MIPS messages

More standardization of messages.

bfd/
* elfxx-mips.c: Standardize error/warning messages.
binutils/
* testsuite/binutils-all/mips/mips-reginfo-n32.d,
* testsuite/binutils-all/mips/mips-reginfo.d: Update.
gas/
* testsuite/gas/mips/reginfo-2.l: Update.
ld/
* testsuite/ld-mips-elf/attr-gnu-4-12.d,
* testsuite/ld-mips-elf/attr-gnu-4-13.d,
* testsuite/ld-mips-elf/attr-gnu-4-14.d,
* testsuite/ld-mips-elf/attr-gnu-4-16.d,
* testsuite/ld-mips-elf/attr-gnu-4-17.d,
* testsuite/ld-mips-elf/attr-gnu-4-18.d,
* testsuite/ld-mips-elf/attr-gnu-4-19.d,
* testsuite/ld-mips-elf/attr-gnu-4-21.d,
* testsuite/ld-mips-elf/attr-gnu-4-23.d,
* testsuite/ld-mips-elf/attr-gnu-4-24.d,
* testsuite/ld-mips-elf/attr-gnu-4-25.d,
* testsuite/ld-mips-elf/attr-gnu-4-26.d,
* testsuite/ld-mips-elf/attr-gnu-4-27.d,
* testsuite/ld-mips-elf/attr-gnu-4-28.d,
* testsuite/ld-mips-elf/attr-gnu-4-29.d,
* testsuite/ld-mips-elf/attr-gnu-4-31.d,
* testsuite/ld-mips-elf/attr-gnu-4-32.d,
* testsuite/ld-mips-elf/attr-gnu-4-34.d,
* testsuite/ld-mips-elf/attr-gnu-4-35.d,
* testsuite/ld-mips-elf/attr-gnu-4-36.d,
* testsuite/ld-mips-elf/attr-gnu-4-37.d,
* testsuite/ld-mips-elf/attr-gnu-4-38.d,
* testsuite/ld-mips-elf/attr-gnu-4-39.d,
* testsuite/ld-mips-elf/attr-gnu-4-41.d,
* testsuite/ld-mips-elf/attr-gnu-4-42.d,
* testsuite/ld-mips-elf/attr-gnu-4-43.d,
* testsuite/ld-mips-elf/attr-gnu-4-45.d,
* testsuite/ld-mips-elf/attr-gnu-4-46.d,
* testsuite/ld-mips-elf/attr-gnu-4-47.d,
* testsuite/ld-mips-elf/attr-gnu-4-48.d,
* testsuite/ld-mips-elf/attr-gnu-4-49.d,
* testsuite/ld-mips-elf/attr-gnu-4-52.d,
* testsuite/ld-mips-elf/attr-gnu-4-53.d,
* testsuite/ld-mips-elf/attr-gnu-4-54.d,
* testsuite/ld-mips-elf/attr-gnu-4-58.d,
* testsuite/ld-mips-elf/attr-gnu-4-59.d,
* testsuite/ld-mips-elf/attr-gnu-4-61.d,
* testsuite/ld-mips-elf/attr-gnu-4-62.d,
* testsuite/ld-mips-elf/attr-gnu-4-63.d,
* testsuite/ld-mips-elf/attr-gnu-4-64.d,
* testsuite/ld-mips-elf/attr-gnu-4-68.d,
* testsuite/ld-mips-elf/attr-gnu-4-69.d,
* testsuite/ld-mips-elf/attr-gnu-4-71.d,
* testsuite/ld-mips-elf/attr-gnu-4-72.d,
* testsuite/ld-mips-elf/attr-gnu-4-73.d,
* testsuite/ld-mips-elf/attr-gnu-4-74.d,
* testsuite/ld-mips-elf/attr-gnu-4-78.d,
* testsuite/ld-mips-elf/attr-gnu-4-79.d,
* testsuite/ld-mips-elf/attr-gnu-4-81.d,
* testsuite/ld-mips-elf/attr-gnu-4-89.d,
* testsuite/ld-mips-elf/attr-gnu-8-12.d,
* testsuite/ld-mips-elf/attr-gnu-8-21.d,
* testsuite/ld-mips-elf/bal-jalx-pic-micromips-n32.d,
* testsuite/ld-mips-elf/bal-jalx-pic-micromips-n64.d,
* testsuite/ld-mips-elf/bal-jalx-pic-micromips.d,
* testsuite/ld-mips-elf/bal-jalx-pic-n32.d,
* testsuite/ld-mips-elf/bal-jalx-pic-n64.d,
* testsuite/ld-mips-elf/bal-jalx-pic.d,
* testsuite/ld-mips-elf/mode-change-error-1.d,
* testsuite/ld-mips-elf/unaligned-branch-2.d,
* testsuite/ld-mips-elf/unaligned-branch-ignore-2.d,
* testsuite/ld-mips-elf/unaligned-branch-ignore-micromips.d,
* testsuite/ld-mips-elf/unaligned-branch-ignore-mips16.d,
* testsuite/ld-mips-elf/unaligned-branch-ignore-r6-1.d,
* testsuite/ld-mips-elf/unaligned-branch-micromips.d,
* testsuite/ld-mips-elf/unaligned-branch-mips16.d,
* testsuite/ld-mips-elf/unaligned-branch-r6-1.d,
* testsuite/ld-mips-elf/unaligned-branch-r6-2.d,
* testsuite/ld-mips-elf/unaligned-branch.d,
* testsuite/ld-mips-elf/unaligned-jalx-1.d,
* testsuite/ld-mips-elf/unaligned-jalx-3.d,
* testsuite/ld-mips-elf/unaligned-jalx-addend-1.d,
* testsuite/ld-mips-elf/unaligned-jalx-addend-3.d,
* testsuite/ld-mips-elf/unaligned-jalx-addend-micromips-1.d,
* testsuite/ld-mips-elf/unaligned-jalx-addend-mips16-1.d,
* testsuite/ld-mips-elf/unaligned-jalx-micromips-1.d,
* testsuite/ld-mips-elf/unaligned-jalx-mips16-1.d,
* testsuite/ld-mips-elf/unaligned-jump-micromips.d,
* testsuite/ld-mips-elf/unaligned-jump-mips16.d,
* testsuite/ld-mips-elf/unaligned-jump.d: Update.

ELF linker messages

This fixes a number of cases where we capitalized error messages or
printed a full-stop, and corrects "Dwarf" to "DWARF".

bfd/
* dwarf2.c, * elf-attrs.c, * elf-eh-frame.c, * elf.c, * elf32-gen.c,
* elflink.c: Standardize error/warning messages.  Replace use of
linker callback einfo with _bfd_error_handler when possible.
ld/
* testsuite/ld-elf/indirect.exp,
* testsuite/ld-elf/pr22649-2ab-mips.msg,
* testsuite/ld-elf/pr22649-2cd-mips.msg,
* testsuite/ld-elf/pr22649.msg,
* testsuite/ld-elf/tls_common.exp,
* testsuite/ld-elfcomm/elfcomm.exp,
* testsuite/ld-arm/attr-merge-incompatible.d,
* testsuite/ld-tic6x/attr-compatibility-gnu-other.d,
* testsuite/ld-tic6x/attr-compatibility-other-gnu.d,
* testsuite/ld-tic6x/attr-compatibility-other-other.d,
* testsuite/ld-i386/warn1.d: Update expected error/warning messages.

unrecognized/unsupported reloc message

It must get boring translating all the variants we have of
unrecognized/unsupported/invalid/unexpected reloc number.  This patch
cuts down on the number of variations.

* aoutx.h, * coff-alpha.c, * coff-i860.c, * coff-m68k.c,
* coff-mcore.c, * coff-ppc.c, * coff-rs6000.c, * coff-sh.c,
* coff-tic4x.c, * coff-tic54x.c, * coff-tic80.c, * coff-w65.c,
* elf-bfd.h, * elf-m10300.c, * elf.c, * elf32-avr.c, * elf32-bfin.c,
* elf32-cr16.c, * elf32-cr16c.c, * elf32-cris.c, * elf32-crx.c,
* elf32-d10v.c, * elf32-d30v.c, * elf32-dlx.c, * elf32-epiphany.c,
* elf32-fr30.c, * elf32-frv.c, * elf32-i370.c, * elf32-i386.c,
* elf32-i960.c, * elf32-ip2k.c, * elf32-iq2000.c, * elf32-lm32.c,
* elf32-m32c.c, * elf32-m32r.c, * elf32-m68hc11.c, * elf32-m68hc12.c,
* elf32-m68k.c, * elf32-mcore.c, * elf32-mep.c, * elf32-metag.c,
* elf32-microblaze.c, * elf32-mips.c, * elf32-moxie.c,
* elf32-msp430.c, * elf32-mt.c, * elf32-nds32.c, * elf32-or1k.c,
* elf32-pj.c, * elf32-ppc.c, * elf32-rl78.c, * elf32-rx.c,
* elf32-s390.c, * elf32-sh.c, * elf32-spu.c, * elf32-tic6x.c,
* elf32-v850.c, * elf32-vax.c, * elf32-visium.c, * elf32-wasm32.c,
* elf32-xgate.c, * elf32-xtensa.c, * elf64-alpha.c,
* elf64-ia64-vms.c, * elf64-mips.c, * elf64-mmix.c,
* elf64-ppc.c, * elf64-s390.c, * elf64-sparc.c, * elf64-x86-64.c,
* elfn32-mips.c, * elfnn-aarch64.c, * elfnn-ia64.c,
* elfnn-riscv.c, * elfxx-mips.c, * elfxx-riscv.c, * elfxx-riscv.h,
* elfxx-sparc.c, * elfxx-sparc.h, * reloc.c: Standardize
unrecognized/unsupported reloc message.

PPC error/warning messages

The GNU coding standard says error messages should be of the form
  program:sourcefile:lineno: message
or
  program: message

and
"The string message should not begin with a capital letter when it
follows a program name and/or file name, because that isn’t the
beginning of a sentence. (The sentence conceptually starts at the
beginning of the line.) Also, it should not end with a period."

This patch does that for ppc, and removes some British spelling.
I've also switched some error output from using the linker callback
einfo to _bfd_error_handler, due to improved compilation time
argument checking now done for the latter function.

bfd/
* elf32-ppc.c: Standardize error/warning messages.  Use
_bfd_error_handler rather than einfo when einfo features not used.
* elf64-ppc.c: Likewise.
ld/
* testsuite/ld-powerpc/attr-gnu-12-21.d: Update.
* testsuite/ld-powerpc/attr-gnu-4-12.d: Update.
* testsuite/ld-powerpc/attr-gnu-4-13.d: Update.
* testsuite/ld-powerpc/attr-gnu-4-21.d: Update.
* testsuite/ld-powerpc/attr-gnu-4-23.d: Update.
* testsuite/ld-powerpc/attr-gnu-4-31.d: Update.
* testsuite/ld-powerpc/attr-gnu-4-32.d: Update.
* testsuite/ld-powerpc/attr-gnu-8-23.d: Update.

Fix double space expected in cp_test_ptype_class

I noticed some failures of some buildbot slaves, e.g.:

FAIL: gdb.cp/nested-types.exp: ptype S10 (limit = 1) // wrong nested type enum definition: enum S10::E10 {S10::A10, S10::B10, S10::C10};

The issue is that they have an older gcc (not c++11 by default?) that
doesn't emit the enum underlying type information.  When the
enum type is printed by ptype, it looks like this:

  enum S10::E10 {S10::A10, S10::B10, S10::C10};

instead of this on older gccs:

  enum S10::E10 : unsigned int {S10::A10, S10::B10, S10::C10};

The regex that matches this is in cp_test_ptype_class, and is

  enum $nested_name (: (unsigned )?int)? \{

If the "unsigned int" portion is not present, then it requires the
string to have two spaces between the enum name and opening bracket.
The fix is simply to move the trailing space inside the ? group.

gdb/testsuite/ChangeLog:

* lib/cp-support.exp (cp_test_ptype_class): Move space inside
parentheses.

Automatic date update in version.in

Remove most cleanups from linux-thread-db.c

This removes most (but not all) cleanups from linux-thread-db.c.
std::string and std::vector are used in place of manual memory
management.

The remaining cleanup in linux-thread-db.c uses
make_cleanup_free_char_ptr_vec, which requires a somewhat bigger
change.

Regression tested by the buildbot.

ChangeLog
2018-02-24  Tom Tromey  <tom@tromey.com>

* linux-thread-db.c (try_thread_db_load_from_pdir_1)
(try_thread_db_load_from_dir, thread_db_load_search): Use
std::string.
(info_auto_load_libthread_db_compare): Return bool.  Change
argument types.
(info_auto_load_libthread_db): Use std::vector, std::string.
Remove cleanups.

Remove cleanups from check_fast_tracepoint_sals

This changes the gdbarch fast_tracepoint_valid_at method to use a
std::string as its out parameter, and then updates all the uses.  This
allows removing a cleanup from breakpoint.c.

Regression tested by the buildbot.

ChangeLog
2018-02-24  Tom Tromey  <tom@tromey.com>

* i386-tdep.c (i386_fast_tracepoint_valid_at): "msg" now a
std::string.
* gdbarch.sh (fast_tracepoint_valid_at): Change "msg" to a
std::string*.
* gdbarch.c: Rebuild.
* gdbarch.h: Rebuild.
* breakpoint.c (check_fast_tracepoint_sals): Use std::string.
* arch-utils.h (default_fast_tracepoint_valid_at): Update.
* arch-utils.c (default_fast_tracepoint_valid_at): "msg" now a
std::string*.

Automatic date update in version.in

GDB/testsuite: Fix a typo in $actual_line

Fix a commit 883fd55ab104 ("Record nested types") issue:

ERROR: tcl error sourcing .../gdb/testsuite/gdb.cp/nested-types.exp.
ERROR: can't read "actual_linejj": no such variable
    while executing
"append txt " definition: $actual_linejj""
    (procedure "cp_test_ptype_class" line 324)
    invoked from within
"cp_test_ptype_class $name "ptype $name (limit = $limit)" $key  $name $children"    (procedure "test_nested_limit" line 28)
    invoked from within
"test_nested_limit -1 false"
    (file ".../gdb/testsuite/gdb.cp/nested-types.exp" line 310)
    invoked from within
"source .../gdb/testsuite/gdb.cp/nested-types.exp"
    ("uplevel" body line 1)
    invoked from within
"uplevel #0 source .../gdb/testsuite/gdb.cp/nested-types.exp"
    invoked from within
"catch "uplevel #0 source $test_file_name""
testcase .../gdb/testsuite/gdb.cp/nested-types.exp completed in 9 seconds

caused by $actual_line having been accidentally referred to as
$actual_linejj in one place.

gdb/testsuite/
* lib/cp-support.exp (cp_test_ptype_class): Fix a typo in the
name of a variable: $actual_linejj -> $actual_line.

dwarf: Make sect_offset 64-bits

Does anybody have an opinion about this?  It would be nice to unbreak
the "default" build with clang (i.e. without passing special -Wno-error=
flags).

Here's a version rebased on today's master.

From 47d28075117fa2ddb93584ec50881e33777a85e5 Mon Sep 17 00:00:00 2001
From: Simon Marchi <simon.marchi@ericsson.com>
Date: Sat, 30 Dec 2017 22:48:18 -0500
Subject: [PATCH] dwarf: Make sect_offset 64-bits

Compiling with Clang 6 shows these errors:

/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:26610:43: error: result of comparison of constant 4294967296 with expression of type 'typename std::underlying_type<sect_offset>::type' (a
ka 'unsigned int') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
      if (to_underlying (per_cu.sect_off) >= (static_cast<uint64_t> (1) << 32))
          ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/emaisin/src/binutils-gdb/gdb/dwarf2read.c:26618:43: error: result of comparison of constant 4294967296 with expression of type 'typename std::underlying_type<sect_offset>::type' (a
ka 'unsigned int') is always false [-Werror,-Wtautological-constant-out-of-range-compare]
      if (to_underlying (per_cu.sect_off) >= (static_cast<uint64_t> (1) << 32))
          ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The code in question checks if there is any offset exceeding 32 bits,
and therefore if we need to use the 64-bit DWARF format when writing the
.debug_names section.  The type we use currently to represent section
offsets is an unsigned int (32-bits), which means a value of this type
will never exceed 32 bits, hence the errors above.

There are many signs that we want to support 64-bits DWARF (although I
haven't tested), such as:

 - We correctly read initial length fields (read_initial_length)
 - We take that into account when reading offsets (read_offset_1)
 - The check_dwarf64_offsets function

However, I don't see how it can work if sect_offset is a 32-bits type.
Every time we record a section offset, we risk truncating the value.
And if a file uses the 64-bit DWARF format, it's most likely because
there are such offset values that overflow 32 bits.

Because of this, I think the way forward is to change sect_offset to be
a uint64_t.  It will be able to represent any offset, regardless of the
bitness of the DWARF info.

This patch was regtested on the buildbot.

gdb/ChangeLog:

* gdbtypes.h (sect_offset): Change type to uint64_t.
(sect_offset_str): New function.
* dwarf2read.c (create_addrmap_from_aranges): Use
sect_offset_str.
(error_check_comp_unit_head): Likewise.
(create_debug_type_hash_table): Likewise.
(read_cutu_die_from_dwo): Likewise.
(init_cutu_and_read_dies): Likewise.
(init_cutu_and_read_dies_no_follow): Likewise.
(process_psymtab_comp_unit_reader): Likewise.
(partial_die_parent_scope): Likewise.
(peek_die_abbrev): Likewise.
(process_queue): Likewise.
(dwarf2_physname): Likewise.
(read_namespace_alias): Likewise.
(read_import_statement): Likewise.
(create_dwo_cu_reader): Likewise.
(create_cus_hash_table): Likewise.
(lookup_dwo_cutu): Likewise.
(inherit_abstract_dies): Likewise.
(read_func_scope): Likewise.
(read_call_site_scope): Likewise.
(dwarf2_add_member_fn): Likewise.
(read_common_block): Likewise.
(read_module_type): Likewise.
(read_typedef): Likewise.
(read_subrange_type): Likewise.
(load_partial_dies): Likewise.
(read_partial_die): Likewise.
(find_partial_die): Likewise.
(read_str_index): Likewise.
(dwarf2_string_attr): Likewise.
(build_error_marker_type): Likewise.
(lookup_die_type): Likewise.
(dump_die_shallow): Likewise.
(follow_die_ref): Likewise.
(dwarf2_fetch_die_loc_sect_off): Likewise.
(dwarf2_fetch_constant_bytes): Likewise.
(follow_die_sig): Likewise.
(get_signatured_type): Likewise.
(get_DW_AT_signature_type): Likewise.
(dwarf2_find_containing_comp_unit): Likewise.
(set_die_type): Likewise.

PR22881, null pointer dereference in assign_file_positions_for_non_load_sections

PR 22881
* elf.c (assign_file_positions_for_non_load_sections): Remove RELRO
segment if no matching LOAD segment.

Document the assembler's .dc, .dcb and .ds directives.

* doc/as.texinfo (Pseudo Ops): Add nodes for .dc, .dcb and .ds.

nds32: Support target directive .ict_model.

Fix up one ChangeLog date.

Automatic date update in version.in

New plugin interface to get list of symbols wrapped with --wrap option.

2018-02-22  Sriraman Tallam  <tmsriram@google.com>

* plugin.cc (get_wrap_symbols): New plugin interface.
(load): Add get_wrap_symbols to transfer vector.
* plugin-api.h (ld_plugin_get_wrap_symbols): New plugin interface.
* testsuite/plugin_test.c (onload): Call and check get_wrap_symbols
interface.
* testsuite/plugin_test_wrap_symbols.sh: New test script.
* testsuite/plugin_test_wrap_symbols_1.cc: New file.
* testsuite/plugin_test_wrap_symbols_2.cc: New file.
* testsuite/Makefile.am (plugin_test_wrap_symbols): New test.
* testsuite/Makefile.in: Regenerate.

RISC-V: Make disassebler work for --enable-targets=all config.

opcodes/
* disassemble.c (ARCH_riscv): Define if ARCH_all.

Diagnose when trying to assemble conditional FP16 vmovx and vins

This patch makes GAS emit a warning when trying to assemble the Armv8.2
FP16 instructions VMOVX and VINS with condition codes. The Armv8-A
Reference Manual specifies these instructions without conditional codes
and says that if they are found in an IT block that they are CONSTRAINED
UNPREDICABLE.

gas/ChangeLog:
2018-02-22  Andre Vieira  <andre.simoesdiasvieira@arm.com>

* config/tc-arm.c (do_neon_movhf): If conditional error out when in arm
mode and emit warning in thumb mode.
* testsuite/gas/arm/armv8-2-fp16-scalar-bad.s: Add new tests.
* testsuite/gas/arm/armv8-2-fp16-scalar-bad.l: Idem.

x86: Add {rex} pseudo prefix

Add {rex} pseudo prefix to generate a REX byte for integer and legacy
vector instructions if possible.  Note that this differs from the rex
prefix which generates REX prefix unconditionally.

gas/

* config/tc-i386.c (_i386_insn): Add rex_encoding.
(md_assemble): When i.rex_encoding is true, generate a REX byte
if possible.
(parse_insn): Set i.rex_encoding for {rex}.
* doc/c-i386.texi: Document {rex}.
* testsuite/gas/i386/x86-64-pseudos.s: Add {rex} tests.
* testsuite/gas/i386/x86-64-pseudos.d: Updated.

opcodes/

* i386-opc.tbl: Add {rex},
* i386-tbl.h: Regenerated.

Fix memory access violation when attempting to shorten a suffixed micromips instruction during lookup.

PR 22014
* config/tc-mips.c (mips_lookup_insn): Use memmove to strip the
instruction size suffix.

Automatic date update in version.in

Fix a typo.

gdb/testsuite/ChangeLog:

* gdb.arch/amd64-i386-address.exp: Fix a typo.

Add "common-defs.h" include to files in arch/ subdir not yet including it.

This fixes a build breakage on FreeBSD hosts.

gdb/ChangeLog:

* arch/aarch64.c: Include "common-defs.h".
* arch/amd64.c: Likewise.
* arch/i386.c: Likewise.

Remove a cleanup from parse_expression_for_completion

This removes a cleanup from parse_expression_for_completion, by
changing various expression-completion functions to use
gdb::unique_xmalloc_ptry rather than explicit malloc+free.

Regression tested by the buildbot.

gdb/ChangeLog
2018-02-21  Tom Tromey  <tom@tromey.com>

* value.h: (extract_field_op): Update.
* eval.c (extract_field_op): Return a const char *.
* expression.h (parse_expression_for_completion): Update.
* completer.c (complete_expression): Update.
(add_struct_fields): Make fieldname const.
* parse.c (expout_completion_name): Now a unique_xmalloc_ptr.
(mark_completion_tag, parse_exp_in_context_1): Update.
(parse_expression_for_completion): Change "name" to
unique_xmalloc_ptr*.

Remove a cleanup from call_function_by_hand_dummy

This removes a cleanup from call_function_by_hand_dummy, replacing
manual allocation with std::vector.

Regression tested by the buildbot.

gdb/ChangeLog
2018-02-21  Tom Tromey  <tom@tromey.com>

* infcall.c (call_function_by_hand_dummy): Use std::vector.

Pass readable_regcache to gdbarch method read_pc

We can pass readable_regcache to gdbarch method read_pc where it is
allowed to do read from regcache.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* avr-tdep.c (avr_read_pc): Change parameter type to
readable_regcache.
* gdbarch.sh (read_pc): Likewise.
* gdbarch.c: Re-generated.
* gdbarch.h: Re-generated.
* hppa-tdep.c (hppa_read_pc): Change parameter type to
readable_regcache.
* ia64-tdep.c (ia64_read_pc): Likewise.
* mips-tdep.c (mips_read_pc): Likewise.
* spu-tdep.c (spu_read_pc): Likewise.

Move register_dump to regcache-dump.c

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* Makefile.in (COMMON_SFILES): Add regcache-dump.c
* regcache-dump.c: New file.
* regcache.c: Move register_dump to regcache-dump.c.
(maintenance_print_registers): Likewise.
(maintenance_print_raw_registers): Likewise.
(maintenance_print_cooked_registers): Likewise.
(maintenance_print_register_groups): Likewise.
(maintenance_print_remote_registers): Likewise.
(_initialize_regcache): Likewise.
* regcache.h (register_dump): Moved from regcache.c.

Remove regcache::m_readonly_p

Now, m_readonly_p is always false, so we can remove it, and regcache no
longer includes pseudo registers.  Some regcache methods are lift up to
its parent class, like reg_buffer or detached_regcache.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* regcache.c (regcache::regcache): Update.
(regcache::invalidate): Move it to detached_regcache::invalidate.
(get_thread_arch_aspace_regcache): Update.
(regcache::raw_update): Update.
(regcache::cooked_read): Remove some code.
(regcache::cooked_read_value): Likewise.
(regcache::raw_write): Remove assert on m_readonly_p.
(regcache::raw_supply_integer): Move it to
detached_regcache::raw_supply_integer.
(regcache::raw_supply_zeroed): Likewise.
* regcache.h (detached_regcache) <raw_supply_integer>: New
declaration.
<raw_supply_zeroed, invalidate>: Likewise.
(regcache) <raw_supply_integer, raw_supply_zeroed>: Removed.
<invalidate>: Likewise.
<m_readonly_p>: Removed.

No longer create readonly regcache

Nowadays, we create a readonly regcache in get_return_value, and pass it
to gdbarch_return_value to get the return value.  In theory, we can pass a
readable_regcache instance and get the return value, because we don't need
to modify the regcache.  Unfortunately, gdbarch_return_value is designed
to multiplex regcache, according to READBUF and WRITEBUF.

 # If READBUF is not NULL, extract the return value and save it in this
 # buffer.
 #
 # If WRITEBUF is not NULL, it contains a return value which will be
 # stored into the appropriate register.

In fact, gdbarch_return_value should be split to three functions, 1) only
return return_value_convention, 2) pass regcache_readonly and readbuf, 3)
pass regcache and writebuf.  These changes are out of the scope of this
patch series, so I pass regcache to gdbarch_return_value even for read,
and trust each gdbarch backend doesn't modify regcache.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* infcmd.c (get_return_value): Let stop_regs point to
get_current_regcache.
* regcache.c (regcache::regcache): Remove.
(register_dump_reg_buffer): New class.
(regcache_print): Adjust.
* regcache.h (regcache): Remove constructors.

Replace regcache::dump with class register_dump

Nowadays, we need to dump registers contents from "readwrite" regcache and
"readonly" regcache,

  if (target_has_registers)
    get_current_regcache ()->dump (out, what_to_dump);
  else
    {
      /* For the benefit of "maint print registers" & co when
         debugging an executable, allow dumping a regcache even when
         there is no thread selected / no registers.  */
      regcache dummy_regs (target_gdbarch ());
      dummy_regs.dump (out, what_to_dump);
    }

since we'll have two different types/classes for "readwrite" regcache and
"readonly" regcache, we have to move dump method to their parent class,
reg_buffer.  However, the functionality of "dump" looks unnecessary to
reg_buffer (because some dump modes like regcache_dump_none,
regcache_dump_remote and regcache_dump_groups don't need reg_buffer at
all, they need gdbarch to do the dump), so I decide to move "dump" into a
separate classes, and each sub-class is about each mode of dump.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* regcache.c (class register_dump): New class.
(register_dump_regcache, register_dump_none): New class.
(register_dump_remote, register_dump_groups): New class.
(regcache_print): Update.
* regcache.h (regcache_dump_what): Move it to regcache.c.
(regcache) <dump>: Remove.

Class detached_regcache

jit.c uses the regcache in a slightly different way, the regcache dosen't
write through to target, but it has read and write methods.  If I apply
regcache in record-full.c, it has the similar use pattern.  This patch
adds a new class detached_regcache, a register buffer, but can be
read and written.

Since jit.c doesn't want to write registers through to target, it uses
regcache as a readonly regcache (because only readonly regcache
disconnects from the target), but it adds a hole in regcache
(raw_set_cached_value) in order to modify a readonly regcache.  This patch
fixes this hole completely.

regcache inherits detached_regcache, and detached_regcache inherits
readable_regcache.  The ideal design is that both detached_regcache and
readable_regcache inherit reg_buffer, and regcache inherit
detached_regcache and regcache_read (virtual inheritance).  I concern
about the performance overhead of virtual inheritance, so I don't do it in
the patch.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* jit.c (struct jit_unwind_private) <regcache>: Change its type to
 reg_buffer_rw *.
(jit_unwind_reg_set_impl): Call raw_supply.
(jit_frame_sniffer): Use reg_buffer_rw.
* record-full.c (record_full_core_regbuf): Change its type.
(record_full_core_open_1): Use reg_buffer_rw.
(record_full_close): Likewise.
(record_full_core_fetch_registers): Use regcache->raw_supply.
(record_full_core_store_registers): Likewise.
* regcache.c (regcache::get_register_status): Move it to
reg_buffer.
(regcache_raw_set_cached_value): Remove.
(regcache::raw_set_cached_value): Remove.
(regcache::raw_write): Call raw_supply.
(regcache::raw_supply): Move it to reg_buffer_rw.
* regcache.h (regcache_raw_set_cached_value): Remove.
(reg_buffer_rw): New class.

Class readonly_detached_regcache

This patch adds a new class (type) for readonly regcache, which is
created via regcache::save.  readonly_detached_regcache inherits
readable_regcache.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* dummy-frame.c (dummy_frame_cache) <prev_regcache>: Use
readonly_detached_regcache.
(dummy_frame_prev_register): Use regcache->cooked_read.
* frame.c (frame_save_as_regcache): Change return type.
(frame_pop): Update.
* frame.h (frame_save_as_regcache): Update declaration.
* inferior.h (get_infcall_suspend_state_regcache): Update
declaration.
* infrun.c (infcall_suspend_state) <registers>: use
readonly_detached_regcache.
(save_infcall_suspend_state): Don't use regcache_dup.
(get_infcall_suspend_state_regcache): Change return type.
* linux-fork.c (struct fork_info) <savedregs>: Change to
readonly_detached_regcache.
<pc>: New field.
(fork_save_infrun_state): Don't use regcache_dup.
(info_checkpoints_command): Adjust.
* mi/mi-main.c (register_changed_p): Update declaration.
(mi_cmd_data_list_changed_registers): Use
readonly_detached_regcache.
(register_changed_p): Change parameter type to
readonly_detached_regcache.
* ppc-linux-tdep.c (ppu2spu_cache) <regcache>: Use
readonly_detached_regcache.
(ppu2spu_sniffer): Construct a new readonly_detached_regcache.
* regcache.c (readonly_detached_regcache::readonly_detached_regcache):
New.
(regcache::save): Move it to reg_buffer.
(regcache::restore): Change parameter type.
(regcache_dup): Remove.
* regcache.h (reg_buffer) <save>: New method.
(readonly_detached_regcache): New class.
* spu-tdep.c (spu2ppu_cache) <regcache>: Use
readonly_detached_regcache.
(spu2ppu_sniffer): Construct a new readonly_detached_regcache.

Remove regcache_save and regcache_cpy

... instead we start to use regcache methods save and restore.  It is
quite straightforward to replace regcache_save with regcache->save.

regcache_cpy has some asserts, some of them not necessary, like

 gdb_assert (src != dst);

because we already assert !m_readonly_p and src->m_readonly_p, so
src isn't dst.  Some of the asserts are moved to ::restore.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* frame.c (frame_save_as_regcache): Use regcache method save.
(frame_pop): Use regcache method restore.
* infrun.c (restore_infcall_suspend_state): Likewise.
* linux-fork.c (fork_load_infrun_state): Likewise.
* ppc-linux-tdep.c (ppu2spu_sniffer): User regcache method
save.
* regcache.c (regcache_save): Remove.
(regcache::restore): More asserts.
(regcache_cpy): Remove.
* regcache.h (regcache_save): Remove the declaration.
(regcache::restore): Move from private to public.
Remove the friend declaration of regcache_cpy.
(regcache_cpy): Remove declaration.

class readable_regcache and pass readable_regcache to gdbarch pseudo_register_read and pseudo_register_read_value

pseudo registers are either from raw registers or memory, so
gdbarch methods pseudo_register_read and pseudo_register_read_value
should have regcache object which only have read methods.  In other
words, we should disallow writing to regcache in these two gdbarch
methods.  In order to apply this restriction, this patch adds a new
class readable_regcache, derived from reg_buffer, and it only has
raw_read and cooked_read methods.  regcache is derived from
readable_regcache.  This patch also passes readable_regcache instead of
regcache to gdbarch methods pseudo_register_read and
pseudo_register_read_value.

This patch moves raw_read* and cooked_read* methods to readable_regcache,
which is straightforward.  One thing not straightforward is that I split
regcache::xfer_part to readable_regcache::read_part and regcache::write_part,
because readable_regcache can only have methods to read.

readable_regcache is an abstract base class, and it has a pure virtual
function raw_update, because I don't want readable_regcache know where
these raw registers are from.  They can be from either the target
(readwrite regcache) or the regcache itself (readonly regcache).

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* aarch64-tdep.c (aarch64_pseudo_register_read_value): Change
parameter type to 'readable_regcache *'.
* amd64-tdep.c (amd64_pseudo_register_read_value): Likewise.
* arm-tdep.c (arm_neon_quad_read): Likewise.
(arm_pseudo_read): Likewise.
* avr-tdep.c (avr_pseudo_register_read): Likewise.
* bfin-tdep.c (bfin_pseudo_register_read): Likewise.
* frv-tdep.c (frv_pseudo_register_read): Likewise.
* gdbarch.c: Re-generated.
* gdbarch.h: Re-generated.
* gdbarch.sh (pseudo_register_read): Change parameter type to
'readable_regcache *'.
(pseudo_register_read_value): Likewise.
* h8300-tdep.c (pseudo_from_raw_register): Likewise.
(h8300_pseudo_register_read): Likewise.
* hppa-tdep.c (hppa_pseudo_register_read): Likewise.
* i386-tdep.c (i386_mmx_regnum_to_fp_regnum): Likewise.
(i386_pseudo_register_read_into_value): Likewise.
(i386_pseudo_register_read_value): Likewise.
* i386-tdep.h (i386_pseudo_register_read_into_value): Update
declaration.
* ia64-tdep.c (ia64_pseudo_register_read): Likewise.
* m32c-tdep.c (m32c_raw_read): Likewise.
(m32c_read_flg): Likewise.
(m32c_banked_register): Likewise.
(m32c_banked_read): Likewise.
(m32c_sb_read): Likewise.
(m32c_part_read): Likewise.
(m32c_cat_read): Likewise.
(m32c_r3r2r1r0_read): Likewise.
(m32c_pseudo_register_read): Likewise.
* m68hc11-tdep.c (m68hc11_pseudo_register_read): Likewise.
* mep-tdep.c (mep_pseudo_cr32_read): Likewise.
(mep_pseudo_cr64_read): Likewise.
(mep_pseudo_register_read): Likewise.
* mips-tdep.c (mips_pseudo_register_read): Likewise.
* msp430-tdep.c (msp430_pseudo_register_read): Likewise.
* nds32-tdep.c (nds32_pseudo_register_read): Likewise.
* regcache.c (regcache::raw_read): Move it to readable_regcache.
(regcache::cooked_read): Likewise.
(regcache::cooked_read_value): Likewise.
(regcache_cooked_read_signed):
(regcache::cooked_read): Likewise.
* regcache.h (readable_regcache): New class.
(regcache): Inherit readable_regcache.  Move some methods to
readable_regcache.
* rl78-tdep.c (rl78_pseudo_register_read): Change
parameter type to 'readable_regcache *'.
* rs6000-tdep.c (do_regcache_raw_read): Remove.
(e500_pseudo_register_read): Change parameter type to
'readable_regcache *'.
(dfp_pseudo_register_read): Likewise.
(vsx_pseudo_register_read): Likewise.
(efpr_pseudo_register_read): Likewise.
* s390-tdep.c (s390_pseudo_register_read): Likewise.
* sh-tdep.c (sh_pseudo_register_read): Likewise.
* sh64-tdep.c (pseudo_register_read_portions): Likewise.
(sh64_pseudo_register_read): Likewise.
* sparc-tdep.c (sparc32_pseudo_register_read): Likewise.
* sparc64-tdep.c (sparc64_pseudo_register_read): Likewise.
* spu-tdep.c (spu_pseudo_register_read_spu): Likewise.
(spu_pseudo_register_read): Likewise.
* xtensa-tdep.c (xtensa_register_read_masked): Likewise.
(xtensa_pseudo_register_read): Likewise.

Class reg_buffer

This patch adds a new class reg_buffer, and regcache inherits it.  Class
reg_buffer is a very simple class, which has the buffer for register
contents and status only.  It doesn't have any methods to set contents and
status, and it is expected that its children classes can inherit it and
add different access methods.

Another reason I keep class reg_buffer so simple is that I think
reg_buffer can be even reused in other classes which need to record the
registers contents and status, like frame cache for example.

gdb:

2018-02-21  Yao Qi  <yao.qi@linaro.org>

* regcache.c (regcache::regcache): Call reg_buffer ctor.
(regcache::arch): Move it to reg_buffer::arch.
(regcache::register_buffer): Likewise.
(regcache::assert_regnum): Likewise.
(regcache::num_raw_registers): Likewise.
* regcache.h (reg_buffer): New class.
(regcache): Inherit reg_buffer.

ia64 testsuite changes for --gc-sections

* testsuite/ld-elf/group8b.d: Run test on ia64.  Use xfail rather
than notarget.
* testsuite/ld-elf/pr12851.d: Likewise.
* testsuite/ld-elf/pr12975.d: Likewise.
* testsuite/ld-elf/pr13177.d: Likewise.
* testsuite/ld-elf/pr13195.d: Likewise.
* testsuite/ld-elf/pr17615.d: Likewise.
* testsuite/ld-elf/pr21562a.d: Likewise.
* testsuite/ld-elf/pr21562b.d: Likewise.
* testsuite/ld-elf/pr21562c.d: Likewise.
* testsuite/ld-elf/pr21562d.d: Likewise.
* testsuite/ld-elf/pr21562i.d: Likewise.
* testsuite/ld-elf/pr21562j.d: Likewise.
* testsuite/ld-elf/pr21562k.d: Likewise.
* testsuite/ld-elf/pr21562l.d: Likewise.
* testsuite/ld-elf/pr21562m.d: Likewise.
* testsuite/ld-elf/pr21562n.d: Likewise.
* testsuite/ld-elf/group9a.d: Run test on ia64 and alpha.  Use xfail
rather than notarget.
* testsuite/ld-elf/group9b.d: Likewise.
* testsuite/ld-elf/pr22677.d: Likewise.

Automatic date update in version.in

MIPS16/GAS/testsuite: Add cross-section R_MIPS16_PC16_S1 relocation tests

Add a pair of MIPS16 branch tests to verify correct R_MIPS16_PC16_S1
relocation generation for cross-section references in a single source.
This complements commit c9775dde3277 ("MIPS16: Add R_MIPS16_PC16_S1
branch relocation support").

gas/
* testsuite/gas/mips/mips16-branch-reloc-4.d: New test.
* testsuite/gas/mips/mips16-branch-reloc-5.d: New test.
* testsuite/gas/mips/mips16-branch-reloc-4.s: New test source.
* testsuite/gas/mips/mips16-branch-reloc-5.s: New test source.
* testsuite/gas/mips/mips.exp: Run the new tests.

MIPS16/opcodes: Free up `M' operand code

The `M' and `m' MIPS16 operand codes are functionally the same, denoting
a 7-bit register list that is encoded the same way for both SAVE and
RESTORE.  Use `m' for both instructions then, making `M' available for a
different use.

opcodes/
* mips16-opc.c (decode_mips16_operand) <'M'>: Remove case.
(mips16_opcodes): Replace `M' with `m' for "restore".

include/
* opcode/mips.h: Remove `M' operand code.

gas: xtensa: limit size of auto litpools

Literal movement code may grow auto litpool so big that it won't be
possible to jump around it. Limit the size of auto litpools by 1/2 of
the jump range.

gas/
2018-02-20  Max Filippov  <jcmvbkbc@gmail.com>

* config/tc-xtensa.c (struct litpool_frag): Add new field
literal_count.
(MAX_AUTO_POOL_LITERALS, MAX_EXPLICIT_POOL_LITERALS)
(MAX_POOL_LITERALS): New macro definitions.
(auto_litpool_limit): Initialize to 0.
(md_parse_option): Set auto_litpool_limit in the presence of
--auto-litpools option.
(xtensa_maybe_create_literal_pool_frag): Zero-initialize
literal_count field.
(xg_find_litpool): New function. Make sure that found literal
pool size is within the limit.
(xtensa_move_literals): Extract literal pool search code into
the new function.
* testsuite/gas/xtensa/all.exp: Add auto-litpools-2 test.
* testsuite/gas/xtensa/auto-litpools-2.d: New file.
* testsuite/gas/xtensa/auto-litpools-2.s: New file.
* testsuite/gas/xtensa/auto-litpools.d: Fix up changed
addresses.
* testsuite/gas/xtensa/auto-litpools.s: Change literal value so
that objdump doesn't get out of sync.

remote-sim: Add missing ATTRIBUTE_PRINTF

Fixes:

/home/emaisin/src/binutils-gdb/gdb/remote-sim.c:385:34: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
  vfprintf_filtered (gdb_stdout, format, args);
                                 ^~~~~~
/home/emaisin/src/binutils-gdb/gdb/remote-sim.c:394:34: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
  vfprintf_filtered (gdb_stdout, format, ap);
                                 ^~~~~~
/home/emaisin/src/binutils-gdb/gdb/remote-sim.c:402:34: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
  vfprintf_filtered (gdb_stderr, format, ap);
                                 ^~~~~~
/home/emaisin/src/binutils-gdb/gdb/remote-sim.c:413:11: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
  verror (format, args);
          ^~~~~~
4 errors generated.

gdb/ChangeLog:

* remote-sim.c (gdb_os_printf_filtered, gdb_os_vprintf_filtered,
gdb_os_evprintf_filtered, gdb_os_error): Add ATTRIBUTE_PRINTF.

Fix typo in listing of objcopy's command line options.

binutils* doc/binutils.texi (objcopy): Add missing closing square
parenthesis to listing of objcopy's command line options.

Enable link time garbage collection support for the IA64 target.

As suggested in long ago in a galaxy far, far away [1], I tried
turning it on and it seems to work, as is evidenced by the Mesa
package in Debian/ia64.  Please enable it with the following patch.
[1] https://sourceware.org/ml/binutils/2007-07/msg00241.html

bfd * elfnn-ia64.c (elf_backend_can_gc_sections): Enable.

gnulib: import mkstemp

Older versions of MinGW do not support mkstemp causing:

    gdb/unittests/scoped_fd-selftests.c:37:29: error: \
    'mkstemp' was not declared in this scope
       int fd = mkstemp (filename);
                             ^
    gdb/unittests/scoped_fd-selftests.c: In function 'void
    selftests::scoped_fd::test_release()':
    gdb/unittests/scoped_fd-selftests.c:56:29: error: \
    'mkstemp' was not declared in this scope
       int fd = mkstemp (filename);
                             ^

Import mkstemp from gnulib.

gdb/
* gnulib/update-gnulib.sh (IMPORTED_GNULIB_MODULES): Add mkstemp.
* gnulib/aclocal.m4: Regenerated.
* gnulib/config.in: Regenerated.
* gnulib/configure: Regenerated.
* gnulib/import/Makefile.am: Regenerated.
* gnulib/import/Makefile.in: Regenerated.
* gnulib/import/m4/gnulib-cache.m4: Regenerated.
* gnulib/import/m4/gnulib-comp.m4: Regenerated.
* gnulib/import/m4/mkstemp.m4: Imported.
* gnulib/import/m4/secure_getenv.m4: Imported.
* gnulib/import/m4/tempname.m4: Imported.
* gnulib/import/mkstemp.c: Imported.
* gnulib/import/secure_getenv.c: Imported.
* gnulib/import/tempname.c: Imported.
* gnulib/import/tempname.h: Imported.

btrace, testsuite: do not force BTS

In gdb.btrace/buffer-size.exp we explicitly ask for the BTS recording format.
This may lead to spurious fails on systems where PT is being used by some other
process at the same time.

Set both PT and BTS buffer sizes to 1 and check that whatever recording format
is used will use a 4KB buffer.

testsuite/
* gdb.btrace/buffer-size.exp: Do not force BTS.

Clarify .arch_extension possible values

Documentation for .arch_extension says it accepts the same architectural
extensions as those accepted by -mcpu. Given the name and the fact that
-march for obvious reason also accept the same extensions, I believe
it's worth mentioning that it accepts the same extensions as both
-march and -mcpu. This commit addresses that.

2018-02-20  Thomas Preud'homme  <thomas.preudhomme@arm.com>

gas/
* doc/c-arm.texi (.arch_extension): Mention extensions it accepts are
also the same as -march.

Fix make 3.81 build errors

gdbserver/
* Makefile.in: Switch order of make rules.

Automatic date update in version.in

gdb/doc: Additional information about 'info line'

Extend the documentation of 'info line' command to:

    1. Make 'info line' with no argument more obvious, and make it clearer
       what this does.

    2. Cover what happens when a secod 'info line' with no argument is
       issued.

    3. Extend the example output for 'info line ...' to include
       symbolic addresses.

gdb/doc/ChangeLog:

* gdb.texinfo (Machine Code): Additional information about "info
line" command.

Revert "_bfd_mips_elf_final_link: Notify user about wrong .reginfo size"

Revert commit 58807c48a5a3 ("_bfd_mips_elf_final_link: Notify user about
wrong .reginfo size") now that the size of the `.reginfo' section has
been truly fixed in `_bfd_mips_elf_section_processing', meaning that the
offending condition can be asserted again.

bfd/
Revert
2018-01-12  Vlad Ivanov  <vlad@ivanov.email>

* elfxx-mips.c (_bfd_mips_elf_final_link): Notify user when
.reginfo section has wrong size.

MIPS/BFD: Fix the size of `.reginfo' and `.MIPS.abiflags' sections

Use the SEC_FIXED_SIZE flag to actually fix the size of `.reginfo' and
`.MIPS.abiflags' sections in `_bfd_mips_elf_always_size_sections', as
originally intended, removing link failures such as:

ld: final link failed: Section has no contents

or:

ld: final link failed: Bad value

or:

ld: foo: .reginfo section size should be 24 bytes, actual size is 32

and assertion failures like:

ld: BFD (GNU Binutils) 2.30.51.20180131 assertion fail .../bfd/elfxx-mips.c:14322

in link scenarios involving a linker script that either creates an
output `.reginfo' or `.MIPS.abiflags' section from scratch or produces
either section from different sections.  If such an output section's
size turns out to be incorrect according to the psABI, then the section
is either truncated or padded out to the correct size, as relevant.

This allows people to handle these sections in a link in an unusual way,
while still addressing the issue covered by commit 58807c48a5a3
("_bfd_mips_elf_final_link: Notify user about wrong .reginfo size").

The original arrangement, coming from an unindentified change made to
what was called `mips_elf_always_size_sections' back then, between
commit 02650bd0a97e ("This adds ABI flags to MIPS/ELF object files.")
and commit 252b5132c753 ("19990502 sourceware import"), also missing
from BFD ChangeLog files, assumed that the output section size is not
going to change after return from `bfd_elf_size_dynamic_sections', the
caller of that function, called in turn from `ldemul_before_allocation'
via `gld${EMULATION_NAME}_before_allocation' in ld/emultempl/elf32.em,
and ultimately from `lang_process'.  This is because later on in
`lang_process' processing `lang_size_sections' is called , happily
recalculating the section size, and it has actually already been the
case at the time of commit 252b5132c753 ("19990502 sourceware import"),
so the assumption was clearly incorrect right from the beginning.

bfd/
* elfxx-mips.c (_bfd_mips_elf_always_size_sections): Set
SEC_FIXED_SIZE and SEC_HAS_CONTENTS flags for `.reginfo' and
`.MIPS.abiflags' sections.
(_bfd_mips_elf_final_link): Avoid reading beyond `.reginfo'
section's end.

ld/
* testsuite/ld-mips-elf/reginfo-0.d: New test.
* testsuite/ld-mips-elf/reginfo-0r.d: New test.
* testsuite/ld-mips-elf/reginfo-1.d: New test.
* testsuite/ld-mips-elf/reginfo-1r.d: New test.
* testsuite/ld-mips-elf/reginfo-2.d: New test.
* testsuite/ld-mips-elf/reginfo-2r.d: New test.
* testsuite/ld-mips-elf/mips-abiflags-0.d: New test.
* testsuite/ld-mips-elf/mips-abiflags-0r.d: New test.
* testsuite/ld-mips-elf/mips-abiflags-1.d: New test.
* testsuite/ld-mips-elf/mips-abiflags-1r.d: New test.
* testsuite/ld-mips-elf/mips-abiflags-2.d: New test.
* testsuite/ld-mips-elf/mips-abiflags-2r.d: New test.
* testsuite/ld-mips-elf/reginfo-0.ld: New test linker script.
* testsuite/ld-mips-elf/reginfo-1.ld: New test linker script.
* testsuite/ld-mips-elf/mips-abiflags-0.ld: New test linker
script.
* testsuite/ld-mips-elf/mips-abiflags-1.ld: New test linker
script.
* testsuite/ld-mips-elf/reginfo-1.s: New test source.
* testsuite/ld-mips-elf/reginfo-2.s: New test source.
* testsuite/ld-mips-elf/mips-elf.exp: Run the new tests.

LD: Support fixed-size sections some psABIs may require

Define a SEC_FIXED_SIZE section flag for target backends to use for
output sections whose size has been fixed in the psABI.  The size of
such sections will not be changed anyhow by the generic linker and it is
up to the target backend to get their size right.

bfd/
* section.c (SEC_FIXED_SIZE): New macro.
* bfd-in2.h: Regenerate.

ld/
* ldlang.c (insert_pad): Do not change output section's size if
SEC_FIXED_SIZE is set in the flags.
(size_input_section): Likewise.
(lang_size_sections_1): Likewise.
(lang_reset_memory_regions): Likewise.

BFD: Remove unused SEC_HAS_GOT_REF section flag

Remove the SEC_HAS_GOT_REF section flag no longer in use since commit
a252afa4cdff ("Fix linking of PIC code on PA"),
<https://sourceware.org/ml/binutils/2003-08/msg00467.html>, to make the
bit position available for reuse.

bfd/
* section.c (SEC_HAS_GOT_REF): Remove macro.
* bfd-in2.h: Regenerate.

Add attribute printf to _bfd_error_handler

and fix a few stray errors.

* elf-attrs.c (_bfd_elf_parse_attributes): Correct _bfd_error_handler
arguments.
* elfxx-mips.c (_bfd_mips_elf_final_link): Likewise.
* elfnn-riscv.c (_bfd_riscv_relax_align): Likewise.
(_bfd_riscv_relax_pc): Likewise and fix typos.
* libbfd-in.h (_bfd_error_handler): Add attribute printf.
* libbfd.h: Regenerate.

Don't use %ll

* dwarf2.c (read_section): Don't use 'll' format modifier.
(find_abstract_instance): Likewise.
* elfcore.h (elf_core_file_p): Likewise.

%L conversions

* bfd-in.h: Include inttypes.h or if not available define
PRId64, PRIu64 and PRIx64.
* bfd.c (_bfd_doprnt, _bfd_doprnt_scan): Remove support for L
modifier to print bfd_vma.
* coff-arm.c, * coff-mcore.c, * coff-ppc.c, * coff-rs6000.c,
* coff-sh.c, * coff-tic80.c, * coffcode.h, * coffgen.c, * cofflink.c,
* compress.c, * dwarf2.c, * elf-m10300.c, * elf.c, * elf32-arc.c,
* elf32-arm.c, * elf32-bfin.c, * elf32-cris.c, * elf32-hppa.c,
* elf32-i386.c, * elf32-ip2k.c, * elf32-lm32.c, * elf32-m32r.c,
* elf32-m68k.c, * elf32-metag.c, * elf32-nds32.c, * elf32-nios2.c,
* elf32-ppc.c, * elf32-rx.c, * elf32-s390.c, * elf32-score.c,
* elf32-score7.c, * elf32-sh.c, * elf32-sh64.c, * elf32-spu.c,
* elf32-tic6x.c, * elf32-tilepro.c, * elf32-v850.c, * elf32-vax.c,
* elf32-xtensa.c, * elf64-alpha.c, * elf64-hppa.c, * elf64-ia64-vms.c,
* elf64-mmix.c, * elf64-s390.c, * elf64-sh64.c, * elf64-x86-64.c,
* elfcode.h, * elfcore.h, * elflink.c, * elfnn-aarch64.c,
* elfnn-ia64.c, * elfnn-riscv.c, * elfxx-mips.c, * elfxx-sparc.c,
* elfxx-tilegx.c, * ieee.c, * ihex.c, * mach-o.c, * merge.c, * mmo.c,
* peXXigen.c, * xcofflink.c: Replace use of Lx modifier with PRIx64,
and cast input to uint64_t, and similarly for Ld and Lu.
* bfd-in2.h: Regenerate.

Use %pI, %pR, %pS, %pT in place of %I, %R, %S and %T.

bfd/
* elf32-arm.c, * elf32-hppa.c, * elf32-lm32.c, * elf32-m32r.c,
* elf32-metag.c, * elf32-nds32.c, * elf32-or1k.c, * elf32-ppc.c,
* elf32-s390.c, * elf32-sh.c, * elf32-tic6x.c, * elf32-tilepro.c,
* elf64-ppc.c, * elf64-s390.c, * elflink.c, * elfnn-aarch64.c,
* elfnn-riscv.c, * elfxx-sparc.c, * elfxx-tilegx.c, * elfxx-x86.c,
* reloc.c: Replace use of %R and %T in format strings passed to
einfo and friends by %pR and %pT.
ld/
* ldmisc.c (vfinfo) Handle %pI, %pR, %pS and %pT in place of
%I, %R, %S and %T.
* ldcref.c, * ldctor.c, * ldemul.c, * ldexp.c, * ldgram.y,
* ldlang.c, * ldlex.l, * ldmain.c, * ldmisc.c, * pe-dll.c,
* emultempl/sh64elf.em: Replace use of of %I, %R, %S and %T in
format strings passed to einfo and friends by %pI, %pR, %pS and %pT.

Use %pA and %pB in messages rather than %A and %B

First step towards compiler verification of _bfd_error_handler
arguments, and better verification of translated messages.

bfd/
* bfd.c (_bfd_doprnt, _bfd_doprnt_scan): Handle %pA and %pB in place
of %A and %B.
* aout-adobe.c: Update all messages using %A and %B.
* aout-cris.c: Likewise.
* aoutx.h: Likewise.
* archive.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.
* compress.c: Likewise.
* cpu-arm.c: Likewise.
* ecoff.c: Likewise.
* elf-attrs.c: Likewise.
* elf-eh-frame.c: Likewise.
* elf-ifunc.c: Likewise.
* elf-m10300.c: Likewise.
* elf-properties.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-sparc.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-wasm32.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.
* elfnn-riscv.c: Likewise.
* elfxx-mips.c: Likewise.
* elfxx-sparc.c: Likewise.
* elfxx-tilegx.c: Likewise.
* elfxx-x86.c: Likewise.
* hpux-core.c: Likewise.
* ieee.c: Likewise.
* ihex.c: Likewise.
* libbfd.c: Likewise.
* linker.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.
* peicode.h: Likewise.
* reloc.c: Likewise.
* rs6000-core.c: Likewise.
* srec.c: Likewise.
* stabs.c: Likewise.
* vms-alpha.c: Likewise.
* xcofflink.c: Likewise.
ld/
* ldmisc.c (vfinfo): Handle %pA and %pB in place of %A and %B.
* ldcref.c: Update all messages using %A and %B.
* ldexp.c: Likewise.
* ldlang.c: Likewise.
* ldmain.c: Likewise.
* ldmisc.c: Likewise.
* pe-dll.c: Likewise.
* plugin.c: Likewise.
* emultempl/beos.em: Likewise.
* emultempl/cr16elf.em: Likewise.
* emultempl/elf32.em: Likewise.
* emultempl/m68kcoff.em: Likewise.
* emultempl/m68kelf.em: Likewise.
* emultempl/mmo.em: Likewise.
* emultempl/nds32elf.em: Likewise.
* emultempl/pe.em: Likewise.
* emultempl/pep.em: Likewise.
* emultempl/spuelf.em: Likewise.
* emultempl/sunos.em: Likewise.
* emultempl/xtensaelf.em: Likewise.

ld: Add -z separate-code tests to frame.exp

On x86, "-z separate-code" leads to assertion fail at bfd/elf.c:5917.
Alsp skip tests if -shared isn't supported.

PR ld/22845
* testsuite/ld-elf/frame.exp: Skip if -shared isn't supported.
Add tests for "-z noseparate-code" and "-z separate-code".
Remove unsupported -shared check.

[ARM] Fix bxns mask

Bit 7 of BXNS is a fixed bit which distinguish it from BLXNS. Yet it is
not set in the disassembler entry mask. This commit fixes that.

2018-02-19  Thomas Preud'homme  <thomas.preudhomme@arm.com>

opcodes/
* arm-dis.c (thumb_opcodes): Fix BXNS mask.

Fix mistake in the declaration of the --include-all-whitespace option of the strings utility.

* strings.c (long_options): Include-all-whitespace does not take
        an extra agument.

Add common/ dir in build directories

gdb/
* Makefile.in: (COMMON_SFILES): Add common/*.c files.
(SFILES): Remove common/*.c files.
(COMMON_OBS): Remove some *.o files built from common/*.c files.
* common/common.host: Add common reference.
* configure.ac: Likewise.
* configure: Regenerate.

gdbserver/
* Makefile.in: Add common directory in build.
* configure.ac: Add common reference.
* configure: Regenerate.

PT_LOAD and PT_GNU_RELRO segment overlap

Commit 325ba6fb34 excluded degenerate zero length PT_LOAD segments,
but that only fixed part of the problem, which was that the load
segment limits were not calculated properly.

PR 22845
* elf.c (IS_TBSS): Define.
(_bfd_elf_map_sections_to_segments): Use IS_TBSS.
(assign_file_positions_for_non_load_sections): Revert last change.
Properly calculate load segment limits to compare against relro limits.

Automatic date update in version.in

Automatic date update in version.in

ld-elf/ehdr_start: Pass --build-id to ld

ld-elf/ehdr_start fails with -z separate-code.  Since there is no data
LOAD segment before code LOAD segment:

There are 2 program headers, starting at offset 64

Program Headers:
  Type           Offset   VirtAddr           PhysAddr           FileSiz  MemSiz   Flg Align
  LOAD           0x200000 0x0000000000600000 0x0000000000600000 0x000010 0x000010 R E 0x200000
  LOAD           0x400000 0x0000000000800000 0x0000000000800000 0x000008 0x000008 R   0x200000

 Section to Segment mapping:
  Segment Sections...
   00     .text
   01     .rodata

the program header isn't included in any LOAD segment.  As the result,
reference to __ehdr_start is resolved to zero.  Pass --build-id to ld
to add a data LOAD segment before code LOAD segment to put the program
header in the  data LOAD segment.

PR ld/22845
* testsuite/ld-elf/ehdr_start.d: Pass --build-id to ld.

ld: Add -z separate-code test for zero size section

PR ld/22845
* testsuite/ld-elf/binutils.exp (tls_opts): Add tests for
"-z noseparate-code" and "-z separate-code".

ld-elf/eh4: Pass -z max-page-size=0x200000 -z noseparate-code to ld

-z separate-code creates separate code LOAD segment, aligns it to the
maximum page size and places .plt section before .text section.  But
ld-elf/eh4 passes -Ttext 0x400 to linker to place .text section at
address 0x400, which is impossible for linker to accomplish:

$ ld -shared -Ttext 0x400 -z separate-code -o x.so eh4.o
ld: section .eh_frame LMA [0000000000200000,000000000020006b] overlaps section .plt LMA [0000000000200000,000000000020001f]

Since ld-elf/eh4 also checks exact addresses, this patch passes
-z max-page-size=0x200000 -z noseparate-code to ld.

PR ld/22845
* ld-elf/eh4.d: Pass -z max-page-size=0x200000 -z noseparate-code
to ld.

Add .nop assembler directive

Implement the '.nop SIZE[, CONTROL]' assembler directive, which emits
SIZE bytes filled with no-op instructions.  SIZE is absolute expression.
The optional CONTROL byte controls how no-op instructions should be
generated.  If the comma and @var{control} are omitted, CONTROL is
assumed to be zero.

For Intel 80386 and AMD x86-64 targets, CONTROL byte specifies the size
limit of a single no-op instruction.  The valid values of CONTROL byte
are between 0 and 8 for 16-bit mode, between 0 and 10 for 32-bit mode,
between 0 and 11 for 64-bit mode.  When 0 is used, the no-op size limit
is set to the maximum supported size.

2 new relax states, rs_space_nop and rs_fill_nop, are added to enum
_relax_state, which are similar to rs_space and rs_fill, respectively,
but they fill with no-op instructions, instead of a single byte.  A
target backend must override the default md_generate_nops to generate
proper no-op instructions.  Otherwise, an error of unimplemented .nop
directive will be issued whenever .nop directive is used.

* NEWS: Mention .nop directive.
* as.h (_relax_state): Add rs_space_nop and rs_fill_nop.
* read.c (potable): Add .nop.
(s_nop): New function.
* read.h (s_nop): New prototype.
* write.c (cvt_frag_to_fill): Handle rs_space_nop and
rs_fill_nop.
(md_generate_nops): New function.
(relax_segment): Likewise.
(write_contents): Use md_generate_nops for rs_fill_nop.
* config/tc-i386.c (alt64_11): New.
(alt64_patt): Likewise.
(md_convert_frag): Handle rs_space_nop.
(i386_output_nops): New function.
(i386_generate_nops): Likewise.
(i386_align_code): Call i386_output_nops.
* config/tc-i386.h (i386_generate_nops): New.
(md_generate_nops): Likewise.
* doc/as.texinfo: Document .nop directive.
* testsuite/gas/i386/i386.exp: Run .nop directive tests.
* testsuite/gas/i386/nop-1.d: New file.
* testsuite/gas/i386/nop-1.s: Likewise.
* testsuite/gas/i386/nop-2.d: Likewise.
* testsuite/gas/i386/nop-2.s: Likewise.
* testsuite/gas/i386/nop-3.d: Likewise.
* testsuite/gas/i386/nop-3.s: Likewise.
* testsuite/gas/i386/nop-4.d: Likewise.
* testsuite/gas/i386/nop-4.s: Likewise.
* testsuite/gas/i386/nop-5.d: Likewise.
* testsuite/gas/i386/nop-5.s: Likewise.
* testsuite/gas/i386/nop-6.d: Likewise.
* testsuite/gas/i386/nop-6.s: Likewise.
* testsuite/gas/i386/nop-bad-1.l: Likewise.
* testsuite/gas/i386/nop-bad-1.s: Likewise.
* testsuite/gas/i386/x86-64-nop-1.d: Likewise.
* testsuite/gas/i386/x86-64-nop-2.d: Likewise.
* testsuite/gas/i386/x86-64-nop-3.d: Likewise.
* testsuite/gas/i386/x86-64-nop-4.d: Likewise.
* testsuite/gas/i386/x86-64-nop-5.d: Likewise.
* testsuite/gas/i386/x86-64-nop-6.d: Likewise.

Automatic date update in version.in

Ignore degenerate PT_LOAD segments

Fixes a failure triggered by -z separate-code.  p_memsz is tested
rather than p_filesz for objcopy --only-keep-debug where p_filesz is
set to zero.

PR 22845
* elf.c (assign_file_positions_for_non_load_sections): Ignore
degenerate zero size PT_LOAD segments when finding one overlapping
the PT_GNU_RELRO segment.

x86-64: Add -z max-page-size=0x200000 -z noseparate-code to linker tests

Add -z max-page-size=0x200000 -z noseparate-code since these tests
check for exact addresses.

* testsuite/ld-x86-64/bnd-branch-1-now.d: Add  -z
max-page-size=0x200000 -z noseparate-code.
* testsuite/ld-x86-64/bnd-ifunc-1-now.d: Likewise.
* testsuite/ld-x86-64/bnd-ifunc-2-now.d: Likewise.
* testsuite/ld-x86-64/bnd-ifunc-2.d: Likewise.
* testsuite/ld-x86-64/bnd-plt-1-now.d: Likewise.
* testsuite/ld-x86-64/bnd-plt-1.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-1-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-1.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2a-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2a.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2b-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2b.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2c-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2c.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2d-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-2d.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3a-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3a.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3b-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3b.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3c-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3c.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3d-x32.d: Likewise.
* testsuite/ld-x86-64/ibt-plt-3d.d: Likewise.
* testsuite/ld-x86-64/load1a.d: Likewise.
* testsuite/ld-x86-64/load1b.d: Likewise.
* testsuite/ld-x86-64/load1c.d: Likewise.
* testsuite/ld-x86-64/load1d.d: Likewise.
* testsuite/ld-x86-64/pie3.d: Likewise.
* testsuite/ld-x86-64/pr14207.d: Likewise.
* testsuite/ld-x86-64/pr17618.d: Likewise.
* testsuite/ld-x86-64/pr19162.d: Likewise.
* testsuite/ld-x86-64/pr19636-2d.d: Likewise.
* testsuite/ld-x86-64/pr19636-2l.d: Likewise.
* testsuite/ld-x86-64/pr20253-1b.d: Likewise.
* testsuite/ld-x86-64/pr20253-1d.d: Likewise.
* testsuite/ld-x86-64/pr20253-1f.d: Likewise.
* testsuite/ld-x86-64/pr20253-1h.d: Likewise.
* testsuite/ld-x86-64/pr20253-1j.d: Likewise.
* testsuite/ld-x86-64/pr20253-1l.d: Likewise.
* testsuite/ld-x86-64/pr20830a-now.d: Likewise.
* testsuite/ld-x86-64/pr20830a.d: Likewise.
* testsuite/ld-x86-64/pr20830b-now.d: Likewise.
* testsuite/ld-x86-64/pr20830b.d: Likewise.
* testsuite/ld-x86-64/pr21038a-now.d: Likewise.
* testsuite/ld-x86-64/pr21038a.d: Likewise.
* testsuite/ld-x86-64/pr21038b-now.d: Likewise.
* testsuite/ld-x86-64/pr21038b.d: Likewise.
* testsuite/ld-x86-64/pr21038c-now.d: Likewise.
* testsuite/ld-x86-64/pr21038c.d: Likewise.

x86-64: Update tests for -z separate-code

"-z separate-code" generates different addresses.  Update these tests
to accept any addresses.

* testsuite/ld-x86-64/bnd-ifunc-1.d: Updated.
* testsuite/ld-x86-64/ilp32-4.d: Likewise.

mpx.exp: Add -z max-page-size=0x200000 -z noseparate-code

Add -z max-page-size=0x200000 -z noseparate-code since these tests
check for exact addresses.

* testsuite/ld-x86-64/mpx.exp: Add -z max-page-size=0x200000
-z noseparate-code.

x86-64.exp: Add -z noseparate-code -z max-page-size=0x200000

Add -z noseparate-code -z max-page-size=0x200000 since these tests
check for exact addresses and don't expect extra PT_LOAD segment.  But
don't add them to nacl targets since they generate different addresses.

* testsuite/ld-x86-64/x86-64.exp: Add -z noseparate-code
-z max-page-size=0x200000, excluding NaCl target.

x86: Add -z noseparate-code to IFUNC tests

Add -z noseparate-code since these tests check for exact addresses.

* testsuite/ld-ifunc/ifunc-2-i386-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-local-i386-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-local-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-21-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-22-i386.d: Likewise.
* testsuite/ld-ifunc/pr17154-i386-now.d: Likewise.
* testsuite/ld-ifunc/pr17154-i386.d: Likewise.

x86-64: Add -z max-page-size=0x200000 -z noseparate-code to IFUNC tests

Add -z max-page-size=0x200000 -z noseparate-code since these tests
check for exact addresses.

* testsuite/ld-ifunc/ifunc-2-local-x86-64-now.d: Add
 -z max-page-size=0x200000 -z noseparate-code.
* testsuite/ld-ifunc/ifunc-2-x86-64-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-21-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-22-x86-64.d: Likewise.
* testsuite/ld-ifunc/pr17154-x86-64-now.d: Likewise.
* testsuite/ld-ifunc/pr17154-x86-64.d: Likewise.

i386: Update IFUNC tests for PLT address

i386 generates

 161: e8 ea ff ff ff        call   150 <*ABS*@plt>

instead of

 1e1: e8 ea ff ff ff        callq  1d0 <*ABS*+0x1e0@plt>

* testsuite/ld-ifunc/ifunc-1-local-x86.d: Update PLT address
for i386.
* testsuite/ld-ifunc/ifunc-1-x86.d: Likewise.
* testsuite/ld-ifunc/ifunc-3a-x86.d: Likewise.

x86: Update IFUNC tests for -z separate-code

"-z separate-code" generates different PLT addresses.  Update these tests
to accept any PLT addresses.

* testsuite/ld-ifunc/ifunc-1-local-x86.d: Updated.
* testsuite/ld-ifunc/ifunc-1-x86.d: Likewise.
* testsuite/ld-ifunc/ifunc-3a-x86.d: Likewise.

i386: Add -z noseparate-code to linker tests

Add -z noseparate-code since these tests check for exact addresses.

* testsuite/ld-i386/ibt-plt-1.d: Add -z noseparate-code.
* testsuite/ld-i386/ibt-plt-2a.d: Likewise.
* testsuite/ld-i386/ibt-plt-2b.d: Likewise.
* testsuite/ld-i386/ibt-plt-2c.d: Likewise.
* testsuite/ld-i386/ibt-plt-2d.d: Likewise.
* testsuite/ld-i386/ibt-plt-3a.d: Likewise.
* testsuite/ld-i386/ibt-plt-3b.d: Likewise.
* testsuite/ld-i386/ibt-plt-3c.d: Likewise.
* testsuite/ld-i386/ibt-plt-3d.d: Likewise.
* testsuite/ld-i386/load1.d: Likewise.
* testsuite/ld-i386/pie1.d: Likewise.
* testsuite/ld-i386/pr20244-1a.d: Likewise.
* testsuite/ld-i386/pr20244-1b.d: Likewise.
* testsuite/ld-i386/pr20244-2a.d: Likewise.
* testsuite/ld-i386/pr20244-2b.d: Likewise.
* testsuite/ld-i386/pr20244-2c.d: Likewise.
* testsuite/ld-i386/pr20244-4a.d: Likewise.
* testsuite/ld-i386/pr20244-4b.d: Likewise.
* testsuite/ld-i386/pr20830.d: Likewise.

i386: Update tests for -z separate-code

"-z separate-code" generates:

0x00002080 00200000 00000000 00000000 00000000 . ..............

There is a space ' ', instead of '.'. Update these tests to expect ".*".

* testsuite/ld-i386/pr19636-1a.d: Updated.
* testsuite/ld-i386/pr19636-1b.d: Likewise.
* testsuite/ld-i386/pr19636-1j.d: Likewise.
* testsuite/ld-i386/pr19636-1k.d: Likewise.
* testsuite/ld-i386/pr19636-2a.d: Likewise.
* testsuite/ld-i386/pr19636-2b.d: Likewise.

i386.exp: Add -z noseparate-code

These tests fail due to one extra PT_LOAD segment with -z separate-code.

* testsuite/ld-i386/i386.exp: Add -z noseparate-code.

New class allocate_on_obstack

This patch adds a new class allocate_on_obstack, and let dwarf2_per_objfile
inherit it, so that dwarf2_per_objfile is automatically allocated on
obstack, and "delete dwarf2_per_objfile" doesn't de-allocate any space.

gdb:

2018-02-16  Yao Qi  <yao.qi@linaro.org>

* block.c (block_namespace_info): Inherit allocate_on_obstack.
(block_initialize_namespace): Use new.
* dwarf2read.c (dwarf2_per_objfile): Inherit allocate_on_obstack.
(dwarf2_free_objfile): Use delete.
* gdbtypes.c (type_pair): Inherit allocate_on_obstack.
(copy_type_recursive): Use new.
* gdb_obstack.h (allocate_on_obstack): New.

Add -z noseparate-code to ld-elf tests

These tests fail due to one extra PT_LOAD segment with -z separate-code.

* testsuite/ld-elf/pr19162.d: Add -z noseparate-code.
* testsuite/ld-elf/textaddr1.d: Likewise.
* testsuite/ld-elf/textaddr2.d: Likewise.
* testsuite/ld-elf/textaddr4.d: Likewise.
* testsuite/ld-elf/textaddr6.d: Likewise.

Remove bfd stub function casts.

This patch defines a bunch of new functions to use in the BFD target
structs rather than casting bfd_false or bfd_true and similar stub
functions.  I've also renamed the stub functions to reflect their
parameters and put "error" in the name if they set bfd_error.  The
latter change is important since there were quite a few uses of
bfd_false where setting bfd_error was inappropriate, for example in
elf_backend_allow_non_load_phdr and is_target_special_symbol.

* libbfd.c (_bfd_bool_bfd_false_error): Rename from bfd_false.
(_bfd_bool_bfd_true): Rename from bfd_true.
(_bfd_ptr_bfd_null_error): Rename from bfd_nullvoidptr.
(_bfd_int_bfd_0): Rename from bfd_0.
(_bfd_uint_bfd_0): Rename from bfd_0u.
(_bfd_long_bfd_0): Rename from bfd_0l.
(_bfd_long_bfd_n1_error): Rename from _bfd_n1.
(_bfd_void_bfd): Rename from bfd_void.
(_bfd_bool_bfd_false, _bfd_bool_bfd_asymbol_false),
(_bfd_bool_bfd_link_false_error),
(_bfd_bool_bfd_link_true, _bfd_bool_bfd_bfd_true),
(_bfd_bool_bfd_uint_true, _bfd_bool_bfd_ptr_true),
(_bfd_bool_bfd_asection_bfd_asection_true),
(_bfd_bool_bfd_asymbol_bfd_asymbol_true),
(_bfd_void_bfd_link, _bfd_void_bfd_asection): New functions.
* archive.c (_bfd_noarchive_get_elt_at_index),
(_bfd_noarchive_openr_next_archived_file),
(_bfd_noarchive_construct_extended_name_table),
(_bfd_noarchive_write_ar_hdr, _bfd_noarchive_truncate_arname),
(_bfd_noarchive_write_armap): New functions.
* archures.c (_bfd_nowrite_set_arch_mach): New function.
* coff-alpha.c (alpha_ecoff_swap_coff_aux_in),
(alpha_ecoff_swap_coff_sym_in, alpha_ecoff_swap_coff_lineno_in),
(alpha_ecoff_swap_coff_aux_out, alpha_ecoff_swap_coff_sym_out),
(alpha_ecoff_swap_coff_lineno_out),
(alpha_ecoff_swap_coff_reloc_out): New functions.
* coff-mips.c (mips_ecoff_swap_coff_aux_in),
(mips_ecoff_swap_coff_sym_in, mips_ecoff_swap_coff_lineno_in),
(mips_ecoff_swap_coff_aux_out, mips_ecoff_swap_coff_sym_out),
(mips_ecoff_swap_coff_lineno_out),
(mips_ecoff_swap_coff_reloc_out): New functions.
* coffcode.h (coff_set_alignment_hook): Replace define with
new function.
(symname_in_debug_hook): Likewise.
* ecoff.c (_bfd_ecoff_set_alignment_hook): New function.
* elfxx-target.h (elf_backend_allow_non_load_phdr): Default to 0.
* elf.c (assign_file_positions_except_relocs): Test
elf_backend_allow_non_load_phdr for NULL.
* elflink.c (_bfd_elf_omit_section_dynsym_default): Rename from
_bfd_elf_link_omit_section_dynsym.  Update uses.
(_bfd_elf_omit_section_dynsym_all): New function.
* elf-bfd.h (_bfd_elf_link_omit_section_dynsym): Delete.
(_bfd_elf_omit_section_dynsym_default): Declare.
(_bfd_elf_omit_section_dynsym_all): Declare.
* linker.c (_bfd_nolink_sizeof_headers, _bfd_nolink_bfd_relax_section),
(_bfd_nolink_bfd_get_relocated_section_contents),
(_bfd_nolink_bfd_lookup_section_flags),
(_bfd_nolink_bfd_is_group_section, _bfd_nolink_bfd_discard_group),
(_bfd_nolink_bfd_link_hash_table_create),
(_bfd_nolink_bfd_link_just_syms),
(_bfd_nolink_bfd_copy_link_hash_symbol_type),
(_bfd_nolink_bfd_link_split_section),
(_bfd_nolink_section_already_linked),
(_bfd_nolink_bfd_define_common_symbol),
(_bfd_nolink_bfd_define_start_stop): New functions.
* reloc.c (_bfd_norelocs_bfd_reloc_type_lookup),
(_bfd_norelocs_bfd_reloc_name_lookup),
(_bfd_nodynamic_canonicalize_dynamic_reloc): New functions.
* section.c (_bfd_nowrite_set_section_contents): New function.
* syms.c (_bfd_nosymbols_canonicalize_symtab),
(_bfd_nosymbols_print_symbol, _bfd_nosymbols_get_symbol_info),
(_bfd_nosymbols_get_symbol_version_string),
(_bfd_nosymbols_bfd_is_local_label_name),
(_bfd_nosymbols_get_lineno, _bfd_nosymbols_find_nearest_line),
(_bfd_nosymbols_find_line, _bfd_nosymbols_find_inliner_info),
(_bfd_nosymbols_bfd_make_debug_symbol),
( _bfd_nosymbols_read_minisymbols),
( _bfd_nosymbols_minisymbol_to_symbol),
(_bfd_nodynamic_get_synthetic_symtab): New functions.
* libbfd-in.h: Declare new functions.  Update existing defines,
removing casts.
* aix386-core.c: Update to use new hooks.  Formatting.
* aout-adobe.c: Likewise.
* aout-arm.c: Likewise.
* aout-target.h: Likewise.
* aout-tic30.c: Likewise.
* aoutf1.h: Likewise.
* binary.c: Likewise.
* bout.c: Likewise.
* cisco-core.c: Likewise.
* coff-alpha.c: Likewise.
* coff-i386.c: Likewise.
* coff-i860.c: Likewise.
* coff-i960.c: Likewise.
* coff-ia64.c: Likewise.
* coff-mips.c: Likewise.
* coff-ppc.c: Likewise.
* coff-rs6000.c: Likewise.
* coff-sh.c: Likewise.
* coff-tic30.c: Likewise.
* coff-tic54x.c: Likewise.
* coff-x86_64.c: Likewise.
* coff64-rs6000.c: Likewise.
* coffcode.h: Likewise.
* elf-m10300.c: Likewise.
* elf32-cr16.c: Likewise.
* elf32-lm32.c: Likewise.
* elf32-m32r.c: Likewise.
* elf32-metag.c: Likewise.
* elf32-score.c: Likewise.
* elf32-score7.c: Likewise.
* elf32-tilepro.c: Likewise.
* elf32-xstormy16.c: Likewise.
* elf32-xtensa.c: Likewise.
* elf64-alpha.c: Likewise.
* elf64-hppa.c: Likewise.
* elf64-ia64-vms.c: Likewise.
* elf64-mmix.c: Likewise.
* elf64-sh64.c: Likewise.
* elfnn-ia64.c: Likewise.
* elfxx-sparc.c: Likewise.
* elfxx-target.h: Likewise.
* elfxx-tilegx.c: Likewise.
* elfxx-x86.h: Likewise.
* hp300hpux.c: Likewise.
* hppabsd-core.c: Likewise.
* hpux-core.c: Likewise.
* i386msdos.c: Likewise.
* i386os9k.c: Likewise.
* ieee.c: Likewise.
* ihex.c: Likewise.
* irix-core.c: Likewise.
* libaout.h: Likewise.
* libecoff.h: Likewise.
* mach-o-target.c: Likewise.
* mach-o.c: Likewise.
* mipsbsd.c: Likewise.
* mmo.c: Likewise.
* netbsd-core.c: Likewise.
* nlm-target.h: Likewise.
* oasys.c: Likewise.
* osf-core.c: Likewise.
* pdp11.c: Likewise.
* pe-mips.c: Likewise.
* pe-x86_64.c: Likewise.
* pef.c: Likewise.
* plugin.c: Likewise.
* ppcboot.c: Likewise.
* ptrace-core.c: Likewise.
* sco5-core.c: Likewise.
* som.c: Likewise.
* sparclynx.c: Likewise.
* srec.c: Likewise.
* tekhex.c: Likewise.
* trad-core.c: Likewise.
* verilog.c: Likewise.
* versados.c: Likewise.
* vms-alpha.c: Likewise.
* vms-lib.c: Likewise.
* wasm-module.c: Likewise.
* xsym.c: Likewise.
* libbfd.h: Regenerate.

Fix symbol resolution with linker plugins for defsym symbols.

2018-02-07  Sriraman Tallam  <tmsriram@google.com>

* expression.cc (Symbol_expression::set_expr_sym_in_real_elf):
New method.
(Unary_expression::set_expr_sym_in_real_elf): New method.
(Binary_expression::set_expr_sym_in_real_elf): New method.
(Trinary_expression::set_expr_sym_in_real_elf): New method.
* plugin.cc (get_symbol_resolution_info): Fix symbol resolution if
defined or used in defsyms.
* plugin.h (Plugin_manager::is_defsym_def): New method.
(Plugin_manager::Plugin_manager): Initialize defsym_defines_set_.
(Plugin_manager::defsym_defines_set_): New member.
(Plugin_manager::Defsym_defines_set): New typedef.
* script.cc (Script_options::set_defsym_uses_in_real_elf): New method.
(Script_options::find_defsym_defs): New method.
* script.h (Expression::set_expr_sym_in_real_elf): New method.
(Symbol_assignment::is_defsym): New method.
(Symbol_assignment::value): New method.
(Script_options::find_defsym_defs): New method.
(Script_options::set_defsym_uses_in_real_elf): New method.
* testsuite/Makefile.am (plugin_test_defsym): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/plugin_test.c: Check for new symbol resolution.
* testsuite/plugin_test_defsym.sh: New script.
* testsuite/plugin_test_defsym.c: New test source.

Automatic date update in version.in

RISC-V: Fix relocation failure with zero address sections.

bfd/
* elfnn-riscv.c (_bfd_riscv_relax_section): Ifdef out check to ignore
symbols whose section address is zero.

RISC-V: Give error for ignored pcrel_lo addend.

bfd/
* elfnn-riscv.c (riscv_elf_relocate_section): Use bfd_reloc_dangerous
when pcrel_lo reloc has an addend.  Use reloc_dangerous callback for
bfd_reloc_dangerous.  Use einfo instead of warning callback for errors.
Add %X%P to error messages.

ld/
* testsuite/ld-riscv-elf/ld-riscv-elf.exp: Run pcrel-lo-addend test.
* testsuite/ld-riscv-elf/pcrel-lo-addend.d: New.
* testsuite/ld-riscv-elf/pcrel-lo-addend.s: New.

Fix AArch32 build attributes for Armv8.4-A.

The build attribute number for Armv8.4-A is currently incorrectly set to that of Armv8-M.
This patch fixes that by setting it as part of the Armv8-A family and adds a test for it.

gas/
2018-02-15  Tamar Christina  <tamar.christina@arm.com>

* config/tc-arm.c (cpu_arch_ver): Renumber ARM_ARCH_V8_4A.
* testsuite/gas/arm/attr-march-armv8_4-a.d: New.

PR ld/22832 on SPARC.

The fix for PR ld/22727 on SPARC passed TRUE as the 'create' argument
in the call to bfd_link_hash_lookup.  It turns out this was a bad idea
because, if the symbol is created at this point, the link will abort
later in elf_link_output_extsym.  This changes the TRUE into a FALSE
and puts an assertion on the result of the call, making it easier to
debug the issue; that's exactly in keeping with what Gold does.

bfd/
* elfxx-sparc.c (_bfd_sparc_elf_check_relocs) <R_SPARC_TLS_GD_CALL>:
Pass FALSE instead of TRUE as 'create' argument to bfd_link_hash_lookup
and assert that the result of the call is not NULL.

Reset inferior::control on inferior exit

When we kill an inferior, the inferior is not deleted.  What is more, it
is reused when the new process is created, so we need to reset inferior's
state when it exits.

gdb:

2018-02-15  Yao Qi  <yao.qi@linaro.org>

PR gdb/22849
* inferior.c (exit_inferior_1): Reset inf->control.

delete ada-lang.c::ada_to_fixed_value_create advance declaration

This advance declaration really isn't necesary, since the implementation
of this function comes before the first reference to it.

gdb/ChangeLog:

        * ada-lang.c (ada_to_fixed_value_create): Delete advance
        declaration.

Tested by rebuilding GDB.

Automatic date update in version.in

Fix GDB crash after Quit thrown from unwinder sniffer

I ran into a GDB crash in gdb.base/bp-cmds-continue-ctrl-c.exp in my
multi-target branch, which turns out exposed a bug that exists in
master too.

That testcase has a breakpoint with a "continue" command associated.
Then the breakpoint is constantly being hit.  At the same time, the
testcase is continualy interrupting the program with Ctrl-C, and
re-resuming it, in a loop.

Running that testcase manually under Valgrind, after a few sequences
of 'Ctrl-C' + 'continue', I got:

 Breakpoint 1, Quit
 (gdb) ==21270== Invalid read of size 8
 ==21270==    at 0x4D8185: pyuw_this_id(frame_info*, void**, frame_id*) (py-unwind.c:461)
 ==21270==    by 0x6D426A: compute_frame_id(frame_info*) (frame.c:505)
 ==21270==    by 0x6D43B7: get_frame_id(frame_info*) (frame.c:537)
 ==21270==    by 0x84F3B8: scoped_restore_current_thread::scoped_restore_current_thread() (thread.c:1678)
 ==21270==    by 0x718E3D: fetch_inferior_event(void*) (infrun.c:4076)
 ==21270==    by 0x7067C9: inferior_event_handler(inferior_event_type, void*) (inf-loop.c:43)
 ==21270==    by 0x45BEF9: handle_target_event(int, void*) (linux-nat.c:4419)
 ==21270==    by 0x6C4255: handle_file_event(file_handler*, int) (event-loop.c:733)
 ==21270==    by 0x6C47F8: gdb_wait_for_event(int) (event-loop.c:859)
 ==21270==    by 0x6C3666: gdb_do_one_event() (event-loop.c:322)
 ==21270==    by 0x6C3712: start_event_loop() (event-loop.c:371)
 ==21270==    by 0x746801: captured_command_loop() (main.c:329)
 ==21270==  Address 0x0 is not stack'd, malloc'd or (recently) free'd
 ==21270==
 ==21270==
 ==21270== Process terminating with default action of signal 11 (SIGSEGV): dumping core
 ==21270==  Access not within mapped region at address 0x0
 ==21270==    at 0x4D8185: pyuw_this_id(frame_info*, void**, frame_id*) (py-unwind.c:461)
 ==21270==    by 0x6D426A: compute_frame_id(frame_info*) (frame.c:505)
 ==21270==    by 0x6D43B7: get_frame_id(frame_info*) (frame.c:537)
 ==21270==    by 0x84F3B8: scoped_restore_current_thread::scoped_restore_current_thread() (thread.c:1678)
 ==21270==    by 0x718E3D: fetch_inferior_event(void*) (infrun.c:4076)
 ==21270==    by 0x7067C9: inferior_event_handler(inferior_event_type, void*) (inf-loop.c:43)
 ==21270==    by 0x45BEF9: handle_target_event(int, void*) (linux-nat.c:4419)
 ==21270==    by 0x6C4255: handle_file_event(file_handler*, int) (event-loop.c:733)
 ==21270==    by 0x6C47F8: gdb_wait_for_event(int) (event-loop.c:859)
 ==21270==    by 0x6C3666: gdb_do_one_event() (event-loop.c:322)
 ==21270==    by 0x6C3712: start_event_loop() (event-loop.c:371)
 ==21270==    by 0x746801: captured_command_loop() (main.c:329)
 ==21270==  If you believe this happened as a result of a stack
 ==21270==  overflow in your program's main thread (unlikely but
 ==21270==  possible), you can try to increase the size of the
 ==21270==  main thread stack using the --main-stacksize= flag.
 ==21270==  The main thread stack size used in this run was 8388608.
 ==21270==

Above, when we get to compute_frame_id, fi->unwind is non-NULL,
meaning, we found an unwinder, in this case the Python unwinder, but
somehow, fi->prologue_cache is left NULL.  pyuw_this_id then crashes
because it assumes fi->prologue_cache is non-NULL:

  static void
  pyuw_this_id (struct frame_info *this_frame, void **cache_ptr,
struct frame_id *this_id)
  {
    *this_id = ((cached_frame_info *) *cache_ptr)->frame_id;
                                      ^^^^^^^^^^

'*cache_ptr' here is 'fi->prologue_cache'.

There's a quit() call in pyuw_sniffer that I believe is the one that
sometimes triggers the crash above.  The crash can be reproduced
easily with this hack to force a quit out of the python unwinder:

 --- a/gdb/python/py-unwind.c
 +++ b/gdb/python/py-unwind.c
 @@ -497,6 +497,8 @@ pyuw_sniffer (const struct frame_unwind *self, struct frame_info *this_frame,
    struct gdbarch *gdbarch = (struct gdbarch *) (self->unwind_data);
    cached_frame_info *cached_frame;

 +  quit ();
 +
    gdbpy_enter enter_py (gdbarch, current_language);

    TRACE_PY_UNWIND (3, "%s (SP=%s, PC=%s)\n", __FUNCTION__,

After that quit is thrown, any subsequent operation that involves
unwinding results in GDB crashing with SIGSEGV like above.

The problem is that this commit:

  commit 30a9c02feff56bd58a276c2a7262f364baa558ac
  CommitDate: Sun Oct 8 23:16:42 2017 -0600
  Subject: Remove cleanup from frame_prepare_for_sniffer

missed that we need to call frame_cleanup_after_sniffer before
rethrowing the exception too.

Without the fix, the "bt" added to
gdb.base/bp-cmds-continue-ctrl-c.exp in this commit makes GDB crash:

  Running src/gdb/testsuite/gdb.base/bp-cmds-continue-ctrl-c.exp ...
  ERROR: Process no longer exists

gdb/ChangeLog:
2018-02-14  Pedro Alves  <palves@redhat.com>

* frame-unwind.c (frame_unwind_try_unwinder): Always call
frame_cleanup_after_sniffer on exception.

gdb/testsuite/ChangeLog:
2018-02-14  Pedro Alves  <palves@redhat.com>

* gdb.base/bp-cmds-continue-ctrl-c.exp (do_test): Test "bt" after
getting a "Quit".

Constify target_so_ops::bfd_open

This constifies the bfd_open method of struct target_so_ops.

gdb/ChangeLog
2018-02-14  Tom Tromey  <tom@tromey.com>

* solist.h (struct target_so_ops) <bfd_open>: Make pathname
const.
(solib_bfd_open): Make pathname const.
* solib.c (solib_bfd_open): Make pathname const.
* solib-spu.c (spu_bfd_fopen): Make name const.
(spu_bfd_open): Make pathname const.
* solib-darwin.c (darwin_bfd_open): Make pathname const.
* solib-aix.c (solib_aix_bfd_open): Make pathname const.

Change openp et al to use a unique_xmalloc_ptr

This changes openp, source_full_path_of, and find_and_open_source to
take a unique_xmalloc_ptr, rather than a char*, as an outgoing
argument type.  This simplifies the API, ownership-wise, and allows
for the removal of some cleanups.

gdb/ChangeLog
2018-02-14  Tom Tromey  <tom@tromey.com>

* symfile.c (symfile_bfd_open): Update.
* source.h (openp, source_full_path_of, find_and_open_source):
Change argument type to unique_xmalloc_ptr.
* source.c (openp): Take a unique_xmalloc_ptr.
(source_full_path_of, find_and_open_source): Likewise.
(open_source_file, symtab_to_fullname): Update.
* solist.h (struct target_so_ops) <find_and_open_solib>: Take a
unique_xmalloc_ptr.
* solib.c (solib_find_1): Use unique_xmalloc_ptr.
(exec_file_find): Update.
* psymtab.c (psymtab_to_fullname): Update.
* nto-tdep.h (nto_find_and_open_solib): Update.
* nto-tdep.c (nto_find_and_open_solib): Change temp_path to a
unique_xmalloc_ptr.
* exec.c (exec_file_attach): Update.
* dwarf2read.c (try_open_dwop_file): Use unique_xmalloc_ptr.
* cli/cli-cmds.c (find_and_open_script): Use unique_xmalloc_ptr.

Move some declarations to source.h

I noticed a few declarations in defs.h that really could be put into
source.h.  I think it's generally preferable to something out of
defs.h unless it is needed by most of the files in gdb.

gdb/ChangeLog
2018-02-14  Tom Tromey  <tom@tromey.com>

* solib.c: Include source.h.
* nto-tdep.c: Include source.h.
* mi/mi-cmd-env.c: Include source.h.
* infcmd.c: Include source.h.
* exec.c: Include source.h.
* defs.h (enum openp_flag, openp, source_full_path_of, mod_path)
(add_path, directory_switch, source_path, init_source_path): Move
declarations...
* source.h (enum openp_flag, openp, source_full_path_of, mod_path)
(add_path, directory_switch, source_path, init_source_path):
...here.

Return unique_xmalloc_ptr from some solib.c functions

This changes a couple of solib.c functions -- exec_file_find and
solib_find -- to return a unique_xmalloc_ptr, and then fixes up the
users.  This allows the removal of some cleanups.

This also changes solib_bfd_open to not take ownership of its
argument.  I think this change is somewhat cleaner.

gdb/ChangeLog
2018-02-14  Tom Tromey  <tom@tromey.com>

* solist.h (exec_file_find, solib_find): Return
unique_xmalloc_ptr.
(solib_bfd_fopen): Take a const char *.
* solib.c (solib_find_1): Return unique_xmalloc_ptr.
(exec_file_find, solib_find): Likewise.
(solib_bfd_fopen): Do not take ownership of "pathname".
(solib_bfd_open): Use unique_xmalloc_ptr.
* solib-darwin.c (darwin_bfd_open): Use unique_xmalloc_ptr.
* solib-aix.c (solib_aix_bfd_open): Use unique_xmalloc_ptr.
* infrun.c (follow_exec): Use unique_xmalloc_ptr.
* exec.c (exec_file_locate_attach): Use unique_xmalloc_ptr.