gdb/remote.c: Eliminate target_buf/target_buf_size hack

This finally eliminates an old hack left in place when tracepoint RSP
support was migrated from tracepoint.c to remote.c, back in
35b1e5cca081 ("Make tracepoint operations go through target vector.")
over 7 years ago.

Tested on x86_64 GNU/Linux.

gdb/ChangeLog:
2017-09-29  Pedro Alves  <palves@redhat.com>

* remote.c (target_buf, target_buf_size): Delete.
(remote_get_noisy_reply): Remove buf_p and sizeof_buf parameters.
Use the connection's packet buffer instead.
All callers adjusted.
(_initialize_remote): Remove references to target_buf and
target_buf_size.

gdbserver/libthread_db: Don't ignore memory reading failures

If we had this in place before, then the regression fixed by the
previous commit would have been been visible is all test runs.  E.g.:

  Running src/gdb/testsuite/gdb.threads/multi-create-ns-info-thr.exp ...
  FAIL: gdb.threads/multi-create-ns-info-thr.exp: continue to breakpoint 6

Debugging manually we'd see this:
  gdbserver: Cannot get thread handle for LWP 1467: generic error

Instead of:
  gdbserver: PID mismatch!  Expected 27472, got 27471

which is misleading - gdbserver didn't 27471, that was stale stack
data from previous function invocations.

gdb/gdbserver/ChangeLog:
2017-09-29  Pedro Alves  <palves@redhat.com>

* proc-service.c (ps_pdread): Return PS_ERR if reading memory
fails.

Fix gdbserver regression exposed by gdb.threads/multi-create-ns-info-thr.exp

Commit 8629910955a7 ("Add thread_db_notice_clone to gdbserver")
introduced calls into libthread_db without making sure that the
current thread is pointing to a know-stopped thread.  This resulted in
sometimes thread_db_notice_clone failing->find_one_thread failing like
this, as seen when running gdb.threads/multi-create-ns-info-thr.exp:

~~~
  Thread <6> executing
  Thread <7> executing
  gdbserver: PID mismatch!  Expected 27472, got 27471
  gdbserver: Cannot find thread after clone.

  Thread <1000> executing
  Thread <1001> executing
~~~

Things go south from here and sometimes that ends up resulting in
gdbserver crashing and the test failing.

gdb/gdbserver/ChangeLog:
2017-09-29  Pedro Alves  <palves@redhat.com>

* linux-low.c (handle_extended_wait): Pass parent thread instead
of process to thread_db_notice_clone.
* linux-low.h (thread_db_notice_clone): Replace parent process
parameter with parent thread parameter.
* thread-db.c (find_one_thread): Add comment.
(thread_db_notice_clone): Replace parent process parameter with
parent thread parameter.  Temporarily switch to the parent thread.

Fail when string merge can't alloc memory

I was looking at Debian bug #874674 again today, and think I might
have spotted the problem.  It appears that merge.c tries to cope with
memory allocation failures in some circumstances, but doesn't quite
manage to get everything right.  This patch will make ld report memory
allocation failures instead of silently not merging strings.

* merge.c (merge_strings): Return FALSE on malloc failure.
(_bfd_merge_sections): Return failures from record_section and
merge_strings.

Automatic date update in version.in

Move utils-selftests.c -> gdb/unittests/

This file was only under gdb/ currently because it predates the
gdb/unittests/ directory.

gdb/ChangeLog:
2017-09-28  Pedro Alves  <palves@redhat.com>

* Makefile.in (SUBDIR_UNITTESTS_SRCS): Add
unittests/common-utils-selftests.c.
(SUBDIR_UNITTESTS_OBS): Add common-utils-selftests.o.
(COMMON_OBS): Remove utils-selftests.o.
* utils-selftests.c: Move to ...
* unittests/common-utils-selftests.c: ... here and rename self
test to "string_printf".

(open_and_init_dwp_file): Protect against some segvs

This is a "tiny patch", no assignment required.

2017-09-28  Alexander Shaposhnikov <alexander.v.shaposhnikov@gmail.com>

* dwarf2read.c (open_and_init_dwp_file): Protect against dwp_file
having NULL cus or tus.

Skip the PR 14918 linker test for ARM targets.

 I am applying a patch that has been lying around in the Fedora
  binutils sources for a while.  It skips the PR14918 linker test for
  ARM based targets.  This test checks that libgcc is not included in a
  link of an empty executable.  This works for most targets, but on the
  ARM the crt1.o startup code calls __libc_csu_init which is in
  /usr/lib/libc_nonshared.a(elf-init.oS).  This in turn needs
  __aeabi_unwind_cpp_pr0@@GCC_3.5 which is provided by libgcc_s.so.1,
  and so the test fails.

PR22220, BFD linker wrongly marks symbols as PREVAILING_DEF_IRONLY

non_ir_ref_dynamic wasn't being set in the case where we have a
versioned dynamic symbol definition with a non-versioned matching IR
symbol.

bfd/
PR 22220
* elflink.c (_bfd_elf_merge_symbol): Set non_ir_ref_dynamic in
a case where plugin_notice isn't called.
ld/
* testsuite/ld-plugin/pr22220.h,
* testsuite/ld-plugin/pr22220lib.cc,
* testsuite/ld-plugin/pr22220lib.ver,
* testsuite/ld-plugin/pr22220main.cc: New test.
* testsuite/ld-plugin/lto.exp: Run it.

Automatic date update in version.in

Complete tdep move to convert_typed_floating

Many tdep files need to perform conversions between two floating-point
types, usually when accessing FP registers.  Most targets now use the
convert_typed_floating helper routine to do so.  However, a small number
still use the old method of converting via a DOUBLEST.  Since we want
to get rid of DOUBLEST, these targets need to be moved to the new
method as well.

The main obstacle is that for convert_typed_floating we need an actual
*type*, not just a floatformat.

In arm-tdep.c, this is very straightforward, since there is already a
type using the ARM extended floatformat.

For sh-tdep.c and sh64-tdep.c, no such type already exists, so I've
added one to the gdbarch_tdep struct as done on other targets.

gdb/ChangeLog
2017-09-27  Ulrich Weigand  <uweigand@de.ibm.com>

* arm-tdep.c: (convert_from_extended): Remove.
(convert_to_extended): Likewise.
(arm_extract_return_value): Use convert_typed_floating.
(arm_store_return_value): Likewise.

* sh-tdep.h (struct gdbarch_tdep): Add sh_littlebyte_bigword_type.
* sh-tdep.c: Do not include "floatformat.h".
(sh_littlebyte_bigword_type): New function.
(sh_register_convert_to_virtual): Use convert_typed_floating.
(sh_register_convert_to_raw): Likewise.
* sh64-tdep.c: (struct gdbarch_tdep): Add sh_littlebyte_bigword_type.
(sh64_littlebyte_bigword_type): New function.
(sh64_extract_return_value): Use convert_typed_floating.
(sh64_register_convert_to_virtual): Likewise.
(sh64_register_convert_to_raw): Likewise.

Simplify floatformat_from_type

For historical reasons, the TYPE_FLOATFORMAT element is still set to hold
an array of two floatformat structs, one for big-endian and the other for
little-endian.  When accessing the element via floatformat_from_type,
the code would check the type's byte order and return the appropriate
floatformat.

However, these days this is quite unnecessary, since the type's byte order
is already known at the time the type is allocated and the floatformat is
installed into TYPE_FLOATFORMAT.  Therefore, we can just install the correct
version here.

Also, moves the (now trivially simple) floatformat_from_type accessor to
gdbtypes.{c,h}, since it doesn't really need to be in doublest.c now.

gdb/ChangeLog
2017-09-27  Ulrich Weigand  <uweigand@de.ibm.com>

* doublest.h (floatformat_from_type): Move to gdbtypes.h.
* doublest.c (floatformat_from_type): Move to gdbtypes.c.

* gdbtypes.h (union type_specific): Make field floatformat hold
just a single struct floatformat, not an array.
(floatformat_from_type): Move here.
* gdbtypes.c (floatformat_from_type): Move here.  Update to
changed TYPE_FLOATFORMAT definition.
(verify_floatformat): Update to changed TYPE_FLOATFORMAT.
(recursive_dump_type): Likewise.
(init_float_type): Install correct floatformat for byte order.
(arch_float_type): Likewise.

Make init_type/arch_type take a size in bits

This changes the interfaces to init_type and arch_type to take the
type length in bits as input (instead of as bytes).  The routines
assert that the length is a multiple of TARGET_CHAR_BIT.

For consistency, arch_flags_type is changed likewise, so that now
all type creation interfaces always use length in bits.

All callers are updated in the straightforward manner.

The assert actually found a bug in read_range_type, where the
init_integer_type routine was called with a wrong argument (probably
a bug introduced with the conversion to use init_integer_type).

gdb/ChangeLog
2017-09-27  Ulrich Weigand  <uweigand@de.ibm.com>

* gdbtypes.c (init_type): Change incoming argument from
length-in-bytes to length-in-bits.  Assert length is a
multiple of TARGET_CHAR_BITS.
(arch_type, arch_flags_type): Likewise.
(init_integer_type): Update call to init_type.
(init_character_type): Likewise.
(init_boolean_type): Likewise.
(init_float_type): Likewise.
(init_decfloat_type): Likewise.
(init_complex_type): Likewise.
(init_pointer_type): Likewise.
(objfile_type): Likewise.
(arch_integer_type): Update call to arch_type.
(arch_character_type): Likewise.
(arch_boolean_type): Likewise.
(arch_float_type): Likewise.
(arch_decfloat_type): Likewise.
(arch_complex_type): Likewise.
(arch_pointer_type): Likewise.
(gdbtypes_post_init): Likewise.

* dwarf2read.c (dwarf2_init_float_type): Update call to init_type.
(read_base_type): Likewise.
* mdebugread.c (basic_type): Likewise.
* stabsread.c (dbx_init_float_type): Likewise.
(rs6000_builtin_type): Likewise.
(read_range_type): Likewise.  Also, fix call to init_integer_type
with erroneous length argument.

* ada-lang.c (ada_language_arch_info): Update call to arch_type.
* d-lang.c (build_d_types): Likewise.
* f-lang.c (build_fortran_types): Likewise.
* go-lang.c (build_go_types): Likewise.
* opencl-lang.c (build_opencl_types): Likewise.
* jit.c (finalize_symtab): Likewise.
* gnu-v3-abi.c (build_gdb_vtable_type): Likewise.
(build_std_type_info_type): Likewise.
* target-descriptions.c (tdesc_gdb_type): Likewise.  Also,
update call to arch_flags_type.

* linux-tdep.c (linux_get_siginfo_type_with_fields): Update call to
arch_type.
* fbsd-tdep.c (fbsd_get_siginfo_type): Likewise.
* windows-tdep.c (windows_get_tlb_type): Likewise.

* avr-tdep.c (avr_gdbarch_init): Update call to arch_type.
* ft32-tdep.c (ft32_gdbarch_init): Likewise.
* m32c-tdep.c (make_types): Likewise.
* rl78-tdep.c (rl78_gdbarch_init): Likewise.
(rl78_psw_type): Update call to arch_flags_type.
* m68k-tdep.c (m68k_ps_type): Update call to arch_flags_type.
* rx-tdep.c (rx_psw_type): Likewise.
(rx_fpsw_type): Likewise.
* sparc-tdep.c (sparc_psr_type): Likewise.
(sparc_fsr_type): Likewise.
* sparc64-tdep.c (sparc64_pstate_type): Likewise.
(sparc64_ccr_type): Likewise.
(sparc64_fsr_type): Likewise.
(sparc64_fprs_type): Likewise.

Add support for the new names of the RISC-V fmv.x.s and fmv.s.x instructions, vis: fmv.x.w and fmv.w.x.

PR 22179
opcodes * riscv-opc.c (riscv_opcodes): Add fmv.x.w and fmv.w.x as the new
names for the fmv.x.s and fmv.s.x instructions respectively.

gas * testsuite/gas/riscv/fmv.x.s: New file: Tests the support for the
renamed fmv.x.s and fmv.s.x instructions.
* testsuite/gas/riscv/fmv.x.d: New file: Test driver.

Constify find_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* findcmd.c (find_command): Constify.

Constify some commands in ada-tasks.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* ada-tasks.c (task_command_1, task_command): Constify.

Constify some commands in symtab.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* symtab.c (maintenance_print_symbol_cache)
(maintenance_flush_symbol_cache)
(maintenance_print_symbol_cache_statistics): Constify.

Constify some commands in inferior.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* inferior.c (detach_inferior_command, kill_inferior_command)
(inferior_command): Constify.

Constify some commands in regcache.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* regcache.c (regcache_print, maintenance_print_registers)
(maintenance_print_raw_registers)
(maintenance_print_cooked_registers)
(maintenance_print_register_groups)
(maintenance_print_remote_registers): Constify.

Constify some commands in printcmd.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* printcmd.c (map_display_numbers, undisplay_command)
(enable_disable_display_command, enable_display_command)
(disable_display_command): Constify.

Constify some commands in breakpoint.c

This also makes delete_command static; but now I wonder if it is used
in Insight and should not be touched.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* breakpoint.h (delete_command): Don't declare.
* breakpoint.c (delete_command, enable_once_command)
(enable_count_command, enable_delete_command, breakpoint_1)
(maintenance_info_breakpoints, stopin_command, stopat_command)
(delete_command, delete_trace_command, save_breakpoints)
(save_breakpoints_command, save_tracepoints_command): Constify.

Constify some commands in macrocmd.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* macrocmd.c (macro_expand_command, macro_expand_once_command)
(skip_ws, extract_identifier, macro_define_command)
(macro_undef_command, macro_list_command): Constify.

Constify some commands in infcmd.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* infcmd.c (environment_info, set_environment_command)
(unset_environment_command, path_info, info_proc_cmd_1)
(info_proc_cmd_mappings, info_proc_cmd_stat)
(info_proc_cmd_status, info_proc_cmd_cwd, info_proc_cmd_cmdline)
(info_proc_cmd_exe, info_proc_cmd_all): Constify.

Constify some commands in i386-tdep.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* i386-tdep.c (i386_mpx_info_bounds, i386_mpx_set_bounds):
Constify.

Constify add_symbol_file_from_memory_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* symfile-mem.c (add_symbol_file_from_memory_command): Constify.

Constify demangle_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* demangle.c (demangle_command): Constify.

Constify maintenance_info_program_spaces_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* progspace.c (maintenance_info_program_spaces_command):
Constify.

Constify some commands in compile.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* compile/compile.c (check_raw_argument, compile_file_command)
(compile_code_command, compile_print_command): Constify.

Constify maintenance_print_reggroups

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* reggroups.c (maintenance_print_reggroups): Constify.

Constify save_gdb_index_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* dwarf2read.c (save_gdb_index_command): Constify.

Constify info_probes_stap_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* stap-probe.c (info_probes_stap_command): Constify.

Constify unset_exec_wrapper_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* fork-child.c (unset_exec_wrapper_command): Constify.

Constify some commands in btrace.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* btrace.c (get_uint, get_context_size, no_chunk)
(maint_btrace_packet_history_cmd)
(maint_btrace_clear_packet_history_cmd, maint_btrace_clear_cmd)
(maint_info_btrace_cmd): Constify.

Constify delete_bookmark_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* reverse.c (delete_bookmark_command): Constify.

Constify some commands in remote.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* remote.c (set_memory_packet_size)
(set_memory_write_packet_size, show_memory_write_packet_size)
(set_memory_read_packet_size, show_memory_read_packet_size)
(compare_sections_command, packet_command, remote_put_command)
(remote_get_command, remote_delete_command): Constify.

Constify some commands in mips-tdep.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* mips-tdep.c (show_mipsfpu_command, set_mipsfpu_single_command)
(set_mipsfpu_double_command, set_mipsfpu_none_command)
(set_mipsfpu_auto_command): Constify.

Constify cd_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cli/cli-cmds.h (cd_command): Constify.
* cli/cli-cmds.c (cd_command): Constify.

Constify some commands in thread.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* thread.c (thread_name_command, thread_find_command): Constify.

Constify some commands in probes.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* probe.c (enable_probes_command, disable_probes_command):
Constify.

Constify some commands in exec.c, plus symbol_file_command

Note that this commit also changes deprecated_file_changed_hook -- not
used in the tree, but Insight will require a (presumably minor)
change.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* symfile.c (symbol_file_command): Constify.
* gdbcore.h (deprecated_file_changed_hook): Constify.
* exec.c (deprecated_file_changed_hook, exec_file_command)
(file_command): Constify.
* defs.h (symbol_file_command): Constify.

Constify some commands in remote-fileio.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* remote-fileio.c (set_system_call_allowed)
(show_system_call_allowed): Constify.

Constify some commands in tracepoint.c

In addition to the constification, this fixes a command-repeat bug.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* tracepoint.c (delete_trace_variable_command)
(tfind_end_command, tfind_start_command, tfind_pc_command)
(tfind_tracepoint_command, tfind_line_command)
(tfind_range_command, tfind_outside_command): Constify.

Constify some commands in ax-gdb.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* ax-gdb.c (maint_agent_printf_command, agent_command)
(agent_eval_command): Constify.

Constify some linespec functions

This changes a few linespec functions to work on "const char *" and
then fixes up all the callers.  This allows further constification
elsewhere.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* tracepoint.c (info_scope_command): Constify.
* python/python.c (gdbpy_decode_line): Constify.
* python/py-breakpoint.c (bppy_init): Constify.
* mi/mi-cmd-break.c (mi_cmd_break_insert_1): Constify.
* location.h: (new_linespec_location)
(string_to_event_location_basic, string_to_event_location):
Constify.
* location.c (new_linespec_location)
(string_to_event_location_basic, string_to_event_location):
Constify.
* linespec.h (decode_line_with_current_source)
(decode_line_with_last_displayed, linespec_lex_to_end): Constify.
* linespec.c (linespec_lex_to_end)
(decode_line_with_current_source)
(decode_line_with_last_displayed): Constify.
* guile/scm-breakpoint.c (gdbscm_register_breakpoint_x):
Constify.
* cli/cli-cmds.c (edit_command, list_command): Constify.
* breakpoint.h (until_break_command, watch_command_wrapper)
(awatch_command_wrapper, rwatch_command_wrapper)
(init_ada_exception_breakpoint): Constify.
* breakpoint.c (break_command_1, dprintf_command)
(break_range_command, watch_command_wrapper)
(rwatch_command_wrapper, awatch_command_wrapper)
(until_break_command, init_ada_exception_breakpoint)
(strace_marker_create_sals_from_location, trace_command)
(ftrace_command, strace_command, struct tracepoint): Constify.
* ax-gdb.c (agent_command_1): Constify.
* ada-lang.c (ada_exception_sal): Constify.

Constify some commands in record.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* record.c (cmd_record_delete, cmd_record_stop, cmd_record_save)
(cmd_record_goto_begin, cmd_record_goto_end, get_insn_number)
(get_context_size, no_chunk, get_insn_history_modifiers)
(cmd_record_insn_history, get_call_history_modifiers)
(cmd_record_call_history): Constify.

Constify some commands in source.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* source.c (show_substitute_path_command)
(unset_substitute_path_command, set_substitute_path_command):
Constify.

Constify commands maint.c, plus maintenance_print_type

In addition to the constification, this fixes a command-repeat bug.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* typeprint.c (maintenance_print_type): Constify.
* maint.c (maintenance_dump_me, maintenance_demangle)
(maintenance_time_display, maintenance_info_sections)
(maintenance_print_statistics, maintenance_deprecate)
(maintenance_undeprecate): Constify.
(maintenance_do_deprecate): Constify.  Use std::string.
(maintenance_selftest): Constify.
* gdbtypes.h (maintenance_print_type): Constify.

Constify unwind_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* hppa-tdep.c (unwind_command): Constify.

Constify some commands in target-descriptions.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* target-descriptions.c (unset_tdesc_filename_cmd)
(maint_print_c_tdesc_cmd, maintenance_check_xml_descriptions):
Constify.

Constify maintenance_print_dummy_frames

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* dummy-frame.c (maintenance_print_dummy_frames): Constify.

Constify some commands in tui.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* tui/tui.c (tui_enable_command, tui_disable_command): Constify.

Constify tui_reg_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.c (tui_reg_command): Constify.

Constify some commands in skip.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* skip.c (skip_file_command, skip_function_command)
(skip_enable_command, skip_disable_command, skip_delete_command):
Constify.

Constify some commands in record-btrace.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* record-btrace.c (cmd_record_btrace_bts_start)
(cmd_record_btrace_pt_start): Constify.

Constify some commands in symmisc.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* symmisc.c (maintenance_print_symbols)
(maintenance_print_msymbols, maintenance_print_objfiles)
(maintenance_info_symtabs, maintenance_check_symtabs)
(maintenance_expand_symtabs, maintenance_info_line_tables):
Constify.

Constify new_ui_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* top.c (new_ui_command): Constify.

Constify some commands in symfile.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* symfile.c (add_symbol_file_command)
(remove_symbol_file_command, list_overlays_command)
(map_overlay_command, unmap_overlay_command)
(overlay_auto_command, overlay_manual_command)
(overlay_off_command, overlay_load_command): Constify.

Constify some commands in spu-tdep.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* spu-tdep.c (info_spu_event_command, info_spu_signal_command)
(info_spu_mailbox_command, info_spu_dma_command)
(info_spu_proxydma_command): Constify.

Constify some commands in cli-logging.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cli/cli-logging.c (set_logging_on, set_logging_off): Constify.

Constify user_defined_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cli/cli-script.c (user_defined_command): Constify.

Constify commands in cli-dump.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cli/cli-dump.c (dump_memory_command, dump_value_command)
(dump_srec_memory, dump_srec_value, dump_ihex_memory)
(dump_ihex_value, dump_verilog_memory, dump_verilog_value)
(dump_tekhex_memory, dump_tekhex_value, dump_binary_memory)
(dump_binary_value, append_binary_memory, append_binary_value):
Constify.
(struct dump_context) <func>: Constify.
(add_dump_command): Update.

Constify some commands in  cli-cmds.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cli/cli-cmds.c (show_version, show_configuration)
(source_command, show_user): Constify.

Constify maintenance_print_target_stack

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* target.c (maintenance_print_target_stack): Constify.

Constify interpreter_exec_cmd

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* interps.c (interpreter_exec_cmd): Constify.

Constify cmd_record_full_restore

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* record-full.c (cmd_record_full_restore): Constify.

Constify some functions in memattr.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* memattr.c (enable_mem_command, disable_mem_command)
(delete_mem_command): Constify.

Constify show_convenience

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* value.c (show_convenience): Constify.

Constify core_file_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* gdbcore.h (core_file_command): Update.
* corefile.c (core_file_command): Constify.

Constify maintenance_print_user_registers

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* user-regs.c (maintenance_print_user_registers): Constify.

Constify maintenance_cplus_namespace

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cp-namespace.c (maintenance_cplus_namespace): Constify.

Constify first_component_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cp-support.c (first_component_command): Constify.

Constify some functions in psymtab.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* psymtab.c (maintenance_print_psymbols)
(maintenance_info_psymtabs, maintenance_check_psymtabs):
Constify.

Constify display_tib

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* windows-tdep.c (display_tib): Constify.

Constify two functions in linux-fork.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* linux-fork.c (delete_checkpoint_command)
(detach_checkpoint_command): Constify.

Constify two functions in cp-abi.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cp-abi.c (set_cp_abi_cmd, show_cp_abi_cmd): Constify.

Constify dump_arc_instruction_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* arc-tdep.c (dump_arc_instruction_command): Constify.

Constify two functions in valprint.c

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* valprint.c (set_radix, show_radix): Constify.

Constify info_probes_dtrace_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* dtrace-probe.c (info_probes_dtrace_command): Constify.

Constify not_just_help_class_command

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* command.h (not_just_help_class_command): Update.
* cli/cli-decode.h (not_just_help_class_command): Update.
* cli/cli-decode.c (not_just_help_class_command): Constify.

Constify add_cmd gdb_bfd.c

This constifies a command function in gdb_bfd.c.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* gdb_bfd.c (maintenance_info_bfds): Constify.

Add add_cmd function overloads

This adds two add_cmd overloads: one whose callback takes a const char *,
and one that doesn't accept a function at all.  The no-function overload
was introduced to avoid ambiguity when NULL was passed as the function.

Long term the goal is for all commands to take const arguments, and
for the non-const variants to be removed entirely.

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* cli/cli-decode.c (add_cmd, set_cmd_cfunc): New function
overloads.
(do_add_cmd): Rename from add_cmd.  Don't call set_cmd_cfunc.
(do_const_cfunc): New function.
(cmd_cfunc_eq): New overload.
(cli_user_command_p): Check do_const_cfunc.
* cli/cli-decode.h (struct cmd_list_element) <function>: New field
const_cfunc.
* command.h (add_cmd): Add const overload and no-function
overload.
(set_cmd_cfunc): Add const overload.
(cmd_const_cfunc_ftype): Declare.
(cmd_cfunc_eq): Add const overload.
* breakpoint.c, cli-cmds.c, cli-dump.c, guile/scm-cmd.c,
python/py-cmd.c, target.c, tracepoint.c: Use no-function add_cmd
overload.

Add support for __VA_OPT__

C++2a adds a "__VA_OPT__" feature that can be used to control the
pesky "," emission when the final (variable) argument of a variadic
macro is empty.  This patch implements this feature for gdb.  (A patch
to implement it for gcc is pending.)

gdb/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* macroexp.c (get_next_token_for_substitution): New function.
(substitute_args): Call it.  Check for __VA_OPT__.

gdb/testsuite/ChangeLog
2017-09-27  Tom Tromey  <tom@tromey.com>

* gdb.base/macscp.exp: Add __VA_OPT__ tests.

Prevent an infinite loop in the DWARF parsing code when encountering a CU structure with a small negative size.

PR 22219
* dwarf.c (process_debug_info): Add a check for a negative
cu_length field.

PR22216, infinite loop in readelf process_symbol_table

This should make readelf bombproof given a fuzzed DT_HASH.  Also
removes a bogus check that would have resulted in wrong histograms.

PR 22216
* readelf.c (process_symbol_table): Check that DT_HASH symbol
chains are only visited once, and report an error if not.  Display
invalid symbol index if chain is out of range.  Use the same logic
when calculating histograms rather than the PR 17531 fix.  Delete
bogus check that chained index is less than number of buckets.

nds32: Fix a tautological comparison.

Automatic date update in version.in

x86-64: Don't pass output_bfd to info->callbacks->minfo

Don't pass output_bfd to info->callbacks->minfo when dumping local IFUNC
functions in the map file.

PR ld/22199
* elf64-x86-64.c (elf_x86_64_finish_dynamic_symbol): Don't pass
output_bfd to info->callbacks->minfo.

dwarf2read: Restrict ICC workaround to ICC<14

GDB has a workaround for DWARF output by ICC, related to missing
DW_AT_declaration on incomplete types.  The bug was fixed in ICC 14,
so this commit adjusts GDB accordingly.

For the version check, this adds a new parser function for the ICC
producer string.  While at it, it also adds unit tests for the
producer parsing covering the new function and preexisting parsers.

gdb/ChangeLog:
2017-09-26  Walfred Tedeschi  <walfred.tedeschi@intel.com>
    Pedro Alves <palves@redhat.com>

* dwarf2read.c (dwarf2_cu): Remove field producer_is_icc and add
producer_is_icc_lt_14.
(producer_is_icc_lt_14): New function.
(check_producer): Add code for checking version of ICC.
(producer_is_icc): Move to producer.c.
(read_structure_type): Restrict ICC workaround to ICC<14.
* producer.c: Include selftest.h.
(producer_is_icc, producer_parsing_tests, _initialize_producer):
New functions.
* producer.h (producer_is_icc): New declaration.

Move GDB producer parsing routines to a separate file

gdb/ChangeLog:
2017-09-26  Walfred Tedeschi  <walfred.tedeschi@intel.com>

* Makefile.in (SFILES): Add producer.c.
(COMMON_OBS): Add producer.o
* amd64-tdep.c (producer.h): Add new include.
* dwarf2read.c (producer.h): Add new include.
* producer.c: New file.
* producer.h: New file.
* utils.c (producer_is_gcc, producer_is_gcc_ge_4): Move to
producer.c.
* utils.h (producer_is_gcc, producer_is_gcc_ge_4): Move to
producer.h.

Use "switch_to_thread" more thoroughly on gdbserver

This is a technical debt that I left when I ported "switch_to_thread"
to gdbserver.  It's a simple patch that converts occurences of:

  current_thread = find_thread_ptid (ptid);

to:

  switch_to_thread (ptid);

This patch also converts a simple "if" to a "gdb_assert" on
gdbserver's version of "switch_to_thread":

  gdb_assert (ptid != minus_one_ptid);

This change makes the code more similar to what GDB already does.

Regtested on BuildBot.

gdb/gdbserver/ChangeLog:
2017-09-26  Sergio Durigan Junior  <sergiodj@redhat.com>

* gdbthread.h: Include "common-gdbthread.h".
* inferiors.c (switch_to_thread): Use "gdb_assert" instead of
"if" when validating the ptid.
* remote-utils.c: Include "gdbthread.h".
(prepare_resume_reply): Use "switch_to_thread".
* target.c (done_accessing_memory): Likewise.

Allow linking GDB with ncursesw

Triggered by https://launchpad.net/bugs/1275210, to be able to cope
with UTF-8 characters in gdbtui.

Reference:
  https://sourceware.org/ml/gdb-patches/2017-09/msg00356.html

gdb/ChangeLog:
2017-09-26  Matthias Klose  <doko@ubuntu.com>

* configure.ac: Search ncursesw before ncurses.
Check ncursesw/ncurses.h before ncurses/ncurses.h.
* gdb_curses.h: Include <ncursesw/ncurses.h>
* config.in, configure: Regenerate.

Fix an assertion failure when parsing a fuzzed x86_64 ELF binary.

PR 22172
* elf64-x86-64.c (elf_x86_64_info_to_howto): Do not trigger an
assertion failure if elf_x86_64_rtype_to_howto has already issued
an error message.

Fix incorrect register mask.

PR gold/22213
* sparc.cc (Target_sparc): Fix incorrect register mask.

Tidy reading data in read_formatted_entries

Using read_attribute_value accomplishes two things: It checks for
unexpected formats, and ensures the buffer pointer always increments.

PR 22210
* dwarf2.c (read_formatted_entries): Use read_attribute_value to
read data.

Avoid needless resource usage when processing a corrupt DWARF directory or file name table.

PR 22210
* dwarf2.c (read_formatted_entries): Fail early if we know that
the loop parsing data entries will overflow the end of the
section.

Remove support for Solaris < 10 (PR gdb/22185)

Given that GCC has obsoleted/removed support for Solaris 9 in GCC 4.9/5 in 2013:

    https://gcc.gnu.org/gcc-4.9/changes.html
    https://gcc.gnu.org/ml/gcc-patches/2013-05/msg00728.html

and the last gdb version that can be compiled with gcc 4.9 is 7.12.1 only when
configured with --disable-build-with-cxx, it's time to obsolete/remove support
for Solaris < 10.

This patch does this, simplifying configure.nat along the way (only a single
sol2 configuration with variants for i386 and sparc).

Some configure checks for older Solaris versions can go, too, and the check
for libthread_db.so.1 removed:

* Since Solaris 10, dlopen has moved to libc and libdl.so is just a
  filter on ld.so.1, so no need to check.

* $RDYNAMIC is already handled above (and is a no-op with Solaris ld
  anyway).

Both proc-service.c and sol-thread.c lose support for (Solaris-only)
PROC_SERVICE_IS_OLD.

The attached revised patch has been tested on sparcv9-sun-solaris2.10,
sparcv9-sun-solaris2.11.4, amd64-pc-solaris2.10, amd64-pc-solaris2.11.4,
and x86_64-pc-linux-gnu.

I've also started an i386-pc-solaris2.9 build to check that it really
stops as expected.

PR gdb/22185
* configure.host <*-*-solaris2.[01], *-*-solaris2.[2-9]*>: Mark as
obsolete.
Use gdb_host sol2 for i[34567]86-*-solaris2*, x86_64-*-solaris2*.
Remove i386sol2 support.
* configure.nat <i386sol2>: Remove.
<sol2-64>: Fold into ...
<sol2>: ... this.
Move common settings to default section.
Add sol-thread.o.
* configure.tgt <i[34567]86-*-solaris2.1[0-9]*,
x86_64-*-solaris2.1[0-9]*>: Rename to ...
<i[34567]86-*-solaris2*, x86_64-*-solaris2*>: ... this.
<i[34567]86-*-solaris*>: Remove.
<sparc-*-solaris2.[0-6], sparc-*-solaris2.[0-6].*>: Remove.

* configure.ac: Remove wctype in libw check.
(_MSE_INT_H): Don't define on Solaris 7-9.
<solaris*>: Remove libthread_db.so.1 check.
* configure: Regenerate.
* config.in: Regenerate.

* proc-service.c: Remove PROC_SERVICE_IS_OLD handling.
(gdb_ps_prochandle_t, gdb_ps_read_buf_t, gdb_ps_write_buf_t)
(gdb_ps_size_t): Remove.
Use base types in users.
* sol-thread.c: Likewise, also for gdb_ps_addr_t.

* NEWS (Changes since GDB 8.0): Document Solaris 2.0-9 removal.

Fix gdb 8.1 Solaris/SPARC compilation (PR build/22206)

When testing my Solaris < 10 removal patch on Solaris/SPARC, I found
that gdb mainline is currently broken there due to the recent SPARC M7
ADI patches:

/vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c:1876:0: error: "PSR_ICC" redefined [-Werror]
 #define PSR_ICC  0x00f00000
 ^
In file included from /usr/include/v7/sys/privregs.h:24:0,
                 from /usr/include/sys/regset.h:420,
                 from /usr/include/sys/ucontext.h:21,
                 from /usr/include/sys/signal.h:231,
                 from /usr/include/sys/procset.h:23,
                 from /usr/include/sys/wait.h:25,
                 from /usr/include/stdlib.h:21,
                 from build-gnulib/import/stdlib.h:36,
                 from /vol/src/gnu/gdb/gdb/local/gdb/common/common-defs.h:53,
                 from /vol/src/gnu/gdb/gdb/local/gdb/defs.h:28,
                 from /vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c:20:
/usr/include/v7/sys/psr.h:35:0: note: this is the location of the previous definition
 #define PSR_ICC  0x00F00000 /* integer condition codes */
 ^
/vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c:1878:0: error: "PSR_IMPL" redefined [-Werror]
 #define PSR_IMPL 0xf0000000
 ^
In file included from /usr/include/v7/sys/privregs.h:24:0,
                 from /usr/include/sys/regset.h:420,
                 from /usr/include/sys/ucontext.h:21,
                 from /usr/include/sys/signal.h:231,
                 from /usr/include/sys/procset.h:23,
                 from /usr/include/sys/wait.h:25,
                 from /usr/include/stdlib.h:21,
                 from build-gnulib/import/stdlib.h:36,
                 from /vol/src/gnu/gdb/gdb/local/gdb/common/common-defs.h:53,
                 from /vol/src/gnu/gdb/gdb/local/gdb/defs.h:28,
                 from /vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c:20:
/usr/include/v7/sys/psr.h:41:0: note: this is the location of the previous definition
 #define PSR_IMPL 0xF0000000 /* implementation */
 ^

Comparing Solaris 11.4 <v7/sys/psr.h> and sparc64-tdep.c, there are more
inconsistencies:

<v7/sys/psr.h>:

#define PSR_S 0x00000080 /* supervisor mode */
#define PSR_ICC 0x00F00000 /* integer condition codes */
#define PSR_VER 0x0F000000 /* mask version */
#define PSR_IMPL 0xF0000000 /* implementation */
#define PSR_RSV 0x000FC000 /* reserved */

sparc64-tdep.c:

#define PSR_S 0x00000080
#define PSR_ICC 0x00f00000
#define PSR_VERS 0x0f000000
#define PSR_IMPL 0xf0000000
#define PSR_V8PLUS 0xff000000
#define PSR_XCC 0x000f0000

Apart from the capitalization differences that trip g++, the names
differ (PSR_VER vs. PSR_VERS), PSR_XCC is included in Solaris' PSR_RSV,
and there's no PSR_V8PLUS on Solaris either.

/vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c: In function `int adi_tag_fd()':
/vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c:296:63: error: format `%d' expects argument of type `int', but argument 4 has type `pid_t {aka long int}' [-Werror=format=]
   snprintf (cl_name, sizeof(cl_name), "/proc/%d/adi/tags", pid);
                                                               ^
/vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c: In function `bool adi_is_addr_mapped(CORE_ADDR, std::size_t)':
/vol/src/gnu/gdb/gdb/local/gdb/sparc64-tdep.c:314:64: error: format `%d' expects argument of type `int', but argument 4 has type `pid_t {aka long int}' [-Werror=format=]
   snprintf (filename, sizeof filename, "/proc/%d/adi/maps", pid);
                                                                ^

You cannot always print a pid_t, which can be either int or long on
Solaris, as an int.

Obviously, the ADI patch which modifies code shared between all SPARC
targets, hasn't been tested on anything but Linux/SPARC.

The patch below includes the minimal fixes necessary to unbreak the
Solaris/SPARC build.

However, as detailed in the PR, there's more breakage here: apart from
not bothering to implement ADI support on Solaris, the code contains
several more changes to shared/common SPARC code that are simply wrong
on anything but Linux/SPARC.

The patch was tested on sparcv9-sun-solaris2.10 and
sparcv9-sun-solaris2.11.4 (build and gdb/gdb gdb/gdb smoke test only).

PR build/22206
* sparc64-tdep.c (adi_tag_fd): Print pid as long.
(adi_is_addr_mapped): Likewise.
(PSR_ICC): Don't redefine.
(PSR_IMPL): Likewise.

PR22209, invalid memory read in find_abstract_instance_name

This patch adds bounds checking for DW_FORM_ref_addr die refs, and
calculates them relative to the first .debug_info section.  See the
big comment for why calculating relative to the current .debug_info
section was wrong for relocatable object files.

PR 22209
* dwarf2.c (struct comp_unit): Delete sec_info_ptr field.
(find_abstract_instance_name): Calculate DW_FORM_ref_addr relative
to stash->info_ptr_memory, and check die_ref is within that memory.
Set info_ptr_end correctly when another CU is refd.  Check die_ref
for DW_FORM_ref4 etc. is within CU.

Fix address violations when parsing a corrupt DWARF linenumber table.

PR 22154
* dwarf.c (get_line_filename_and_dirname): Add extra checks for
buffer overruns.

LD/testsuite: Fix undefined symbol tests for MIPS and S+core targets

Fix commit 647d4de92e06 ("Test undefined symbols in shared libraries")
and remove "undefined symbols in shared lib (dyn reloc)" test failures
for MIPS and S+core targets.

This test assumes 3 dynamic relocation entries, however the respective
psABIs define an extra single dummy *_NONE relocation at the beginning
of the dynamic relocation table, which results in test output like the
following for MIPS:

Relocation section '.rel.dyn' at offset 0x33c contains 4 entries:
 Offset     Info    Type            Sym.Value  Sym. Name
00000000  00000000 R_MIPS_NONE
00010364  00000903 R_MIPS_REL32      00000000   undef_pfun
00010360  00000a03 R_MIPS_REL32      00000000   undef_data
00010368  00000b03 R_MIPS_REL32      00000000   undef_notype

and S+core:

Relocation section '.rel.dyn' at offset 0x274 contains 4 entries:
 Offset     Info    Type            Sym.Value  Sym. Name
00000000  00000000 R_SCORE_NONE
5ffe0464  00000e12 R_SCORE_REL32     00000000   undef_notype
5ffe045c  00000f12 R_SCORE_REL32     00000000   undef_data
5ffe0460  00001012 R_SCORE_REL32     00000000   undef_pfun

targets respectively, and also n64 MIPS targets pad relocation triplets
they use with extra *_NONE relocations, resulting in output like this:

Relocation section '.rel.dyn' at offset 0x4c0 contains 4 entries:
  Offset          Info           Type           Sym. Value    Sym. Name
000000000000  000000000000 R_MIPS_NONE
                    Type2: R_MIPS_NONE
                    Type3: R_MIPS_NONE
000000010508  000900001203 R_MIPS_REL32      0000000000000000 undef_pfun
                    Type2: R_MIPS_64
                    Type3: R_MIPS_NONE
000000010500  000a00001203 R_MIPS_REL32      0000000000000000 undef_data
                    Type2: R_MIPS_64
                    Type3: R_MIPS_NONE
000000010510  000b00001203 R_MIPS_REL32      0000000000000000 undef_notype
                    Type2: R_MIPS_64
                    Type3: R_MIPS_NONE

Expect the right number of *_NONE relocations then, and adjust the total
number of relocations expected accordingly.

ld/
* testsuite/ld-undefined/undefined.exp: Correct the dyn reloc
test for MIPS and S+core targets.