base_type.
(execute_stack_op): Update.
(ctx_no_get_frame_base, ctx_no_get_frame_cfa)
(ctx_no_get_frame_pc, ctx_no_get_tls_address, ctx_no_dwarf_call)
(ctx_no_get_base_type, ctx_no_push_dwarf_reg_entry_value)
(ctx_no_get_addr_index): Remove; now methods on
dwarf_expr_context.
* dwarf2-frame.c (read_addr_from_reg): Take a frame_info, not a
baton.
(class dwarf_expr_executor): New class.
(get_reg_value, read_mem): Rename, turn into methods.
(execute_stack_op): Use dwarf_expr_executor.

Convert DWARF expr functions to methods

This converts various DWARF expr functions to be members on
dwarf_expr_context, then fixes up the various users.  This results in
somewhat less wordy code and sets the stage for the next patch.

2016-10-21  Tom Tromey  <[email protected]>

* dwarf2loc.c (per_cu_dwarf_call)
(dwarf_expr_push_dwarf_reg_entry_value)
(dwarf2_evaluate_loc_desc_full, dwarf2_locexpr_baton_eval)
(needs_dwarf_reg_entry_value)
(dwarf2_loc_desc_get_symbol_read_needs): Update.
* dwarf2expr.h (dwarf_expr_context) <push_address, eval, fetch,
fetch_address, fetch_in_stack_memory, address_type, grow_stack,
push, stack_empty_p, add_piece, get_base_type, execute_stack_op,
pop>: New method declarations.
(dwarf_expr_push_address, dwarf_expr_eval, dwarf_expr_fetch)
(dwarf_expr_fetch_address, dwarf_expr_fetch_in_stack_memory):
Don't declare.
* dwarf2expr.c (address_type, grow_stack, push, push_address)
(pop, fetch, fetch_address, fetch_in_stack_memory)
(stack_empty_p, add_piece, eval, get_base_type)
(execute_stack_op): Rename.  Turn into methods.
* dwarf2-frame.c (execute_stack_op): Update.

Initial conversion of dwarf_expr_ctx

This is the first step in the conversion of dwarf_expr_ctx to a C++
class.  This conversion is done in steps to make the patches, and the
reviews, a bit simpler.  This patch changes dwarf_expr_ctx to be
stack-allocated and removes the associated cleanup.

2016-10-21  Tom Tromey  <[email protected]>

* dwarf2loc.c (dwarf2_evaluate_loc_desc_full): Stack-allocate
dwarf_expr_context.  Remove cleanups.
(dwarf2_locexpr_baton_eval)
(dwarf2_loc_desc_get_symbol_read_needs):  Likewise.
* dwarf2expr.h (dwarf_expr_context, ~dwarf_expr_context): Add
constructors and destructors.
(new_dwarf_expr_context, free_dwarf_expr_context)
(make_cleanup_free_dwarf_expr_context): Don't declare.
* dwarf2-frame.c (execute_stack_op): Stack-allocate
dwarf_expr_context.  Remove cleanups.
(dwarf_expr_context): Rename from new_dwarf_expr_context.  Turn
into constructor.
(free_dwarf_expr_context, free_dwarf_expr_context_cleanup):
Remove.
(~dwarf_expr_context): Rename from
make_cleanup_free_dwarf_expr_context.  Turn into destructor.

Some cleanup removal in dwarf2loc.c

This removes some cleanups and manual allocation handling in
dwarf2loc.c with std::vector.  Note that this patch has a case where
the vector would normally fall into the "use gdb::unique_ptr"
guidelines -- but here because the vector is immediately initialized,
I moved the initialization into the constructor call, for further
code savings.

2016-10-21  Tom Tromey  <[email protected]>

* dwarf2loc.c: Include <vector>.
(read_pieced_value, write_pieced_value)
(dwarf2_compile_expr_to_ax): Use std::vector.

Remove make_cleanup_restore_current_uiout

This removes make_cleanup_restore_current_uiout in favor of an
RAII-based class.

2016-10-21  Tom Tromey  <[email protected]>

* stack.c (print_stack_frame_to_uiout): Use scoped_restore.
* ui-out.c (make_cleanup_restore_current_uiout)
(restore_current_uiout_cleanup): Remove.
* infrun.c (print_stop_event): Use scoped_restore.
* ui-out.h (make_cleanup_restore_current_uiout): Don't declare.

Use gdb::unique_ptr in elf_read_minimal_symbols

This changes elf_read_minimal_symbols to use gdb::unique_ptr rather
than an explicit allocation.  This removes a cleanup.

2016-10-21  Tom Tromey  <[email protected]>

* elfread.c (elf_read_minimal_symbols): Use gdb::unique_ptr.

Replace two xmallocs with unique_ptr

This replaces a couple of uses of xmalloc with gdb::unique_ptr, also
removing a couple of cleanups.

2016-10-21  Tom Tromey  <[email protected]>

* cli/cli-dump.c (dump_memory_to_file): Use gdb::unique_ptr.
(restore_binary_file): Likewise.

Change command stats reporting to use class

This removes make_command_stats_cleanup in favor of an RAII class.
The patch is reasonably straightforward, but keeping the same
semantics without excessive reindentation required splitting
captured_main in two.

2016-10-21  Tom Tromey  <[email protected]>

* maint.h (scoped_command_stats): New class.
(make_command_stats_cleanup): Don't declare.
* maint.c (struct cmd_stats): Remove.
(~scoped_command_stats): Rename from report_command_stats.  Now a
destructor.
(scoped_command_stats): Rename from make_command_stats_cleanup.
Now a constructor.
* main.c (captured_main_1): New function.  Use
scoped_command_stats.
(captured_main): Call captured_main_1.
* event-top.c (command_handler): Use scoped_command_stats.

Remove some cleanups in MI

This patch removes a couple of cleanups from MI by using
gdb::unique_ptr.

2016-10-21  Tom Tromey  <[email protected]>

* mi/mi-main.c (mi_cmd_data_read_memory): Use gdb::unique_ptr.
Remove some cleanups.

Remove make_cleanup_restore_current_ui

This removes make_cleanup_restore_current_ui by converting the last
use.  The last use was in a few functions used to iterate over all
UIs.  This patch replaces these functions with a class, and arranges
for the class destructor to do the needed cleanup.

2016-10-21  Tom Tromey  <[email protected]>

* tui/tui-interp.c (tui_on_normal_stop, tui_on_signal_received)
(tui_on_end_stepping_range, tui_on_signal_exited, tui_on_exited)
(tui_on_no_history, tui_on_user_selected_context_changed):
Update.
* top.h (switch_thru_all_uis): New class.
(SWITCH_THRU_ALL_UIS): Rewrite.
(make_cleanup_restore_current_ui, switch_thru_all_uis_init)
(switch_thru_all_uis_cond, switch_thru_all_uis_next): Don't
declare.
* mi/mi-interp.c (mi_new_thread, mi_thread_exit)
(mi_record_changed, mi_inferior_added, mi_inferior_appeared)
(mi_inferior_exit, mi_inferior_removed, mi_on_signal_received)
(mi_on_end_stepping_range, mi_on_signal_exited, mi_on_exited)
(mi_on_no_history, mi_on_normal_stop, mi_traceframe_changed)
(mi_tsv_created, mi_tsv_deleted, mi_tsv_modified)
(mi_breakpoint_created, mi_breakpoint_deleted)
(mi_breakpoint_modified, mi_output_running_pid, mi_on_resume)
(mi_solib_loaded, mi_solib_unloaded, mi_command_param_changed)
(mi_memory_changed, mi_user_selected_context_changed): Update.
* infrun.c (all_uis_check_sync_execution_done)
(all_uis_on_sync_execution_starting, normal_stop): Update.
* event-top.c (restore_ui_cleanup)
(make_cleanup_restore_current_ui, switch_thru_all_uis_init)
(switch_thru_all_uis_cond, switch_thru_all_uis_next): Remove.
* cli/cli-interp.c (cli_on_normal_stop, cli_on_signal_received)
(cli_on_end_stepping_range, cli_on_signal_exited, cli_on_exited)
(cli_on_no_history, cli_on_user_selected_context_changed):
Update.
* breakpoint.c (watchpoint_check): Update.

Record minimal symbols directly in reader.

This patch changes minimal symbol creation in two ways.  First, it
removes global variables in favor of members of minimal_symbol_reader.
Second, it changes functions like prim_record_minimal_symbol to be
member functions of minimal_symbol_reader.

2016-10-21  Tom Tromey  <[email protected]>

* xcoffread.c (record_minimal_symbol, scan_xcoff_symtab): Add
"reader" argument.  Update.
(xcoff_initial_scan): Update.
* symfile.h (mdebug_build_psymtabs): Add "reader" argument.
* mipsread.c (mipscoff_symfile_read): Update.
(read_alphacoff_dynamic_symtab): Add "reader" argument.  Update.
* minsyms.h (minimal_symbol_reader) <record, record_full>:
Declare.
<m_msym_bunch, m_msym_bunch_index, m_msym_count>: New members.
<record_with_info>: New function, renamed from
prim_record_minimal_symbol_and_info.
* minsyms.c (msym_bunch, msym_bunch_index, msym_count): Remove
globals.
(minimal_symbol_reader): Initialize new members.
(minimal_symbol_reader::record): Renamed from
prim_record_minimal_symbol.
(minimal_symbol_reader::record_full): Renamed from
prim_record_minimal_symbol_full.
(prim_record_minimal_symbol_and_info): Move to minsyms.h; rename.
* mdebugread.c (mdebug_build_psymtabs, parse_partial_symbols)
(record_minimal_symbol): Add "reader" argument.  Update.
(elfmdebug_build_psymtabs): Update.
* machoread.c (macho_symtab_add_minsym, macho_symtab_read): Add
"reader" argument.  Update.
(macho_symfile_read): Update.
* elfread.c (record_minimal_symbol, elf_symtab_read)
(elf_rel_plt_read): Add "reader" argument.  Update.
(elf_read_minimal_symbols): Update.
* dbxread.c (record_minimal_symbol, read_dbx_dynamic_symtab)
(read_dbx_symtab): Add "reader" argument.  Update.
(dbx_symfile_read): Update.
* coffread.c (record_minimal_symbol, coff_symtab_read): Add
"reader" argument.  Update.
(coff_symfile_read): Update.
* coff-pe-read.h (read_pe_exported_syms): Add "reader" argument.
* coff-pe-read.c (add_pe_exported_sym, add_pe_forwarded_sym)
(read_pe_exported_syms): Add "reader" argument.  Update.

Change minimal_symbol_reader to store objfile

This changes minimal_symbol_reader to require the objfile to be
passed to the constructor.  The class now records the objfile and
automatically uses it later in "install".

This is a minor cleanup that will come in useful in the next patch.
It is separate from the first patch to keep that one a bit simpler to
understand.

2016-10-21  Tom Tromey  <[email protected]>

* xcoffread.c (xcoff_initial_scan): Update.
* mipsread.c (mipscoff_symfile_read): Update.
* minsyms.c (minimal_symbol_reader): Add obj argument.
Initialize member.
(install): Remove objfile argument.  Update.
* mdebugread.c (elfmdebug_build_psymtabs): Update.
* machoread.c (macho_symfile_read): Update.
* elfread.c (elf_read_minimal_symbols): Update.
* dbxread.c (dbx_symfile_read): Update.
* coffread.c (coff_symfile_read): Update.
* minsyms.h (minimal_symbol_reader): Add m_objfile member.
(constructor): Add objfile argument.
(minimal_symbol_reader::install): Remove objfile argument.

Introduce minimal_symbol_reader

This patch introduced minimal_symbol_reader, a RAII-based class which
replaces the current make_cleanup_discard_minimal_symbols.

2016-10-21  Tom Tromey  <[email protected]>

* xcoffread.c (xcoff_initial_scan): Use
minimal_symbol_reader.
* mipsread.c (mipscoff_symfile_read): Use
minimal_symbol_reader.
* minsyms.h (minimal_symbol_reader): New class.
(init_minimal_symbol_collection)
(make_cleanup_discard_minimal_symbols, install_minimal_symbols):
Don't declare.
* minsyms.c (minimal_symbol_reader): Renamed from
init_minimal_symbol_collection, turned into constructor.
(~minimal_symbol_reader): Renamed from
do_discard_minimal_symbols_cleanup, turned into destructor.
(make_cleanup_discard_minimal_symbols): Remove.
(minimal_symbol_reader::install): Rename form
install_minimal_symbols.
* mdebugread.c (elfmdebug_build_psymtabs): Use
minimal_symbol_reader.
* machoread.c (macho_symfile_read): Use
minimal_symbol_reader.
* elfread.c (elf_read_minimal_symbols): Use
minimal_symbol_reader.
* dbxread.c (dbx_symfile_read): Use minimal_symbol_reader.
* coffread.c (coff_symfile_read): Use
minimal_symbol_reader.

Use scoped_restore for current_ui

This changes most uses of make_cleanup_restore_current_ui to use
scoped_restore.  The use in switch_thru_all_uis_init still remains;
that is dealt with in a later patch by replacing this iterator with a
real class.

2016-10-21  Tom Tromey  <[email protected]>

* top.c (new_ui_command, wait_sync_command_done)
(gdb_readline_wrapper): Use scoped_restore.
* infrun.c (fetch_inferior_event): Use scoped_restore.
* infcall.c (call_thread_fsm_should_stop): Use scoped_restore.

Use scoped_restore for ui_file

This replaces all the uses of make_cleanup_restore_ui_file with
scoped_restore.

2016-10-21  Tom Tromey  <[email protected]>

* utils.c (make_cleanup_restore_ui_file, do_restore_ui_file)
(struct restore_ui_file_closure): Remove.
* utils.h (make_cleanup_restore_ui_file): Don't declare.
* guile/scm-ports.c (ioscm_with_output_to_port_worker): Use
scoped_restore.
* top.c (execute_command_to_string): Use scoped_restore.

Use RAII to save and restore scalars

This patch replaces many (but not all) uses of
make_cleanup_restore_integer with a simple RAII-based template class.
It also removes the similar restore_execution_direction cleanup in
favor of this new class.  Subsequent patches will replace other
similar cleanups with this class.

The class is typically instantiated using make_scoped_restore.  This
allows for template argument deduction.

2016-10-21  Tom Tromey  <[email protected]>

* common/scoped_restore.h: New file.
* utils.h: Include scoped_restore.h.
* top.c (execute_command_to_string): Use scoped_restore.
* python/python.c (python_interactive_command): Use
scoped_restore.
(python_command, execute_gdb_command): Likewise.
* printcmd.c (do_one_display): Use scoped_restore.
* mi/mi-main.c (exec_continue): Use scoped_restore.
* mi/mi-cmd-var.c (mi_cmd_var_assign): Use scoped_restore.
* linux-fork.c (checkpoint_command): Use scoped_restore.
* infrun.c (restore_execution_direction): Remove.
(fetch_inferior_event): Use scoped_restore.
* compile/compile.c (compile_file_command): Use
scoped_restore.
(compile_code_command, compile_print_command): Likewise.
* cli/cli-script.c (execute_user_command): Use
scoped_restore.
(while_command, if_command, script_from_file): Likewise.
* arm-tdep.c (arm_insert_single_step_breakpoint): Use
scoped_restore.

Fix build failure in xcoffread.c

This changes read_xcoff_symtab to be const-correct.  This fixes a
build failure.

2016-10-21  Tom Tromey  <[email protected]>

* xcoffread.c (read_xcoff_symtab): Make "filestring" const.

X86: Remove pcommit instruction

Remove x86 pcommit instruction support, which has been deprecated:

https://software.intel.com/en-us/blogs/2016/09/12/deprecate-pcommit-instruction

gas/

* config/tc-i386.c (cpu_arch): Remove .pcommit.
* doc/c-i386.texi: Likewise.
* testsuite/gas/i386/i386.exp: Remove pcommit tests.
* testsuite/gas/i386/pcommit-intel.d: Removed.
* testsuite/gas/i386/pcommit.d: Likewise.
* testsuite/gas/i386/pcommit.s: Likewise.
* testsuite/gas/i386/x86-64-pcommit-intel.d: Likewise.
* testsuite/gas/i386/x86-64-pcommit.d: Likewise.
* testsuite/gas/i386/x86-64-pcommit.s: Likewise.

opcodes/

* i386-dis.c (PREFIX_RM_0_0FAE_REG_7): Removed.
(prefix_table): Remove the PREFIX_RM_0_0FAE_REG_7 entry.
(rm_table): Update the RM_0FAE_REG_7 entry.
* i386-gen.c (cpu_flag_init): Remove CPU_PCOMMIT_FLAGS.
(cpu_flags): Remove CpuPCOMMIT.
* i386-opc.h (CpuPCOMMIT): Removed.
(i386_cpu_flags): Remove cpupcommit.
* i386-opc.tbl: Remove pcommit.
* i386-init.h: Regenerated.
* i386-tbl.h: Likewise.

Fix PR 17704.

This fix keeps the section with the highest alignment when folding sections with ICF.

PR gold/17704
* icf.cc (match_sections): Add new parameter section_addraligns.
Check section alignment and keep the section with the strictest
alignment.
(find_identical_sections): New local variable section_addraligns.
Store each section's alignment.
* testsuite/pr17704a_test.s: New file.
* testsuite/Makefile.am (pr17704a_test): New test.
* testsuite/Makefile.in: Regenerate.

Fix symbol table file name on AIX

When xlc -qfuncsect or gcc -ffunction-sections options is used in AIX,
each function csect is associated with each psymtab, so each psymtab
will have it's corresponding filename entries set.

If the pst filename isn't set then we will be seeing the below output
when we set a breakpoint.

(gdb) br main
Breakpoint 1 at 0x10000374: file  _start_ , line 18.

With the fix it will be.

(gdb) br main
Breakpoint 1 at 0x10000518: file test.c, line 24.

Attached patch resolve this issue and correct filename will be set.

gdb/
2016-10-21  Sangamesh Mallayya  <[email protected]>
    Ulrich Weigand  <[email protected]>

* xcoffread.c (read_xcoff_symtab): Make name of current file as
pst->filename instead of _start_ in AIX.

Signed-off-by: Ulrich Weigand <[email protected]>

Delete target_so_ops->special_symbol_handling hook

No one(!) actually implements this hook. So simply delete it.

gdb/ChangeLog:

* solist.h (struct target_so_ops): Delete special_symbol_handling
hook.
* solib.c (solib_add, reload_shared_libraries): Adjust.
* solib-aix.c (solib_aix_special_symbol_handling): Delete
(_initialize_solib_aix): Adjust
* solib-darwin.c (darwin_special_symbol_handling): Delete
(_initialize_darwin_solib): Adjust
* solib-dsbt.c (dsbt_special_symbol_handling): Delete
(_initialize_dsbt_solib): Adjust
* solib-frv.c (frv_special_symbol_handling): Delete
(_initialize_frv_solib): Adjust
* solib-svr4.c (svr4_special_symbol_handling): Delete
(_initialize_svr4_solib): Adjust
* solib-target.c (solib_target_special_symbol_handling): Delete
(_initialize_solib_target): Adjust

Automatic date update in version.in

Check invalid mask registers

In 32-bit, the REX_B bit in the 3-byte VEX prefix is ignored and the
the highest bit in VEX.vvvv is either 1 or ignored.  In 64-bit, we
need to check invalid mask registers.

gas/

PR binutis/20705
* testsuite/gas/i386/i386.exp: Run x86-64-opcode-bad.
* testsuite/gas/i386/x86-64-opcode-bad.d: New file.
* testsuite/gas/i386/x86-64-opcode-bad.s: Likewise.

opcodes/

PR binutis/20705
* i386-dis.c (get_valid_dis386): Ignore the REX_B bit and
the highest bit in VEX.vvvv for the 3-byte VEX prefix in
32-bit mode.  Don't check vex.register_specifier in 32-bit
mode.
(OP_E_register): Check invalid mask registers.
(OP_G): Likewise.
(OP_VEX): Likewise.

testsuite: Fix gcc_compiled for gcc 6 & 7

gdb/testsuite/ChangeLog
2016-10-20  Jan Kratochvil  <[email protected]>

* lib/gdb.exp (get_compiler_info): Generalize gcc_compile regexp.

Regenerate bfd.pot.

Don't configure gdb for vxworks target

VxWorks support was removed in 2004.

commit e84ecc995d6a5e4e9114d3cea61717b8a573afb6
Author: Andrew Cagney <[email protected]>
Date:   Sat Nov 13 23:10:02 2004 +0000

    2004-11-13  Andrew Cagney  <[email protected]>

        * configure.tgt: Delete i[34567]86-*-vxworks*, m68*-netx-*,
        m68*-*-vxworks*, mips*-*-vxworks*, powerpc-*-vxworks*, and
        sparc-*-vxworks*.
        * NEWS: Mention that vxworks was deleted.

This patch adds *-*-vxworks* in a list of targets GDB doesn't
support.

gdb:

2016-10-20  Yao Qi  <[email protected]>

* configure.tgt: Don't configure if target is *-*-vxworks*.

Automatic date update in version.in

testsuite: Fix gdb.base/killed-outside.exp using irrelevant stale options

Fix a commit 4a556533cf02 ("Fix PR11094: JIT breakpoint is not properly
recreated on reruns") regression:

Running .../gdb/testsuite/gdb.base/killed-outside.exp ...
Executing on host: mips-mti-linux-gnu-gcc  -Wl,--no-as-needed  -c -g  -o .../gdb/testsuite/outputs/gdb.base/killed-outside/killed-outside0.o .../gdb/testsuite/gdb.base/killed-outside.c .../gdb/testsuite/outputs/gdb.base/jit-simple/jit-simple-jit.so    (timeout = 300)
spawn mips-mti-linux-gnu-gcc -Wl,--no-as-needed -c -g -o .../gdb/testsuite/outputs/gdb.base/killed-outside/killed-outside0.o .../gdb/testsuite/gdb.base/killed-outside.c .../gdb/testsuite/outputs/gdb.base/jit-simple/jit-simple-jit.so
mips-mti-linux-gnu-gcc: warning: .../gdb/testsuite/outputs/gdb.base/jit-simple/jit-simple-jit.so: linker input file unused because linking not done
output is:
mips-mti-linux-gnu-gcc: warning: .../gdb/testsuite/outputs/gdb.base/jit-simple/jit-simple-jit.so: linker input file unused because linking not done

gdb compile failed, mips-mti-linux-gnu-gcc: warning: .../gdb/testsuite/outputs/gdb.base/jit-simple/jit-simple-jit.so: linker input file unused because linking not done
UNTESTED: gdb.base/killed-outside.exp: failed to prepare

and adjust the call to `prepare_for_testing' by removing a reference to
`options', which is not set in this test case but a stale value is
carried over from `gdb.base/jit-simple.exp' previously executed in a
full test suite run.

gdb/testsuite/
* gdb.base/killed-outside.exp: Remove $options from a call to
`prepare_for_testing'.

Add c-format tags to translatable strings with more than one argument-using formatting token.

* aout-adobe.c: Add missing c-format tags for translatable strings.
* aout-cris.c: Likewise.
* aoutx.h: Likewise.
* bfd.c: Likewise.
* binary.c: Likewise.
* cache.c: Likewise.
* coff-alpha.c: Likewise.
* coff-arm.c: Likewise.
* coff-i860.c: Likewise.
* coff-mcore.c: Likewise.
* coff-ppc.c: Likewise.
* coff-rs6000.c: Likewise.
* coff-sh.c: Likewise.
* coff-tic4x.c: Likewise.
* coff-tic54x.c: Likewise.
* coff-tic80.c: Likewise.
* coff64-rs6000.c: Likewise.
* coffcode.h: Likewise.
* coffgen.c: Likewise.
* cofflink.c: Likewise.
* coffswap.h: Likewise.
* cpu-arm.c: Likewise.
* dwarf2.c: Likewise.
* ecoff.c: Likewise.
* elf-attrs.c: Likewise.
* elf-eh-frame.c: Likewise.
* elf-ifunc.c: Likewise.
* elf-m10300.c: Likewise.
* elf-s390-common.c: Likewise.
* elf.c: Likewise.
* elf32-arc.c: Likewise.
* elf32-arm.c: Likewise.
* elf32-avr.c: Likewise.
* elf32-bfin.c: Likewise.
* elf32-cr16.c: Likewise.
* elf32-cr16c.c: Likewise.
* elf32-cris.c: Likewise.
* elf32-crx.c: Likewise.
* elf32-d10v.c: Likewise.
* elf32-d30v.c: Likewise.
* elf32-epiphany.c: Likewise.
* elf32-fr30.c: Likewise.
* elf32-frv.c: Likewise.
* elf32-gen.c: Likewise.
* elf32-hppa.c: Likewise.
* elf32-i370.c: Likewise.
* elf32-i386.c: Likewise.
* elf32-i960.c: Likewise.
* elf32-ip2k.c: Likewise.
* elf32-iq2000.c: Likewise.
* elf32-lm32.c: Likewise.
* elf32-m32c.c: Likewise.
* elf32-m32r.c: Likewise.
* elf32-m68hc11.c: Likewise.
* elf32-m68hc12.c: Likewise.
* elf32-m68hc1x.c: Likewise.
* elf32-m68k.c: Likewise.
* elf32-mcore.c: Likewise.
* elf32-mep.c: Likewise.
* elf32-metag.c: Likewise.
* elf32-microblaze.c: Likewise.
* elf32-moxie.c: Likewise.
* elf32-msp430.c: Likewise.
* elf32-mt.c: Likewise.
* elf32-nds32.c: Likewise.
* elf32-nios2.c: Likewise.
* elf32-or1k.c: Likewise.
* elf32-pj.c: Likewise.
* elf32-ppc.c: Likewise.
* elf32-rl78.c: Likewise.
* elf32-rx.c: Likewise.
* elf32-s390.c: Likewise.
* elf32-score.c: Likewise.
* elf32-score7.c: Likewise.
* elf32-sh-symbian.c: Likewise.
* elf32-sh.c: Likewise.
* elf32-sh64.c: Likewise.
* elf32-spu.c: Likewise.
* elf32-tic6x.c: Likewise.
* elf32-tilepro.c: Likewise.
* elf32-v850.c: Likewise.
* elf32-vax.c: Likewise.
* elf32-visium.c: Likewise.
* elf32-xgate.c: Likewise.
* elf32-xtensa.c: Likewise.
* elf64-alpha.c: Likewise.
* elf64-gen.c: Likewise.
* elf64-hppa.c: Likewise.
* elf64-ia64-vms.c: Likewise.
* elf64-mmix.c: Likewise.
* elf64-ppc.c: Likewise.
* elf64-s390.c: Likewise.
* elf64-sh64.c: Likewise.
* elf64-sparc.c: Likewise.
* elf64-x86-64.c: Likewise.
* elfcode.h: Likewise.
* elfcore.h: Likewise.
* elflink.c: Likewise.
* elfnn-aarch64.c: Likewise.
* elfnn-ia64.c: Likewise.
* elfxx-mips.c: Likewise.
* elfxx-sparc.c: Likewise.
* elfxx-tilegx.c: Likewise.
* ieee.c: Likewise.
* ihex.c: Likewise.
* libbfd.c: Likewise.
* linker.c: Likewise.
* m68klinux.c: Likewise.
* mach-o.c: Likewise.
* merge.c: Likewise.
* mmo.c: Likewise.
* oasys.c: Likewise.
* pdp11.c: Likewise.
* pe-mips.c: Likewise.
* peXXigen.c: Likewise.
* pei-x86_64.c: Likewise.
* peicode.h: Likewise.
* ppcboot.c: Likewise.
* reloc.c: Likewise.
* sparclinux.c: Likewise.
* srec.c: Likewise.
* stabs.c: Likewise.
* vms-alpha.c: Likewise.
* vms-lib.c: Likewise.
* xcofflink.c: Likewise.

Remove spurious whitespace introduced by previous delta.

Regen gdb/config.in

gdb/ChangeLog:
2016-10-19  Pedro Alves  <[email protected]>

* config.in: Regenerate.

[GAS][ARM]Generate unpredictable warning for pc used in data processing instructions with register-shifted register operand.

gas/

2016-10-19  Renlin Li  <[email protected]>

* config/tc-arm.c (encode_arm_shift): Generate unpredictable warning
for register-shifted register instructions.
* testsuite/gas/arm/shift-bad-pc.d: New.
* testsuite/gas/arm/shift-bad-pc.l: New.
* testsuite/gas/arm/shift-bad-pc.s: New.

Automatic date update in version.in

gdb: no longer define __STDC_CONSTANT_MACROS/__STDC_LIMIT_MACROS

My gnulib fix at:
  https://lists.gnu.org/archive/html/bug-gnulib/2015-11/msg00010.html
was merged upstream meanwhile and our gnulib copy now includes it.

As a concidence, Kevin was telling me today that these macros are
causing a build problem on FreeBSD:

common/common-defs.h:47:0: error: "__STDC_CONSTANT_MACROS" redefined [-Werror]
  #define __STDC_CONSTANT_MACROS 1
  /usr/include/sys/cdefs.h:408:0: note: this is the location of the previous definition
  #define __STDC_CONSTANT_MACROS
(and a similar error for __STDC_LIMIT_MACROS)

The problem seems to be that we should be defining these input macros
before including any system header, but, we're not.

So let's just revert e063da67902e ([C++] Define __STDC_CONSTANT_MACROS
/ __STDC_LIMIT_MACROS for stdint.h).  If this causes a problem
somewhere, we can re-define the macros higher up in the file, before
system headers are included.

gdb/ChangeLog:
2016-10-18  Pedro Alves  <[email protected]>

* common/common-defs.h (__STDC_CONSTANT_MACROS)
(__STDC_LIMIT_MACROS): Delete.

Update gnulib to current upstream master

I tried building gdb with G++ 4.7 and CXX="g++ -std=gnu+11", and that
tripped on a build error:

  In file included from build-gnulib/import/stdio.h:53:0,
   from /home/pedro/gdb/mygit/cxx-convertion/src/gdb/common/common-defs.h:31,
   from /home/pedro/gdb/mygit/cxx-convertion/src/gdb/defs.h:28,
   from /home/pedro/gdb/mygit/cxx-convertion/src/gdb/armobsd-tdep.c:20:
  build-gnulib/import/stddef.h:104:3: error: conflicting declaration ‘typedef union max_align_t max_align_t’
  In file included from build-gnulib/import/stddef.h:55:0,
   from build-gnulib/import/stdio.h:53,
   from /home/pedro/gdb/mygit/cxx-convertion/src/gdb/common/common-defs.h:31,
   from /home/pedro/gdb/mygit/cxx-convertion/src/gdb/defs.h:28,
   from /home/pedro/gdb/mygit/cxx-convertion/src/gdb/armobsd-tdep.c:20:
  /opt/gcc-4.7/lib/gcc/x86_64-unknown-linux-gnu/4.7.4/include/stddef.h:426:3: error: ‘max_align_t’ has a previous declaration as ‘typedef struct max_align_t max_align_t’

Updating gnulib to current master fixes it, since it brings in this
fix:

 http://lists.gnu.org/archive/html/bug-gnulib/2016-04/msg00000.html

Our last update was in August 2015.  This doesn't bring in much added
baggage, it's mostly bug fixes.  It pulled in the "limits-h" module as
automatic dependency, and given that looks potentially useful I added
it to the set of modules we require.

Tested on x86_64 Fedora 23, with g++ 4.7 and g++ 5.3, native and gdbserver.

gdb/ChangeLog:
2016-10-18  Pedro Alves  <[email protected]>

* gnulib/update-gnulib.sh (GNULIB_COMMIT_SHA1): Set to
2692e23a48e21f6daa029e8af9f1a143b7532f47.
* gnulib/configure, gnulib/config.in, gnulib/aclocal.m4:
Regenerate.
* gnulib/import/Makefile: Update.
* gnulib/import/alloca: Update.
* gnulib/import/basename-lgpl: Update.
* gnulib/import/canonicalize-lgpl: Update.
* gnulib/import/config: Update.
* gnulib/import/dirent: Update.
* gnulib/import/dirfd: Update.
* gnulib/import/dirname-lgpl: Update.
* gnulib/import/dirname.h: Update.
* gnulib/import/dosname.h: Update.
* gnulib/import/errno: Update.
* gnulib/import/extra/snippet/arg-nonnull.h: Update.
* gnulib/import/extra/snippet/c++defs.h: Update.
* gnulib/import/extra/snippet/warn-on-use.h: Update.
* gnulib/import/extra/update-copyright: Update.
* gnulib/import/flexmember.h: Update.
* gnulib/import/float+.h: Update.
* gnulib/import/float: Update.
* gnulib/import/float: Update.
* gnulib/import/fnmatch: Update.
* gnulib/import/fnmatch: Update.
* gnulib/import/fnmatch_loop: Update.
* gnulib/import/fpucw.h: Update.
* gnulib/import/frexp: Update.
* gnulib/import/frexpl: Update.
* gnulib/import/gettimeofday: Update.
* gnulib/import/hard-locale: Update.
* gnulib/import/hard-locale.h: Update.
* gnulib/import/inttypes: Update.
* gnulib/import/isnan: Update.
* gnulib/import/isnand-nolibm.h: Update.
* gnulib/import/isnand: Update.
* gnulib/import/isnanl-nolibm.h: Update.
* gnulib/import/isnanl: Update.
* gnulib/import/itold: Update.
* gnulib/import/limits: Update.
* gnulib/import/localcharset: Update.
* gnulib/import/localcharset.h: Update.
* gnulib/import/lstat: Update.
* gnulib/import/m4/00gnulib: Update.
* gnulib/import/m4/absolute-header: Update.
* gnulib/import/m4/alloca: Update.
* gnulib/import/m4/canonicalize: Update.
* gnulib/import/m4/codeset: Update.
* gnulib/import/m4/configmake: Update.
* gnulib/import/m4/dirent_h: Update.
* gnulib/import/m4/dirfd: Update.
* gnulib/import/m4/dirname: Update.
* gnulib/import/m4/double-slash-root: Update.
* gnulib/import/m4/eealloc: Update.
* gnulib/import/m4/errno_h: Update.
* gnulib/import/m4/exponentd: Update.
* gnulib/import/m4/exponentl: Update.
* gnulib/import/m4/extensions: Update.
* gnulib/import/m4/extern-inline: Update.
* gnulib/import/m4/fcntl-o: Update.
* gnulib/import/m4/flexmember: Update.
* gnulib/import/m4/float_h: Update.
* gnulib/import/m4/fnmatch: Update.
* gnulib/import/m4/fpieee: Update.
* gnulib/import/m4/frexp: Update.
* gnulib/import/m4/frexpl: Update.
* gnulib/import/m4/gettimeofday: Update.
* gnulib/import/m4/glibc21: Update.
* gnulib/import/m4/gnulib-cache: Update.
* gnulib/import/m4/gnulib-common: Update.
* gnulib/import/m4/gnulib-comp: Update.
* gnulib/import/m4/gnulib-tool: Update.
* gnulib/import/m4/hard-locale: Update.
* gnulib/import/m4/include_next: Update.
* gnulib/import/m4/inttypes-pri: Update.
* gnulib/import/m4/inttypes: Update.
* gnulib/import/m4/isnand: Update.
* gnulib/import/m4/isnanl: Update.
* gnulib/import/m4/largefile: Update.
* gnulib/import/m4/limits-h: Update.
* gnulib/import/m4/localcharset: Update.
* gnulib/import/m4/locale-fr: Update.
* gnulib/import/m4/locale-ja: Update.
* gnulib/import/m4/locale-zh: Update.
* gnulib/import/m4/longlong: Update.
* gnulib/import/m4/lstat: Update.
* gnulib/import/m4/malloc: Update.
* gnulib/import/m4/malloca: Update.
* gnulib/import/m4/math_h: Update.
* gnulib/import/m4/mbrtowc: Update.
* gnulib/import/m4/mbsinit: Update.
* gnulib/import/m4/mbsrtowcs: Update.
* gnulib/import/m4/mbstate_t: Update.
* gnulib/import/m4/memchr: Update.
* gnulib/import/m4/memmem: Update.
* gnulib/import/m4/mmap-anon: Update.
* gnulib/import/m4/multiarch: Update.
* gnulib/import/m4/nocrash: Update.
* gnulib/import/m4/off_t: Update.
* gnulib/import/m4/pathmax: Update.
* gnulib/import/m4/rawmemchr: Update.
* gnulib/import/m4/readlink: Update.
* gnulib/import/m4/rename: Update.
* gnulib/import/m4/rmdir: Update.
* gnulib/import/m4/signal_h: Update.
* gnulib/import/m4/ssize_t: Update.
* gnulib/import/m4/stat: Update.
* gnulib/import/m4/stdbool: Update.
* gnulib/import/m4/stddef_h: Update.
* gnulib/import/m4/stdint: Update.
* gnulib/import/m4/stdio_h: Update.
* gnulib/import/m4/stdlib_h: Update.
* gnulib/import/m4/strchrnul: Update.
* gnulib/import/m4/string_h: Update.
* gnulib/import/m4/strstr: Update.
* gnulib/import/m4/strtok_r: Update.
* gnulib/import/m4/sys_socket_h: Update.
* gnulib/import/m4/sys_stat_h: Update.
* gnulib/import/m4/sys_time_h: Update.
* gnulib/import/m4/sys_types_h: Update.
* gnulib/import/m4/time_h: Update.
* gnulib/import/m4/unistd_h: Update.
* gnulib/import/m4/warn-on-use: Update.
* gnulib/import/m4/wchar_h: Update.
* gnulib/import/m4/wchar_t: Update.
* gnulib/import/m4/wctype_h: Update.
* gnulib/import/m4/wint_t: Update.
* gnulib/import/malloc: Update.
* gnulib/import/malloca: Update.
* gnulib/import/malloca.h: Update.
* gnulib/import/math: Update.
* gnulib/import/math: Update.
* gnulib/import/mbrtowc: Update.
* gnulib/import/mbsinit: Update.
* gnulib/import/mbsrtowcs-impl.h: Update.
* gnulib/import/mbsrtowcs-state: Update.
* gnulib/import/mbsrtowcs: Update.
* gnulib/import/memchr: Update.
* gnulib/import/memmem: Update.
* gnulib/import/pathmax.h: Update.
* gnulib/import/rawmemchr: Update.
* gnulib/import/readlink: Update.
* gnulib/import/ref-add.sin: Update.
* gnulib/import/ref-del.sin: Update.
* gnulib/import/rename: Update.
* gnulib/import/rmdir: Update.
* gnulib/import/same-inode.h: Update.
* gnulib/import/signal: Update.
* gnulib/import/stat: Update.
* gnulib/import/stdbool: Update.
* gnulib/import/stddef: Update.
* gnulib/import/stdint: Update.
* gnulib/import/stdio: Update.
* gnulib/import/stdlib: Update.
* gnulib/import/str-two-way.h: Update.
* gnulib/import/strchrnul: Update.
* gnulib/import/streq.h: Update.
* gnulib/import/string: Update.
* gnulib/import/stripslash: Update.
* gnulib/import/strnlen1: Update.
* gnulib/import/strnlen1.h: Update.
* gnulib/import/strstr: Update.
* gnulib/import/strtok_r: Update.
* gnulib/import/sys_stat: Update.
* gnulib/import/sys_time: Update.
* gnulib/import/sys_types: Update.
* gnulib/import/time: Update.
* gnulib/import/unistd: Update.
* gnulib/import/unistd: Update.
* gnulib/import/verify.h: Update.
* gnulib/import/wchar: Update.
* gnulib/import/wctype: Update.
* gnulib/import/flexmember.h: New file.
* gnulib/import/hard-locale.c: New file.
* gnulib/import/hard-locale.h: New file.
* gnulib/import/limits.in.h: New file.
* gnulib/import/m4/flexmember.m4: New file.
* gnulib/import/m4/hard-locale.m4: New file.
* gnulib/import/m4/limits-h.m4: New file.

Check addr32flag instead of sizeflag for rip/eip

Since the address size prefix, 0x67, is ignored for MPX instructions in
64-bit mode, we should check addr32flag instead of sizeflag for rip/eip.

PR binutis/20699
* i386-dis.c (OP_E_memory): Check addr32flag instead of sizeflag
for rip/eip.

Remove the remaining SSE5 support

PR binutis/20704
* i386-dis.c (three_byte_table): Remove the remaining SSE5 support.

Introduce gdb::unique_ptr

Many make_cleanup uses in the code base are best eliminated by using a
"owning" smart pointer to manage ownership of the resource
automatically.

The question is _which_ smart pointer.

GDB currently supports building with a C++03 compiler.  We have
std::auto_ptr in C++03, but, as is collective wisdom by now, that's
too easy to misuse, and has therefore been deprecated in C++11 and
finally removed in C++17.

It'd be nice to be able to use std::unique_ptr instead, which is the
modern, safe std::auto_ptr replacement in C++11.

In addition to extra safety -- moving (i.e., transfer of ownership of
the managed pointer between smart pointers) must be explicit --
std::unique_ptr has (among others) one nice feature that std::auto_ptr
doesn't --- ability to specify a custom deleter as template parameter.
In gdb's context, that allows easily creating a smart pointer for
memory allocated with xmalloc -- the smart pointer then knows to
release with xfree instead of delete.  This is particularly
interesting when managing objects allocated in C libraries, and also,
for C++-fying parts of GDB that interact with other parts that still
return objects allocated with xmalloc.

Since std::unique_ptr's API is quite nice, and eventually we'd like to
move to C++11, this patch adds a C++03-compatible smart pointer that
exposes the subset of the std::unique_ptr API that we're interested
in.  An advantage is that whenever we start requiring C++11, we won't
have to learn a new API.  Meanwhile, this allows continuing to support
building with a C++03 compiler.

Since C++03 doesn't support rvalue references (boost gets close to
emulating them, but it's not fully transparent to user code), the
C++03 std::unique_ptr emulation here doesn't try hard to prevent
accidentally moving, which is where most of complication of a more
thorough emulation would be.  Instead, we rely on the fact that GDB
will be usually compiled with a C++11 compiler, and use the real
std::unique_ptr in that case to catch such accidental moves.  IOW, the
goal here is to allow code that would be correct using std::unique_ptr
to be equally correct in C++03 mode, and, just as efficient.

The C++03 version was originally based on GCC 7.0's std::auto_ptr and
then heavily customized to behave more like C++11's std::unique_ptr:

   - Support for custom (stateless) deleters.  (Support for stateful
     deleters could be added, if necessary.)

   - unique_ptr<T[]> partial specialization (auto_ptr<T> does not know
     to use delete[]).

   - Support for all of 'ptr != NULL', 'ptr == NULL' and 'if (ptr)'
     using the safe bool idiom to emulate C++11's explicit bool
     operator.

   - There's no nullptr in C++03, so this allows initialization and
     assignment from NULL instead (std::auto_ptr allows neither).

   - Variable names un-uglified (ie., no leading __ prefix everywhere).

   - Formatting made to follow GDB's coding conventions, including
     comment style.

   - Converting "move" constructors done differently in order to truly
     support:

      unique_ptr<Derived> func_returning_unique_ptr (.....);
      ...
      unique_ptr<Base> ptr = func_returning_unique_ptr (.....);

At this point, it no longer shares much at all with the original file,
but, that's the history.

See comments in the code to find out more.

I thought of putting the "emulation" / shim in the "std" namespace, so
that when we start requiring C++11 at some point, no actual changes to
users of the smart pointer throughout would be necessary.  Putting
things in the std namespace is technically undefined, however in
practice it doesn't cause any issue with any compiler.  However,
thinking that people might be confused with seeing std::unique_ptr and
thinking that we're actually requiring C++11 already, I put the new
types in the "gdb" namespace instead.

For managing xmalloc pointers, this adds a gdb::unique_xmalloc_ptr<T>
"specialization" with a custom xfree deleter.

No actual use of any smart pointer is introduced in this patch.
That'll be done in following patches.

Tested (along with the rest of the series) on:

 - NetBSD 5.1 (gcc70 on the compile farm), w/ gcc 4.1.3
 - x86-64 Fedora 23, gcc 5.3.1 (gnu++03)
 - x86-64 Fedora 23, and gcc 7.0 (gnu++14)

gdb/ChangeLog:
2016-10-18  Pedro Alves  <[email protected]>

* common/common-defs.h: Include "gdb_unique_ptr.h".
* common/gdb_unique_ptr.h: New.

Updated Danish translation for the BFD library.

sim: avr: move changelog entries to subdir

i386-tdep: Verify architecture before proceeding with `set/show mpx'

Make sure the architecture is `bfd_arch_i386' before handling the `set
mpx' and `show mpx' commands, avoiding the issue with `i386_mpx_enabled'
interpreting `gdbarch->tdep' according to the `struct gdbarch_tdep'
definition in i386-tdep.h while indeed in a multi-target configuration
it may have a different layout and cause GDB to crash or at least
misbehave.

gdb/
* i386-tdep.c (i386_mpx_info_bounds): Make sure the architecture
is `bfd_arch_i386' before proceeding.
(i386_mpx_set_bounds): Likewise.

tilegx-tdep: Correct aliasing errors in `tilegx_analyze_prologue'

Fix a load of aliasing build errors:

cc1plus: warnings being treated as errors
.../gdb/tilegx-tdep.c: In function 'CORE_ADDR tilegx_analyze_prologue(gdbarch*, CORE_ADDR, CORE_ADDR, tilegx_frame_cache*, frame_info*)':
.../gdb/tilegx-tdep.c:609: error: dereferencing pointer 'operands' does break strict-aliasing rules
.../gdb/tilegx-tdep.c:592: error: dereferencing pointer 'operands' does break strict-aliasing rules
.../gdb/tilegx-tdep.c:571: error: dereferencing pointer 'operands' does break strict-aliasing rules
[...]
.../gdb/tilegx-tdep.c:601: error: dereferencing pointer '<anonymous>' does break strict-aliasing rules
.../gdb/tilegx-tdep.c:601: note: initialized from here
cc1plus: error: dereferencing pointer 'operands' does break strict-aliasing rules
cc1plus: error: dereferencing pointer 'operands' does break strict-aliasing rules
.../gdb/tilegx-tdep.c:452: note: initialized from here
cc1plus: error: dereferencing pointer 'pretmp.896' does break strict-aliasing rules
cc1plus: note: initialized from here
cc1plus: error: dereferencing pointer 'pretmp.896' does break strict-aliasing rules
cc1plus: note: initialized from here
make[1]: *** [tilegx-tdep.o] Error 1

from an attempt to cast a `long long' pointer to an `int64_t' pointer,
which may not necessarily be compatible types.  Use the `long long' type
for the auxiliary variable then as this is the type of the structure
member referred.

gdb/
* tilegx-tdep.c (tilegx_analyze_prologue): Use the `long long'
type for `operands'.

AArch64/opcodes: Correct an `index' global shadowing error

Fix a commit 4df068de5214 ("Add support for SVE addressing modes") build
regression:

cc1: warnings being treated as errors
.../opcodes/aarch64-dis.c: In function 'aarch64_ext_sve_addr_rr_lsl':
.../opcodes/aarch64-dis.c:1324: error: declaration of 'index' shadows a global declaration
/usr/include/string.h:303: error: shadowed declaration is here
make[3]: *** [aarch64-dis.lo] Error 1

in a way following commit 91d6fa6a035c ("Add -Wshadow to the gcc command
line options used when compiling the binutils.").

opcodes/
* aarch64-dis.c (aarch64_ext_sve_addr_rr_lsl): Rename `index'
local variable to `index_regno'.

NDS32/BFD: Correct an aliasing error in `nds32_elf_check_relocs'

Fix an aliasing build error:

cc1: warnings being treated as errors
.../bfd/elf32-nds32.c: In function 'nds32_elf_check_relocs':
.../bfd/elf32-nds32.c:6644: warning: dereferencing type-punned pointer will break strict-aliasing rules
make[3]: *** [elf32-nds32.lo] Error 1

in a way following commit 6edfbbad0864 ("Fix up gcc4.1 aliasing
warnings"), <https://sourceware.org/ml/binutils/2005-10/msg00071.html>.

bfd/
* elf32-nds32.c (nds32_elf_check_relocs): Avoid aliasing warning
from GCC.

ARM/BFD: Correct an `index' global shadowing error

Fix a commit 5025eb7c0d87 ("Delete relocations associatesd with deleted
exidx entries.") build regression:

cc1: warnings being treated as errors
.../bfd/elf32-arm.c: In function 'elf32_arm_update_relocs':
.../bfd/elf32-arm.c:14951: warning: declaration of 'index' shadows a global declaration
/usr/include/string.h:304: warning: shadowed declaration is here
make[3]: *** [elf32-arm.lo] Error 1

in a way following commit 91d6fa6a035c ("Add -Wshadow to the gcc command
line options used when compiling the binutils.").

bfd/
* elf32-arm.c (elf32_arm_update_relocs): Rename `index' local
variable to `reloc_index'.

Automatic date update in version.in

Fix duplicate test message in mi-trace-save.exp

gdb/testsuite/ChangeLog:

* gdb.trace/mi-trace-save.exp (test_trace_save_wrong_num_args):
Change test message.

Fix comment in mi-trace-save.exp

This fixes a comment I forgot to update in the previous patch.

gdb/testsuite/ChangeLog:

* gdb.trace/mi-trace-save.exp (test_trace_save_wrong_num_args):
Update comment.

Fix -trace-save crash when argument is missing

-trace-save doesn't check whether an argument is passed, leading to a
segfault if you pass nothing.

I added a small test, which only tests the error conditions of
-trace-save.

gdb/ChangeLog:

* mi/mi-main.c (mi_cmd_trace_save): Check if argument is present
before using it.

gdb/testsuite/ChangeLog:

* gdb.trace/mi-trace-save.exp: New file.

gdb: Fix phony iconv build

Cross building gdb for mingw32 on Fedora 23 fails with:

  x86_64-w64-mingw32-g++ -g -O2  [...]  ../../src/gdb/charset.c
  In file included from ../../src/gdb/charset.c:21:0:
  ../../src/gdb/charset.h:134:3: error: 'iconv_t' does not name a type
     iconv_t m_desc;
     ^
  ../../src/gdb/charset.c: In constructor 'wchar_iterator::wchar_iterator(const gdb_byte*, size_t, const char*, size_t)':
  ../../src/gdb/charset.c:600:3: error: 'm_desc' was not declared in this scope
     m_desc = iconv_open (INTERMEDIATE_ENCODING, charset);
     ^
  ../../src/gdb/charset.c: In destructor 'wchar_iterator::~wchar_iterator()':
  ../../src/gdb/charset.c:607:7: error: 'm_desc' was not declared in this scope
     if (m_desc != (iconv_t) -1)
 ^
  ../../src/gdb/charset.c: In member function 'int wchar_iterator::iterate(wchar_iterate_result*, gdb_wchar_t**, const gdb_byte**, size_t*)':
  ../../src/gdb/charset.c:633:25: error: 'm_desc' was not declared in this scope
 size_t r = iconv (m_desc, &inptr, &m_bytes, &outptr, &out_avail);
   ^

This is a regression caused by commit cda6c55bd399 (Turn wchar
iterator into a class).  The problem is that iconv_t is now exposed in
gdb/charset.h, while before it was only used in gdb/charset.c.

gdb/charset.c, under #ifdef PHONY_ICONV, does:

 #undef iconv_t
 #define iconv_t int

So it seems the simplest is to use 'int' in the header file too.

gdb/ChangeLog:
2016-10-17  Pedro Alves  <[email protected]>

* charset.h (class wchar_iterator) [PHONY_ICONV] <m_desc>: Use
'int' as type.

Fixed matching in newly added test.

gas/ChangeLog:

    2016-10-17  Cupertino Miranda  <[email protected]>
* testsuite/arc/dis-inv.d: Fixed target match.

Display unknown notes.  Decode NT_GNU_HWCAP notes.

* readelf.c (apply_relocations): Fail if the symbol table section
linked to by the reloc section does not have either the SHT_SYMTAB
or SHT_DYNSYM type.
(print_gnu_note): Decode the contents of NT_GNU_HWCAP notes.
Print the contents of unknown note types.
(process_note): Add the file and section to the parameter list.
Use print_symbol to display the note name.
Display the contents of unknown note types.
(process_corefile_note_segment): Rename to process_notes_at.
Add section parameter.  Apply relocations to the notes when
loading from a section.  Display section name when processing
notes in a section.
* testsuite/binutils-all/readelf.n: Update expected output.

Removed pseudo invalid instructions opcodes.

The disassember was generating invXXX instructions for cases when in reality we
had llockd or scondd instrutions.

opcodes/ChangeLog:

    Cupertino Miranda  <[email protected]>
arc-tbl.h: Removed any "inv.+" instructions from the table.

gas/ChangeLog:

    Cupertino Miranda  <[email protected]>
        testsuite/arc/dis-inv.s: Test to validate patch.
        testsuite/arc/dis-inv.d: Likewise.

Update list of ELF machine numbers.

include * elf/common.h (DT_SYMTAB_SHNDX): Define.
(EM_CLOUDSHIELD, EM_COREA_1ST, EM_COREA_2ND, EM_OPEN8): Define.
(EM_VIDEOCORE5, EM_56800EX, EM_BA1, EM_BA2, EM_XCORE): Define.
(EM_MCHP_PIC, EM_KM32, EM_KMX32, EM_KMX16, EM_KMX8): Define.
(EM_KVARC, EM_CDP, EM_COGE, EM_COOL, EM_NORC): Define.
(EM_CSR_KALIMBA, EM_Z80, EM_AMDGPU, EM_RISCV): Define.
(ELFOSABI_OPENVOS): Define.
(GRP_MASKOS, GRP_MASKPROC): Define.

binutils * readelf.c (get_dynamic_type): Add DT_SYMTAB_SHNDX.
(get_machine_type): Add EM_CLOUDSHIELD, EM_COREA_1ST,
EM_COREA_2ND, EM_OPEN8, EM_VIDEOCORE5, EM_56800EX, EM_BA1, EM_BA2,
EM_XCORE, EM_MCHP_PIC, EM_KM32, EM_KMX32, EM_KMX16, EM_KMX8,
EM_KVARC, EM_CDP, EM_COGE, EM_COOL, EM_NORC, EM_CSR_KALIMBA,
EM_Z80, EM_AMDGPU, EM_RISCV.
(get_osabi_name): Add ELFOSABI_CLOUDABI and ELFOSABI_OPENVS.
(get_group_flags): Update to handle flags in the
GRP_MASKOS and GRP_MASKPROC ranges.

Sync libiberty sources with gcc mainline.

2016-09-19  Andrew Stubbs  <[email protected]>

* pex-win32.c (argv_to_cmdline): Quote zero-length parameters.
* testsuite/test-pexecute.c (main): Insert check for zero-length parameters.

2016-09-10  Mark Wielaard  <[email protected]>

* cp-demangle.c (d_substitution): Change struct demangle_component
variable name from c to dc.

2016-08-12  Marek Polacek  <[email protected]>

PR c/7652
* cp-demangle.c (d_print_mod): Add FALLTHRU.

2016-08-04  Marcel B?hme  <[email protected]>

PR c++/71696
* cplus-dem.c: Prevent infinite recursion when there is a cycle
in the referencing of remembered mangled types.
(work_stuff): New stack to keep track of the remembered mangled
types that are currently being processed.
(push_processed_type): New method to push currently processed
remembered type onto the stack.
(pop_processed_type): New method to pop currently processed
remembered type from the stack.
(work_stuff_copy_to_from): Copy values of new variables.
(delete_non_B_K_work_stuff): Free stack memory.
(demangle_args): Push/Pop currently processed remembered type.
(do_type): Do not demangle a cyclic reference and push/pop
referenced remembered type.

Automatic date update in version.in

Automatic date update in version.in

Regenerate spu_ovl.o_c

No real changes here.  This is just so that we match current spu_ovl.o
with .shstrtab moved.

* emultempl/spu_ovl.o_c: Regenerate.

Automatic date update in version.in

FINAL/OVERRIDE: Define to empty on g++ < 4.7

final/override were only implemented in g++ 4.7.

include/ChangeLog
2016-10-14  Pedro Alves  <[email protected]>

* ansidecl.h [__cplusplus >= 201103 && GCC_VERSION < 4007] (FINAL,
OVERRIDE): Define as empty.
[__cplusplus < 201103 && GCC_VERSION < 4007] (FINAL): Define as
__final.
[__cplusplus < 201103 && GCC_VERSION >= 4007] (OVERRIDE): Define as
empty.

Move OVERRIDE/FINAL from gcc/coretypes.h to include/ansidecl.h

So that GDB and other projects that share the top level can use them.

Bootstrapped with all default languages + jit on x86-64 Fedora 23.

gcc/ChangeLog:
2016-10-14  Pedro Alves  <[email protected]>

* coretypes.h (OVERRIDE, FINAL): Delete, moved to
include/ansidecl.h.

include/ChangeLog:
2016-10-14  Pedro Alves  <[email protected]>

* ansidecl.h (GCC_FINAL): Delete.
(OVERRIDE, FINAL): New, moved from gcc/coretypes.h.

Fix typos in trace commands doc

gdb/doc/ChangeLog:

* gdb.texinfo (Using Trace Files): Fix typo.
(GDB/MI Tracepoint Commands): Likewise.

Document -ctf switch of -trace-save

The -trace-save MI command supports saving the trace in the CTF format,
as its CLI counterpart, but it's not documented.

gdb/doc/ChangeLog:

* gdb.texinfo (GDB/MI Tracepoint Commands): Document -ctf switch
of -trace-save.

Fix leftover remote test failures from PR binutils/19020

As pointed out in PR binutils/19020, those tests were still failing when doing
remote testing. This is because the binary files weren't being copied over to
the remote host for testing.

FAIL: binutils-all/pr19020a
FAIL: binutils-all/pr19020b

This small patch fixes this up to make things pass.

binutils/ChangeLog:
2016-10-14  Luis Machado  <[email protected]>

* testsuite/lib/utils-lib.exp (run_dump_test): Call remote_download
to copy file to remote host.

Fix set sysroot command on AIX

set sysroot command on AIX has no effect if a program depends on shared
library archives (.a).  Fixed by using solib_find and solib_bfd_fopen
instead of gdb_bfd_open in solib_aix_bfd_open.

gdb/
2016-10-14  Sangamesh Mallayya  <[email protected]>
    Ulrich Weigand  <[email protected]>

* solib-aix.c (solib_aix_bfd_open): Call solib_find so that sysroot
path is set properly if program has a dependency on .a archive and
sysroot is set via set sysroot command.

Signed-off-by: Ulrich Weigand <[email protected]>

Add separate debug info file section

* scripttempl/DWARF.sc: Add .debug_addr.

[ARC] Disassembler: fix LIMM detection for short instructions.

The ARC (short) instructions are using a special register number to
indicate is the instruction uses a long immediate (LIMM).  In the case
of short instruction, this LIMM indicator depends on the ISA version
used. Thus, for ARCv1 processors, the LIMM indicator is 0x3E, the same
value used in "long" instructions.  However, for the ARCv2 processors,
this LIMM indicator is 0x1E.

This patch fixes the LIMM detection for ARCv1 ISA and adds two tests.

gas/
2016-10-13  Claudiu Zissulescu  <[email protected]>

* testsuite/gas/arc/shortlimm_a7.d: New file.
* testsuite/gas/arc/shortlimm_a7.s: Likewise.
* testsuite/gas/arc/shortlimm_hs.d: Likewise.
* testsuite/gas/arc/shortlimm_hs.s: Likewise.

include/
2016-10-13  Claudiu Zissulescu  <[email protected]>

* opcode/arc.h (ARC_OPCODE_ARCV2): New define.

opcodes/
2016-10-13  Claudiu Zissulescu  <[email protected]>

        * arc-dis.c (find_format_from_table): Discriminate LIMM indicator
usage on ISA basis.

btrace: remove leftover comment

Remove a leftover comment on a function that was removed with commit
0568462bbf0f666d5bf9f720e16147da71eec46a.

Include strings.h where available

gdb/ChangeLog

2016-10-14  Eli Zaretskii  <[email protected]>

* common/common-defs.h [HAVE_STRINGS_H]: Include strings.h if
available, to get prototypes of 'strcasecmp' and 'strncasecmp'.

Automatic date update in version.in

ARI: Remove true/false checks

These don't make sense with C++.

gdb/ChangeLog:
2016-10-13  Pedro Alves  <[email protected]>

* contrib/ari/gdb_ari.sh (boolean): Suggest bool instead.
(false, true): Remove checks.

Share proc get_var_address

This patch moves proc get_var_address into lib/gdb.exp, and remove the
duplicate copy in gdb.base/*.exp files.

gdb/testsuite:

2016-10-13  Yao Qi  <[email protected]>

* gdb.base/code_elim.exp (get_var_address): Remove.
* gdb.base/relocate.exp: Likewise.
* gdb.base/shreloc.exp: Likewise.
* lib/gdb.exp (get_var_address): New.

Skip testing structures with floating points

This patch skips some tests related to floating point in structs.exp
if gdb_skip_float_test return false.

gdb/testsuite:

2016-10-13  Yao Qi  <[email protected]>

* gdb.base/structs.exp: Invoke gdb_skip_float_test, and do
floating point tests if $skip_float_test is false.

Use std::string in macho_symfile_read_all_oso

This changes macho_symfile_read_all_oso to use std::string.  This
avoids a cleanup.

2016-10-12  Tom Tromey  <[email protected]>

* machoread.c (macho_symfile_read_all_oso): Use std::string.

Remove unnecessary null_cleanup

This patch removes an unnecessary null_cleanup.

2016-10-12  Tom Tromey  <[email protected]>

* tracepoint.c (trace_dump_command): Remove unnecessary
null_cleanup.

Turn wchar iterator into a class

This changes wchar_iterator from charset.c into a real C++ class, then
updates the users to use the class.  This lets us remove some cleanups
in favor of the class' destructor.

2016-10-12  Tom Tromey  <[email protected]>

* valprint.c (generic_emit_char, count_next_character)
(generic_printstr): Update.
* charset.c (struct wchar_iterator): Move to charset.h.
(wchar_iterator::wchar_iterator): Rename from
make_wchar_iterator, turn into a constructor.
(wchar_iterator::~wchar_iterator): Rename from
do_cleanup_iterator, turn into a destructor.
(make_cleanup_wchar_iterator): Remove.
(wchar_iterator::iterate): Rename from wchar_iterate.  Remove
"iter" argument.  Update.
* charset.h: Include <vector>.
(class wchar_iterator): New class, from old struct
wchar_iterator.
(make_wchar_iterator, make_cleanup_wchar_iterator): Don't
declare.

Change selttest.c to use use std::vector

This patch changes selftest.c to use std::vector rather than VEC.
I think this is a small net plus.

2016-10-12  Tom Tromey  <[email protected]>

* selftest.c: Include <vector>, not "vec.h".
(self_test_function_ptr): Remove.
(tests): Now a std::vector.
(register_self_test, run_self_tests): Update.

Convert tid_range_parser and get_number_or_range to classes

This converts tid_range_parser and get_number_or_range to be classes.
The various tid_range_parser_* and get_number_or_range_* functions
become methods on the respective classes.  Then it updates the users
to follow.

The rationale for the change is that this provides better
encapsulation.  For example, this forced me to think of a better
interface between tid_range_parser and get_number_or_range, since the
former peeked into the latter's internals a bit too much.  That ended
up resulting mostly in these two not-just-straight-1-1 changes:

  void
 -tid_range_parser_skip (struct tid_range_parser *parser)
 +tid_range_parser::skip_range ()
  {
 ...

 -  tid_range_parser_init (parser, parser->range_parser.end_ptr,
 -  parser->default_inferior);
 +  m_range_parser.skip_range ();
 +  init (m_range_parser.string (), m_default_inferior);
  }

 and:

    /* If we successfully parsed a thread number or finished parsing a
       thread range, switch back to assuming the next TID is
       inferior-qualified.  */
 -  if (parser->range_parser.end_ptr == NULL
 -      || parser->range_parser.string == parser->range_parser.end_ptr)
 +  if (!m_range_parser.in_range ())
      {

For the same reason (encapsulation), this moves the enum
tid_range_state definition to within the tid_parser class's scope,
since that is private implementation detail.

While at it, switch to use "bool" for booleans.

gdb/ChangeLog:
2016-10-13  Pedro Alves  <[email protected]>
    Tom Tromey  <[email protected]>

* tid-parse.h (tid_range_parser): New class.
(enum tid_range_state): Move into tid_range_parser's scope.
Remove TID_RANGE_ prefix from all values.
(tid_range_parser_get_tid, tid_range_parser_get_tid_range)
(tid_range_parser_star_range, tid_range_parser_finished)
(tid_range_parser_skip, tid_range_parser_qualified): Don't
declare.
(tid_is_in_list): Update comment.
* tid-parse.c (tid_range_parser::tid_range_parser): New.
(init, finished, get_string, skip, tid_is_qualified)
(get_tid_or_range, get_tid_range, get_tid, star_range): Rename;
turn into methods.
(tid_is_in_list): Adjust.
* cli/cli-utils.h (number_or_range_parser): New class.
(init_number_or_range, get_number_or_range)
(number_range_setup_range): Don't declare.
* cli/cli-utils.c
(number_or_range_parser::number_or_range_parser): New.
(init_number_or_range, get_number_or_range)
(number_range_setup_range): Rename; turn into methods.
(number_is_in_list): Adjust.
* breakpoint.c (map_breakpoint_numbers): Adjust.  Use bool.
(trace_pass_command, get_tracepoint_by_number): Adjust.
* breakpoint.h (get_tracepoint_by_number): Adjust.
* inferior.c (detach_inferior_command, kill_inferior_command)
(remove_inferior_command): Adjust.
* linespec.c (decode_line_2): Adjust.
* memattr.c (mem_enable_command, mem_disable_command)
(mem_delete_command): Adjust.
* printcmd.c (map_display_numbers): Adjust.
* reverse.c (delete_bookmark_command, bookmarks_info): Adjust.
* thread.c (thread_apply_command): Adjust.

Automatic date update in version.in

Fixup gdb.python/py-value.exp for bare-metal aarch64-elf

I noticed that testing aarch64-elf gdb with a physical board
ran into issues with gdb.python/py-value.exp. Further investigation showed
that we were actually trying to dereference a NULL pointer (argv) when trying
to access argv[0].

Being bare-metal, argv is not guaranteed to be valid. So we need to make sure
argv is sane before accessing argv[0].

The following patch fixes up the test program to check for a NULL argv and also
improves the testcase a bit so it doesn't have to work with a hardcoded argc
value.

Regression-tested on x86-64 Ubuntu 16.04.

gdb/testsuite/ChangeLog:

2016-10-12  Luis Machado  <[email protected]>

* gdb.python/py-value.c (main): Check if argv is NULL before using it.
* gdb.python/py-value.exp (test_value_in_inferior): Don't use hardcoded
argc values.
Add 1 to argc so we guarantee distinct initial/modified argc values.

arc: Add support for Newlib

Add support for Newlib as an OS/ABI.  The only thing that is specific to it
relatively to "generic" baremetal target is location of PC register in jump
buffer for longjump support.

Sniffer uses .ivt section to decide if ELF file is for ARC Newlib or not.

gdb/ChangeLog:

* arc-newlib-tdep.c: New file.
* configure.tgt: Add newlib support for ARC.

arc: Add evaluation of long jump targets

Standard get_longjmp_target implementation, similar to what is in arm-tdep.c.
Actual value of jb_pc should be set in init_osabi methods of particular OS/ABI
implementations.

gdb/ChangeLog:

* arc-tdep.h (struct gdbarch_tdep) <jb_pc>: New field.
* arc-tdep.c (arc_get_longjmp_target): New function.
(arc_gdbarch_init): Set get_longjmp_target if jb_pc is non-negative.
(arc_dump_tdep): Print jb_pc.

arc: Add a gdbarch_tdep structure

Add target-specific structure gdbarch_tdep for ARC.

gdb/ChangeLog:

* arc-tdep.h (struct gdbarch_tdep): New.
* arc-tdep.c (arc_gdbarch_init): Allocate gdbarch_tdep.

[AArch64] Track FP registers in prologue analyzer

We don't track FP registers in aarch64 prologue analyzer, so this causes
an internal error when FP registers are saved by "stp" instruction in
prologue (stp d8, d9, [sp,#128]),

 tbreak _Unwind_RaiseException^M
 aarch64-tdep.c:335: internal-error: CORE_ADDR aarch64_analyze_prologue(gdbarch*, CORE_ADDR, CORE_ADDR, aarch64_prologue_cache*): Assertion `inst.operands[0].type == AARCH64_OPND_Rt' failed.^M
 A problem internal to GDB has been detected,

This patch teaches GDB to track FP registers (D registers) in prologue
analyzer.

gdb:

2016-10-12  Yao Qi  <[email protected]>

PR tdep/20682
* aarch64-tdep.c: Replace 32 with AARCH64_D_REGISTER_COUNT.
(aarch64_analyze_prologue): Extend array 'regs' for D registers.
Assert that operand 0 and 1 can be X or D registers.  Update
register number for D registers.  Update registers in frame
cache.
* aarch64-tdep.h (AARCH64_D_REGISTER_COUNT): New macro.

Automatic date update in version.in

missing changelog entry