platform/upstream/binutils.git
5 years agoFix test-cp-name-parser build, parser_fprintf undefined
Pedro Alves [Tue, 12 Mar 2019 16:56:02 +0000 (16:56 +0000)]
Fix test-cp-name-parser build, parser_fprintf undefined

  $ make test-cp-name-parser
  ...
  test-cp-name-parser.o: In function `yy_symbol_print(_IO_FILE*, int, YYSTYPE const*, cpname_state*)':
  build/gdb/cp-name-parser.c.tmp:1335: undefined reference to `parser_fprintf(_IO_FILE*, char const*, ...)'
  build/gdb/cp-name-parser.c.tmp:1339: undefined reference to `parser_fprintf(_IO_FILE*, char const*, ...)'
  test-cp-name-parser.o: In function `yy_stack_print(short*, short*)':
  build/gdb/cp-name-parser.c.tmp:1350: undefined reference to `parser_fprintf(_IO_FILE*, char const*, ...)'
  build/gdb/cp-name-parser.c.tmp:1354: undefined reference to `parser_fprintf(_IO_FILE*, char const*, ...)'
  build/gdb/cp-name-parser.c.tmp:1356: undefined reference to `parser_fprintf(_IO_FILE*, char const*, ...)'
  build/gdb/cp-name-parser.c.tmp:1376: more undefined references to `parser_fprintf(_IO_FILE*, char const*, ...)' follow
  collect2: error: ld returned 1 exit status
  make: *** [Makefile:1833: test-cp-name-parser] Error 1

gdb/ChangeLog:
2019-03-12  Pedro Alves  <palves@redhat.com>

* yy-remap.h [TEST_CPNAMES] (YYFPRINTF): Don't define.

5 years agoFix test-cp-name-parser build, unused variable
Pedro Alves [Tue, 12 Mar 2019 16:56:02 +0000 (16:56 +0000)]
Fix test-cp-name-parser build, unused variable

  $ make test-cp-name-parser
  ...
  CXX    test-cp-name-parser.o
  src/gdb/cp-name-parser.y: In function ‘int gdb::main(int, char**)’:
  src/gdb/cp-name-parser.y:2137:6: error: unused variable ‘len’ [-Werror=unused-variable]
    int len;
^~~
  cc1plus: all warnings being treated as errors

gdb/ChangeLog:
2019-03-12  Pedro Alves  <palves@redhat.com>

* cp-name-parser.y (main): Remove unused 'len' variable.

5 years agoMake null_ptid and minus_one_ptid "const"
Tom Tromey [Mon, 4 Mar 2019 19:05:47 +0000 (12:05 -0700)]
Make null_ptid and minus_one_ptid "const"

This makes null_ptid and minus_one_ptid "const".  I think this is an
improvement because it means they can't be accidentally modified.

2019-03-12  Tom Tromey  <tromey@adacore.com>

* common/ptid.c (null_ptid, minus_one_ptid): Now const.
* common/ptid.h (null_ptid, minus_one_ptid): Now const.

5 years agoChange iterate_over_lwps to take a gdb::function_view
Tom Tromey [Mon, 4 Mar 2019 19:05:03 +0000 (12:05 -0700)]
Change iterate_over_lwps to take a gdb::function_view

This changes iterate_over_lwps to use a gdb::function_view.  This was
needed in order to make null_ptid and minus_one_ptid 'const'.

gdb/ChangeLog
2019-03-12  Tom Tromey  <tromey@adacore.com>

* linux-nat.c (iterate_over_lwps): Update.
(stop_callback): Remove parameter.
(stop_wait_callback, detach_callback, resume_set_callback)
(select_singlestep_lwp_callback, set_ignore_sigint)
(status_callback, resumed_callback, resume_clear_callback)
(kill_callback, kill_wait_callback, linux_nat_stop_lwp): Remove
data parameter.
(linux_nat_target::detach, linux_nat_target::resume)
(linux_stop_and_wait_all_lwps, select_event_lwp)
(linux_nat_filter_event, linux_nat_wait_1)
(linux_nat_target::kill, linux_nat_target::stop)
(linux_nat_target::stop): Update.
(linux_nat_resume_callback): Change type.
(resume_stopped_resumed_lwps, count_events_callback)
(select_event_lwp_callback): Likewise.
(linux_stop_lwp, linux_nat_stop_lwp): Update.
* arm-linux-nat.c (struct update_registers_data): Remove.
(update_registers_callback): Change type.
(arm_linux_insert_hw_breakpoint1): Update.
* nat/x86-linux-dregs.c (update_debug_registers_callback): Remove
parameter.
(x86_linux_dr_set_addr): Update.
(x86_linux_dr_set_control): Update.
* nat/linux-nat.h (iterate_over_lwps_ftype): Remove parameter.
(iterate_over_lwps): Use gdb::function_view.
* nat/aarch64-linux-hw-point.c (struct
aarch64_dr_update_callback_param): Remove.
(debug_reg_change_callback): Change type.
(aarch64_notify_debug_reg_change): Update.
* s390-linux-nat.c (s390_refresh_per_info): Update.

gdb/gdbserver/ChangeLog
2019-03-12  Tom Tromey  <tromey@adacore.com>

* linux-low.c (iterate_over_lwps): Update.

5 years agoDon't use bfd_get_file_size in objdump
Alan Modra [Tue, 12 Mar 2019 05:49:25 +0000 (16:19 +1030)]
Don't use bfd_get_file_size in objdump

Compressed debug sections can have uncompressed sizes that exceed the
original file size, so we can't use bfd_get_file_size.  objdump also
used bfd_get_file_size to limit reloc section size, but I believe the
underlying bug causing the PR22508 out of bounds buffer access was
that we had an integer overflow when calculating the reloc buffer
size.  I've fixed that instead in most of the backends, som and
vms-alpha being the exceptions.  SOM and vmd-alpha have rather more
serious bugs in their slurp_relocs routines that would need fixing
first if we want to fuss about making them safe against fuzzed object
files.

The patch also fixes a number of other potential overflows by using
the bfd_alloc2/malloc2/zalloc2 memory allocation functions.

bfd/
* coffcode.h (buy_and_read): Delete unnecessary forward decl.  Add
nmemb parameter.  Use bfd_alloc2.
(coff_slurp_line_table): Use bfd_alloc2.  Update buy_and_read calls.
Delete assertion.
(coff_slurp_symbol_table): Use bfd_alloc2 and bfd_zalloc2.
(coff_slurp_reloc_table): Use bfd_alloc2.  Update buy_and_read calls.
* coffgen.c (coff_get_reloc_upper_bound): Ensure size calculation
doesn't overflow.
* elf.c (bfd_section_from_shdr): Use bfd_zalloc2.  Style fix.
(assign_section_numbers): Style fix.
(swap_out_syms): Use bfd_malloc2.
(_bfd_elf_get_reloc_upper_bound): Ensure size calculation doesn't
overflow.
(_bfd_elf_make_empty_symbol): Style fix.
(elfobj_grok_stapsdt_note_1): Formatting.
* elfcode.h (elf_object_p): Use bfd_alloc2.
(elf_write_relocs, elf_write_shdrs_and_ehdr): Likewise.
(elf_slurp_symbol_table): Use bfd_zalloc2.
(elf_slurp_reloc_table): Use bfd_alloc2.
(_bfd_elf_bfd_from_remote_memory): Use bfd_malloc2.
* elf64-sparc (elf64_sparc_get_reloc_upper_bound): Ensure
size calculation doesn't overflow.
(elf64_sparc_get_dynamic_reloc_upper_bound): Likewise.
* mach-o.c (bfd_mach_o_get_reloc_upper_bound): Likewise.
* pdp11.c (get_reloc_upper_bound): Copy aoutx.h version.
binutils/
* objdump.c (load_specific_debug_section): Don't compare section
size against file size.
(dump_relocs_in_section): Don't compare reloc size against file size.
Print "failed to read relocs" on bfd_get_reloc_upper_bound error.

5 years agoAdd missing changelogs for previous commits.
Andreas Krebbel [Tue, 12 Mar 2019 13:23:10 +0000 (14:23 +0100)]
Add missing changelogs for previous commits.

5 years agoS/390: arch13: Adjust to recent changes
Andreas Krebbel [Tue, 12 Mar 2019 13:09:55 +0000 (14:09 +0100)]
S/390: arch13: Adjust to recent changes

opcodes/ChangeLog:

2019-03-12  Andreas Krebbel  <krebbel@linux.ibm.com>

* s390-opc.txt: Rename selhhhr to selfhr.  Remove optional operand
from vstrszb, vstrszh, and vstrszf.

gas/ChangeLog:

2019-03-12  Andreas Krebbel  <krebbel@linux.ibm.com>

* testsuite/gas/s390/zarch-arch13.s: Adjust testcase to optable changes.
* testsuite/gas/s390/zarch-arch13.d: Likewise.

5 years agoS/390: arch13: Add instruction descriptions
Andreas Krebbel [Tue, 12 Mar 2019 13:09:55 +0000 (14:09 +0100)]
S/390: arch13: Add instruction descriptions

opcodes/ChangeLog:

2019-03-12  Andreas Krebbel  <krebbel@linux.ibm.com>

* s390-opc.txt: Add instruction descriptions.

5 years agoAutomatic date update in version.in
GDB Administrator [Tue, 12 Mar 2019 00:00:20 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoRemove redundant assignment from dwarf2_find_containing_comp_unit
Tom Tromey [Mon, 11 Mar 2019 16:22:17 +0000 (10:22 -0600)]
Remove redundant assignment from dwarf2_find_containing_comp_unit

dwarf2_find_containing_comp_unit has two assignments to "this_cu" in
quick succession, both of which are just:

  this_cu = dwarf2_per_objfile->all_comp_units[low];

... with no intervening assignments.

This patch removes the second assignment.  I'm checking this in as
obvious.  Tested on x86-64 Fedora 29.

gdb/ChangeLog
2019-03-11  Tom Tromey  <tromey@adacore.com>

* dwarf2read.c (dwarf2_find_containing_comp_unit): Remove
redundant assignment to "this_cu".

5 years agoAutomatic date update in version.in
GDB Administrator [Mon, 11 Mar 2019 00:00:22 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoAutomatic date update in version.in
GDB Administrator [Sun, 10 Mar 2019 00:01:08 +0000 (00:01 +0000)]
Automatic date update in version.in

5 years agoRemove unnecessary cases from rank_one_type's switch
Simon Marchi [Fri, 8 Mar 2019 15:15:12 +0000 (10:15 -0500)]
Remove unnecessary cases from rank_one_type's switch

We return INCOMPATIBLE_TYPE_BADNESS for all these type codes, so we might as
well just let them go to the default case.

Incidentally, this patch also makes this false positive error go away when
compiling with gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0, default compiler on
Ubuntu 18.04.

  CXX    gdbtypes.o
/home/smarchi/src/binutils-gdb/gdb/gdbtypes.c: In function ‘rank rank_one_type(type*, type*, value*)’:
/home/smarchi/src/binutils-gdb/gdb/gdbtypes.c:4259:1: error: control reaches end of non-void function [-Werror=return-type]
 }
 ^

gdb/ChangeLog:

* gdbtypes.c (rank_one_type): Remove unnecessary cases from switch.

5 years agoSplit rank_one_type_parm_set from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:11 +0000 (10:15 -0500)]
Split rank_one_type_parm_set from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_set): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_struct from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:11 +0000 (10:15 -0500)]
Split rank_one_type_parm_struct from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_struct): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_complex from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:11 +0000 (10:15 -0500)]
Split rank_one_type_parm_complex from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_complex): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_float from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:10 +0000 (10:15 -0500)]
Split rank_one_type_parm_float from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_float): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_bool from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:10 +0000 (10:15 -0500)]
Split rank_one_type_parm_bool from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_bool): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_range from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:09 +0000 (10:15 -0500)]
Split rank_one_type_parm_range from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_range): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_char from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:09 +0000 (10:15 -0500)]
Split rank_one_type_parm_char from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_char): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_enum from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:09 +0000 (10:15 -0500)]
Split rank_one_type_parm_enum from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_enum): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_int from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:08 +0000 (10:15 -0500)]
Split rank_one_type_parm_int from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_int): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_func from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:08 +0000 (10:15 -0500)]
Split rank_one_type_parm_func from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_func): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_array from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:08 +0000 (10:15 -0500)]
Split rank_one_type_parm_array from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_array): New function extracted
from...
(rank_one_type): ... this.

5 years agoSplit rank_one_type_parm_ptr from rank_one_type
Simon Marchi [Fri, 8 Mar 2019 15:15:07 +0000 (10:15 -0500)]
Split rank_one_type_parm_ptr from rank_one_type

gdb/ChangeLog:

* gdbtypes.c (rank_one_type_parm_ptr): New function extracted
from...
(rank_one_type): ... this.

5 years agoEnsure 'help set/show print inferior-events' shows the example events.
Philippe Waroquiers [Sat, 9 Mar 2019 11:25:11 +0000 (12:25 +0100)]
Ensure 'help set/show print inferior-events' shows the example events.

Without this patch, the help stops after  'e.g.' :
  (gdb) apropos \(inferior\|thread\) event
  set print inferior-events -- Set printing of inferior events (e.g.
  set print thread-events -- Set printing of thread events (such as thread start and exit)
  show print inferior-events -- Show printing of inferior events (e.g.
  show print thread-events -- Show printing of thread events (such as thread start and exit)

Using the same notation as for the thread evenets (i.e. 'such as') gives:
 (gdb) apropos \(inferior\|thread\) event
 set print inferior-events -- Set printing of inferior events (such as inferior start and exit)
 set print thread-events -- Set printing of thread events (such as thread start and exit)
 show print inferior-events -- Show printing of inferior events (such as inferior start and exit)
 show print thread-events -- Show printing of thread events (such as thread start and exit)

5 years agoSupport styling on native MS-Windows console
Eli Zaretskii [Sat, 9 Mar 2019 06:44:56 +0000 (08:44 +0200)]
Support styling on native MS-Windows console

gdb/ChangeLog:
2019-03-08  Eli Zaretskii  <eliz@gnu.org>

PR/24315
* utils.c (can_emit_style_escape) [_WIN32]: Don't disable styling
on MS-Windows if $TERM is not defined.

* cli/cli-style.c: Set cli_styling to 1 in the MinGW build.

* posix-hdep.c (gdb_console_fputs):
* mingw-hdep.c (rgb_to_16colors, gdb_console_fputs): New
functions.
* ui-file.h (gdb_console_fputs): Add prototype.

* ui-file.c (stdio_file::puts): Call gdb_console_fputs, and fall
back to fputs only if the former returns zero.

5 years agoAutomatic date update in version.in
GDB Administrator [Sat, 9 Mar 2019 00:00:30 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoPR24311, FAIL: S-records with constructors
Alan Modra [Fri, 8 Mar 2019 12:11:55 +0000 (22:41 +1030)]
PR24311, FAIL: S-records with constructors

Not padding string merge section output to its alignment can cause
failures of the S-record tests when input string merge sections are
padded, since the ELF linker output for the single string section
would shrink compared to the SREC linker output.  That might result in
following sections having different addresses.
On the other hand, padding string merge section output when input
string merge sections are *not* padded can also cause failures, in
this case due to the ELF linker output for the string section being
larger (due to padding) than the SREC linker output.

It would be better to write a more robust test, but it is also nice
to leave input unchanged when no string merges occur.

PR 24311
* merge.c (merge_strings): Return secinfo.  Don't pad section
to alignment here.
(_bfd_merge_sections): Pad section to alignment here, if input
sections contributing to merged output all pad to alignment.
Formatting.

5 years agoAutomatic date update in version.in
GDB Administrator [Fri, 8 Mar 2019 00:00:42 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoC++-ify bcache
Tom Tromey [Thu, 7 Mar 2019 11:20:19 +0000 (04:20 -0700)]
C++-ify bcache

This somewhat C++-ifies bcache.  It replaces bcache_xmalloc and
bcache_xfree with constructors; changes some functions into methods;
and changes various structures to include a bcache directly (as
opposed to a pointer to a bcache).

Tested by the buildbot.

gdb/ChangeLog
2019-03-07  Tom Tromey  <tom@tromey.com>

* symmisc.c (print_symbol_bcache_statistics): Update.
(print_objfile_statistics): Update.
* symfile.c (allocate_symtab): Update.
* stabsread.c: Don't include bcache.h.
* psymtab.h (struct psymbol_bcache): Don't declare.
(class psymtab_storage) <psymbol_cache>: Now a bcache.
(psymbol_bcache_init, psymbol_bcache_free)
(psymbol_bcache_get_bcache): Don't declare.
* psymtab.c (struct psymbol_bcache): Remove.
(psymtab_storage::psymtab_storage): Update.
(psymtab_storage::~psymtab_storage): Update.
(psymbol_bcache_init, psymbol_bcache_free)
(psymbol_bcache_get_bcache, psymbol_bcache_full): Remove.
(add_psymbol_to_bcache): Update.
(allocate_psymtab): Update.
* objfiles.h (struct objfile_per_bfd_storage) <filename_cache,
macro_cache>: No longer pointers.
* objfiles.c (get_objfile_bfd_data): Don't call bcache_xmalloc.
(free_objfile_per_bfd_storage): Don't call bcache_xfree.
* macrotab.c (macro_bcache): Update.
* macroexp.c: Don't include bcache.h.
* gdbtypes.c (check_types_worklist): Update.
(types_deeply_equal): Remove TRY/CATCH.  Update.
* elfread.c (elf_symtab_read): Update.
* dwarf2read.c: Don't include bcache.h.
* buildsym.c (buildsym_compunit::get_macro_table): Update.
* bcache.h (bcache, bcache_full, bcache_xffree, bcache_xmalloc)
(print_bcache_statistics, bcache_memory_used): Don't declare.
(struct bcache): Move from bcache.c.  Add constructor, destructor,
methods.  Rename all data members.
* bcache.c (struct bcache): Move to bcache.h.
(bcache::expand_hash_table): Rename from expand_hash_table.
(bcache): Remove.
(bcache::insert): Rename from bcache_full.
(bcache::compare): Rename from bcache_compare.
(bcache_xmalloc): Remove.
(bcache::~bcache): Rename from bcache_xfree.
(bcache::print_statistics): Rename from print_bcache_statistics.
(bcache::memory_used): Rename from bcache_memory_used.

5 years agoFix normal_stop latent bug
Pedro Alves [Thu, 7 Mar 2019 16:35:06 +0000 (16:35 +0000)]
Fix normal_stop latent bug

TARGET_WAITKIND_NO_RESUMED doesn't have an associated event thread, so
we shouldn't be referring to inferior_thread() assuming it points to
one.

This was caught on the multi-target branch, where we always switch to
no-thread-selected whenever we start handling an event, exactly to
catch places that incorrectly use "inferior_ptid/inferior_thread()"
without switching to the right event thread / target.

Here, on the branch, we assert in inferior_thread() because
TARGET_WAITKIND_NO_RESUMED doesn't have an associated event thread, so
inferior_ptid is still null_ptid.

gdb/ChangeLog:
2019-03-07  Pedro Alves  <palves@redhat.com>

* infrun.c (normal_stop): Also check for
TARGET_WAITKIND_NO_RESUMED before referring to inferior_thread().

5 years agogdb: Move value_from_host_double into value.c and make more use of it
Andrew Burgess [Thu, 7 Mar 2019 14:53:37 +0000 (14:53 +0000)]
gdb: Move value_from_host_double into value.c and make more use of it

The function value_from_host_double can be moved from f-lang.c into
value.c as a generally useful function, and then used more widely.

Tested on X86-64/GNU Linux with no regressions.

gdb/ChangeLog:

* f-lang.c (value_from_host_double): Moved to...
* value.c (value_from_host_double): ...here.
* value.h (value_from_host_double): Declare.
* guile/scm-math.c (vlscm_convert_typed_number): Use
value_from_host_double.
(vlscm_convert_number): Likewise.
* guile/scm-value.c (gdbscm_value_to_real): Likewise.
* python/py-value.c (convert_value_from_python): Likewise.

5 years agoAutomatic date update in version.in
GDB Administrator [Thu, 7 Mar 2019 00:00:23 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoUse SCOPE_EXIT in write_gcore_file
Tom Tromey [Mon, 28 Jan 2019 17:51:01 +0000 (10:51 -0700)]
Use SCOPE_EXIT in write_gcore_file

This replaces a TRY/CATCH in write_gcore_file with a use of SCOPE_EXIT
instead.  I find that this is simpler to understand.

2019-03-06  Tom Tromey  <tom@tromey.com>

* gcore.c (write_gcore_file): Use SCOPE_EXIT.

5 years agoRemove free_current_contents
Tom Tromey [Thu, 24 Jan 2019 16:58:29 +0000 (09:58 -0700)]
Remove free_current_contents

free_current_contents is no longer used, so this patch removes it.

2019-03-06  Tom Tromey  <tom@tromey.com>

* utils.h (free_current_contents): Don't declare.
* utils.c (free_current_contents): Remove.

5 years agoRemove basic cleanup code
Tom Tromey [Thu, 24 Jan 2019 16:56:49 +0000 (09:56 -0700)]
Remove basic cleanup code

This removes the basic cleanup code: make_cleanups, do_cleanups,
discard_cleanups, and friends.  This code is no longer needed, as
nothing in gdb makes an ordinary cleanup.  Final cleanups are still
needed.

2019-03-06  Tom Tromey  <tom@tromey.com>

* top.c (quit_force): Update.
* main.c (captured_command_loop): Update.
* common/new-op.c (operator new): Update.
* common/common-exceptions.c (struct catcher)
<save_cleanup_chain>: Remove member.
(exceptions_state_mc_init): Update.
(exception_try_scope_entry): Return nullptr.
(exception_try_scope_exit, exception_rethrow)
(throw_exception_sjlj, throw_exception_cxx): Update.
* common/cleanups.h (make_cleanup, make_cleanup_dtor)
(all_cleanups, do_cleanups, discard_cleanups)
(discard_final_cleanups, save_cleanups, save_final_cleanups)
(restore_cleanups, restore_final_cleanups): Don't declare.
(do_final_cleanups): Remove parameter.
* common/cleanups.c (cleanup_chain, make_cleanup)
(make_cleanup_dtor, all_cleanups, do_cleanups)
(discard_my_cleanups, discard_cleanups)
(discard_final_cleanups, save_my_cleanups, save_cleanups)
(save_final_cleanups, restore_my_cleanups, restore_cleanups)
(null_cleanup): Remove.
(do_final_cleanups): Remove parameter.

5 years agoUse unique_xmalloc_ptr in remote.c
Tom Tromey [Thu, 14 Feb 2019 23:10:01 +0000 (16:10 -0700)]
Use unique_xmalloc_ptr in remote.c

This removes a cleanup from remote.c, replacing it with
unique_xmalloc_ptr.

2019-03-06  Tom Tromey  <tom@tromey.com>

* remote.c (remote_target::remote_parse_stop_reply): Use
unique_xmalloc_ptr.

5 years agoRemove last cleanups from stabsread.c
Tom Tromey [Thu, 24 Jan 2019 16:51:19 +0000 (09:51 -0700)]
Remove last cleanups from stabsread.c

This removes the last cleanups from stabsread.c.  Similar code in
dwarf2read.c was C++-ified, but considering that stabs are deprecated,
it seemed simpler to just change these allocations to use an obstack
and leave the data structures in place.

This patch renames field_info to stabs_field_info -- adding a
constructor here provoked a bug due to the resulting ODR violation.

2019-03-06  Tom Tromey  <tom@tromey.com>

* stabsread.c (struct stabs_field_info): Rename from field_info.
<list, fnlist>: Add initializers.
<obstack>: New member.
(read_member_functions, read_struct_fields, read_baseclasses):
Allocate on obstack.  Don't use cleanups.
(read_one_struct_field, read_member_functions, read_struct_fields)
(read_baseclasses, read_tilde_fields, attach_fn_fields_to_type)
(attach_fields_to_type, read_cpp_abbrev, read_member_functions)
(read_struct_type): Update.

5 years agoRemove last cleanup from linux-namespaces.c
Tom Tromey [Thu, 24 Jan 2019 13:43:45 +0000 (06:43 -0700)]
Remove last cleanup from linux-namespaces.c

This removes the last cleanup from linux-namespaces.c, replacing it
with a use of SCOPE_EXIT.

2019-03-06  Tom Tromey  <tom@tromey.com>

* nat/linux-namespaces.c (linux_mntns_access_fs): Use SCOPE_EXIT.
* common/filestuff.h (make_cleanup_close): Don't declare.
* common/filestuff.c (do_close_cleanup, make_cleanup_close):
Remove.

5 years agoRemove last cleanup solib-aix.c
Tom Tromey [Thu, 24 Jan 2019 13:38:59 +0000 (06:38 -0700)]
Remove last cleanup solib-aix.c

This removes the last cleanup solib-aix.c, replacing it with a use of
make_scope_exit.

2019-03-06  Tom Tromey  <tom@tromey.com>

* solib-aix.c: Use make_scope_exit.

5 years agoRemove last cleanups from solib-svr4.c
Tom Tromey [Thu, 24 Jan 2019 13:41:12 +0000 (06:41 -0700)]
Remove last cleanups from solib-svr4.c

This removes the last cleanups from solib-svr4.c, replacing them with
uses of make_scope_exit.

2019-03-06  Tom Tromey  <tom@tromey.com>

* solib-svr4.c (svr4_parse_libraries, svr4_current_sos_direct):
Use make_scope_exit.

5 years agoRemove cleanup from solib-svr4.c
Tom Tromey [Thu, 24 Jan 2019 05:23:15 +0000 (22:23 -0700)]
Remove cleanup from solib-svr4.c

This removes a cleanup from solib-svr4.c, replacing it with
make_scope_exit.

2019-03-06  Tom Tromey  <tom@tromey.com>

* solib-svr4.c (disable_probes_interface): Remove parameter.
(svr4_handle_solib_event): Use make_scope_exit.

5 years agoRemove last cleanup from gdbserver
Tom Tromey [Thu, 24 Jan 2019 05:20:04 +0000 (22:20 -0700)]
Remove last cleanup from gdbserver

This removes the last cleanup from gdbserver, replacing it with
SCOPE_EXIT.  This could perhaps be done in a different way, but this
approach was direct and obviously correct.

2019-03-06  Tom Tromey  <tom@tromey.com>

* server.c (detach_or_kill_for_exit_cleanup): Remove parameter.
(captured_main): Use SCOPE_EXIT.

5 years agoC++ify remote notification code
Tom Tromey [Thu, 24 Jan 2019 05:16:53 +0000 (22:16 -0700)]
C++ify remote notification code

This C++ifies the remote notification code -- replacing function
pointers with virtual methods and using unique_ptr.  This allows for
the removal of some cleanups.

2019-03-06  Tom Tromey  <tom@tromey.com>

* remote.c (struct stop_reply_deleter): Remove.
(stop_reply_up): Update.
(struct stop_reply): Derive from notif_event.  Don't typedef.
<regcache>: Now a std::vector.
(stop_reply_xfree): Remove.
(stop_reply::~stop_reply): Rename from stop_reply_dtr.
(remote_notif_stop_alloc_reply): Return a unique_ptr.  Use new.
(remote_target::discard_pending_stop_replies): Use delete.
(remote_target::remote_parse_stop_reply): Update.
(remote_target::process_stop_reply): Update.
* remote-notif.h (struct notif_event): Add virtual destructor.
Remove "dtr" member.
(struct notif_client) <alloc_event>: Return a unique_ptr.
(notif_event_xfree): Don't declare.
(notif_event_up): New typedef.
* remote-notif.c (remote_notif_ack, remote_notif_parse): Update.
(notif_event_xfree, do_notif_event_xfree): Remove.
(remote_notif_state_xfree): Update.

5 years agoChange displaced_step_clear_cleanup to a forward_scope_exit
Tom Tromey [Wed, 23 Jan 2019 23:28:28 +0000 (16:28 -0700)]
Change displaced_step_clear_cleanup to a forward_scope_exit

This changes displaced_step_clear_cleanup to be a forward_scope_exit
and updates the callers.

gdb/ChangeLog
2019-03-06  Tom Tromey  <tom@tromey.com>

* infrun.c (displaced_step_clear_cleanup): Now a
forward_scope_exit type.
(displaced_step_prepare_throw): Update.
(displaced_step_fixup): Update.

5 years agoUpdate two cleanup comments
Tom Tromey [Wed, 23 Jan 2019 22:11:42 +0000 (15:11 -0700)]
Update two cleanup comments

This updates another couple of comments to remove mentions of
cleanups.

gdb/ChangeLog
2019-03-06  Tom Tromey  <tom@tromey.com>

* inferior.h (class inferior): Update comment.
* gdbthread.h (class thread_info): Update comment.

5 years agoRemove cleanups from coffread.c
Tom Tromey [Sun, 19 Aug 2018 17:50:44 +0000 (11:50 -0600)]
Remove cleanups from coffread.c

This removes the remaining cleanups from coffread.c.

Tested by the buildbot and also some manual testing.

This version includes the fix provided by Joel.

gdb/ChangeLog
2019-03-06  Joel Brobecker  <brobecker@adacore.com>
    Tom Tromey  <tom@tromey.com>

* stabsread.h (struct stab_section_list): Remove.
(coffstab_build_psymtabs): Update.
* dbxread.c (symbuf_sections): Now a std::vector.
(sect_idx): New global.
(fill_symbuf): Update.
(coffstab_build_psymtabs): Change type of stabsects parameter.
Update.
* coffread.c (struct coff_symfile_info) <stabsects>: Now a
std::vector.
(linetab, linetab_offset, linetab_size, stringtab): Move earlier.
(coff_locate_sections): Update.
(coff_symfile_read): Remove cleanups.  Update.
(init_stringtab): Add storage parameter.
(free_stringtab, free_stringtab_cleanup): Remove.
(init_lineno): Add storage parameter.
(free_linetab, free_linetab_cleanup): Remove.

5 years agoEliminate fork_info::clobber_regs
Pedro Alves [Wed, 6 Mar 2019 18:29:19 +0000 (18:29 +0000)]
Eliminate fork_info::clobber_regs

All fork_save_infrun_state callers pass '1' as CLOBBER_REGS nowadays.

The larger hunk in fork_save_infrun_state is just a reindentation.

gdb/ChangeLog:
2019-03-06  Pedro Alves  <palves@redhat.com>

* linux-fork.c (fork_info::clobber_regs): Delete.
(fork_load_infrun_state): Remove reference to 'clobber_regs'.
(fork_save_infrun_state): Remove 'clobber_regs' parameter.  Update
comment.  Adjust.
(scoped_switch_fork_info::scoped_switch_fork_info)
(checkpoint_command, linux_fork_context): Adjust
fork_save_infrun_state calls.

5 years agolinux-fork.c: rewrite inf_has_multiple_threads
Pedro Alves [Wed, 6 Mar 2019 18:29:19 +0000 (18:29 +0000)]
linux-fork.c: rewrite inf_has_multiple_threads

There's no need to iterate over all threads of all inferiors here.

gdb/ChangeLog:
2019-03-06  Pedro Alves  <palves@redhat.com>

* linux-fork.c (inf_has_multiple_thread_cb): Delete.
(inf_has_multiple_threads): Return 'bool' and rewrite using
inferior_info::threads().

5 years agoC++ify fork_info, use std::list
Pedro Alves [Wed, 6 Mar 2019 18:29:18 +0000 (18:29 +0000)]
C++ify fork_info, use std::list

- Convert new_fork and free_fork to fork_info ctor/dtor.
- Use std::list.

gdb/ChangeLog:
2019-03-06  Pedro Alves  <palves@redhat.com>

* linux-fork.c: Include <list>.
(fork_list): Now a std::list instance.
(fork_info): Add ctor, dtor, and in-class initialize all fields.
(forks_exist_p, find_last_fork): Adjust.
(new_fork): Delete.
(one_fork_p): New.
(add_fork): Adjust.
(free_fork): Delete, folded into fork_info::~fork_info().
(delete_fork, find_fork_ptid, find_fork_id, find_fork_pid):
Adjust.
(init_fork_list): Delete.
(linux_fork_killall, linux_fork_mourn_inferior)
(linux_fork_detach, info_checkpoints_command): Adjust.
(_initialize_linux_fork): No longer call init_fork_list.

5 years agoMake "checkpoint" not rely on inferior_ptid
Pedro Alves [Wed, 6 Mar 2019 18:29:17 +0000 (18:29 +0000)]
Make "checkpoint" not rely on inferior_ptid

Don't rely on "inferior_ptid" deep within add_fork.  In the
multi-target branch, I'm forcing inferior_ptid to null_ptid early in
infrun event handling to make sure we don't inadvertently rely on the
current thread/target when we shouldn't, and that caught some bad or
unnecessary assumptions throughout.

gdb/ChangeLog:
2019-03-06  Pedro Alves  <palves@redhat.com>

* linux-fork.c (new_fork): New, split out of ...
(add_fork): ... this.  Return void.  Move "first fork" special
case from here, to ...
(checkpoint_command): ... here.
* linux-linux.h (add_fork): Return void.

5 years agogdb/fortran: Handle older TYPE*SIZE typenames
Andrew Burgess [Mon, 21 Jan 2019 15:10:55 +0000 (15:10 +0000)]
gdb/fortran: Handle older TYPE*SIZE typenames

This patch adds support for the older TYPE*SIZE typenames that are
still around in older code.

For implementation this currently reuses the kind mechanism, as under
gFortran the kind number is equivalent to the size, however, this is
not necessarily true for all compilers.  If the rules for other
compilers are better understood then this code might need to be
improved slightly to allow for a distinction between size and kind,
however, adding this extra complexity now seems pointless.

gdb/ChangeLog:

* f-exp.y (direct_abs_decl): Handle TYPE*SIZE type names.

gdb/testsuite/ChangeLog:

* gdb.fortran/type-kinds.exp: Extend to cover TYPE*SIZE cases.

5 years agogdb/fortran: Add support for the ABS intrinsic function
Andrew Burgess [Fri, 18 Jan 2019 14:44:48 +0000 (14:44 +0000)]
gdb/fortran: Add support for the ABS intrinsic function

Adds support for the abs intrinsic function, this requires adding a
new pattern to the Fortran parser.  Currently only float and integer
argument types are supported to ABS, complex is still not supported,
this can be added later if needed.

gdb/ChangeLog:

* f-exp.y: New token, UNOP_INTRINSIC.
(exp): New pattern using UNOP_INTRINSIC token.
(f77_keywords): Add 'abs' keyword.
* f-lang.c: Add 'target-float.h' and 'math.h' includes.
(value_from_host_double): New function.
(evaluate_subexp_f): Support UNOP_ABS.

gdb/testsuite/ChangeLog:

* gdb.fortran/intrinsics.exp: Extend to cover ABS.

5 years agogdb/fortran: Use TYPE_CODE_CHAR for character types
Andrew Burgess [Fri, 18 Jan 2019 11:24:24 +0000 (11:24 +0000)]
gdb/fortran: Use TYPE_CODE_CHAR for character types

Switch to using TYPE_CODE_CHAR for character types.  This appears to
have little impact on the test results as gFortran uses the
DW_TAG_string_type to represent all character variables (as far as I
can see).  The only place this has an impact is when the user casts a
variable to a character type, in which case GDB does now use the CHAR
type, and prints the variable as both a value and a character, for
example, before:

    (gdb) p ((character) 97)
    $1 = 97

and after:

    (gdb) p ((character) 97)
    $1 = 97 'a'

gdb/ChangeLog:

* f-lang.c (build_fortran_types): Use TYPE_CODE_CHAR for character
types.

gdb/testsuite/ChangeLog:

* gdb.fortran/type-kinds.exp: Update expected results.

5 years agogdb/fortran: Add builtin 8-byte integer type with (kind=8) support
Andrew Burgess [Thu, 17 Jan 2019 16:31:56 +0000 (16:31 +0000)]
gdb/fortran: Add builtin 8-byte integer type with (kind=8) support

Add a new builtin type, an 8-byte integer, and allow GDB to parse
'integer (kind=8)', returning the new 8-byte integer.

gdb/ChangeLog:

* f-exp.y (convert_to_kind_type): Handle integer (kind=8).
* f-lang.c (build_fortran_types): Setup builtin_integer_s8.
* f-lang.h (struct builtin_f_type): Add builtin_integer_s8 field.

gdb/testsuite/ChangeLog:

* gdb.fortran/type-kinds.exp: Test new integer type kind.

5 years agogdb/fortran: Expand the set of types that support (kind=N)
Andrew Burgess [Thu, 17 Jan 2019 16:30:35 +0000 (16:30 +0000)]
gdb/fortran: Expand the set of types that support (kind=N)

Expand the number of types that can be adjusted with a (kind=N) type
extension.

gdb/ChangeLog:

* f-exp.y (convert_to_kind_type): Handle more type kinds.

gdb/testsuite/ChangeLog:

* gdb.fortran/type-kinds.exp (test_cast_1_to_type_kind): New
function.
(test_basic_parsing_of_type_kinds): Expand types tested.
(test_parsing_invalid_type_kinds): New function.

5 years agogdb/fortran: Add Fortran 'kind' intrinsic and keyword
Andrew Burgess [Wed, 16 Jan 2019 16:42:10 +0000 (16:42 +0000)]
gdb/fortran: Add Fortran 'kind' intrinsic and keyword

The 'kind' keyword has two uses in Fortran, it is the name of a
builtin intrinsic function, and it is also a keyword used to create a
type of a specific kind.

This commit adds support for using kind as an intrinsic function, and
also adds some initial support for using kind to create types of a
specific kind.

This commit only allows the creation of the type 'character(kind=1)',
however, it will be easy enough to extend this in future to support
more type kinds.

The kind of any expression can be queried using the kind intrinsic
function.  At the moment the kind returned corresponds to the size of
the type, this matches how gfortran handles kinds.  However, the
correspondence between kind and type size depends on the compiler
and/or the specific target, so this might not be correct for
everyone.  If we want to support different compilers/targets in future
the code to compute the kind from a type will need to be updated.

gdb/ChangeLog:

* expprint.c (dump_subexp_body_standard): Support UNOP_KIND.
* f-exp.y: Define 'KIND' token.
(exp): New pattern for KIND expressions.
(ptype): Handle types with a kind extension.
(direct_abs_decl): Extend to spot kind extensions.
(f77_keywords): Add 'kind' to the list.
(push_kind_type): New function.
(convert_to_kind_type): New function.
* f-lang.c (evaluate_subexp_f): Support UNOP_KIND.
* parse.c (operator_length_standard): Likewise.
* parser-defs.h (enum type_pieces): Add tp_kind.
* std-operator.def: Add UNOP_KIND.

gdb/testsuite/ChangeLog:

* gdb.fortran/intrinsics.exp: New file.
* gdb.fortran/intrinsics.f90: New file.
* gdb.fortran/type-kinds.exp: New file.

5 years agogdb/fortran: Enable debugging of the Fortran parser
Andrew Burgess [Thu, 17 Jan 2019 14:42:15 +0000 (14:42 +0000)]
gdb/fortran: Enable debugging of the Fortran parser

This commit allows 'set debug parser on' to work for the Fortran
parser.

gdb/ChangeLog:

* f-exp.y (f_parse): Set yydebug.

5 years agogdb/fortran: Add new function to evaluate Fortran expressions
Andrew Burgess [Wed, 16 Jan 2019 16:16:59 +0000 (16:16 +0000)]
gdb/fortran: Add new function to evaluate Fortran expressions

This is an initial restructure, it adds a new function in which
Fortran specific expressions can be evaluated.  No Fortran specific
expressions are added with this commit though, so for now, the new
function just forwards all expressions to the default expression
handler, as such, there should be no user visible changes after this
commit.  However, the new function will be useful in later commits.

gdb/ChangeLog:

* f-lang.c (evaluate_subexp_f): New function.
(exp_descriptor_f): New global.
(f_language_defn): Use exp_descriptor_f instead of
exp_descriptor_standard.

5 years agogdb/fortran: Simplify handling of Fortran dot operations and keywords
Andrew Burgess [Wed, 16 Jan 2019 15:30:54 +0000 (15:30 +0000)]
gdb/fortran: Simplify handling of Fortran dot operations and keywords

Use strncasecmp to compare Fortran dot operations (like .AND.) and for
the keywords list.  This allows for some duplication to be removed
from the token arrays.  I've also performed whitespace cleanup around
the code I've changed.

I have added some tests to ensure that upper and lowercase dot
operations are correctly tested.  The keywords list remains always
lowercase for now.

There should be no user visible changes after this commit.

gdb/ChangeLog:

* f-exp.y (struct token): Add comments.
(dot_ops): Remove uppercase versions and the end marker.
(f77_keywords): Likewise.
(yylex): Use ARRAY_SIZE to iterate over dot_ops, assert all
entries in the dot_ops array are case insensitive, and use
strncasecmp to compare strings.  Also some whitespace cleanup in
this area.  Similar for the f77_keywords array, except entries in
this list might be case sensitive.

gdb/testsuite/ChangeLog:

* gdb.fortran/dot-ops.exp: New file.

5 years agogdb/fortran: Cleanup code for parsing logical constants
Andrew Burgess [Wed, 16 Jan 2019 13:36:46 +0000 (13:36 +0000)]
gdb/fortran: Cleanup code for parsing logical constants

This patch cleans up the code used for parsing the Fortran logical
constants '.TRUE.' and '.FALSE.'.  Instead of listing both upper and
lowercase versions of these strings we now use strncasecmp.

I've also switched to use ARRAY_SIZE for the array iteration, and I've
cleaned up whitespace in the vicinity of the code I've changed.

Finally, I've added a test to ensure that both the upper and lower
case versions of the logical constants are understood by GDB,
something that was missing previously.

There should be no user visible changes after this commit.

gdb/ChangeLog:

* f-exp.y (struct f77_boolean_val): Add comments.
(boolean_values): Remove uppercase versions, and end marker.
(yylex): Use ARRAY_SIZE for iterating over boolean_values array,
and use strncasecmp to achieve case insensitivity.  Additionally,
perform whitespace cleanup around this code.

gdb/testsuite/ChangeLog:

* gdb.fortran/types.exp (test_logical_literal_types_accepted):
Check upper and lower case logical literals.

5 years agogdb/fortran: Remove some duplicate tests
Andrew Burgess [Wed, 16 Jan 2019 13:20:28 +0000 (13:20 +0000)]
gdb/fortran: Remove some duplicate tests

Make the test names unique in gdb.fortran/types.exp by removing a few
duplicate tests.

gdb/testsuite/ChangeLog:

* gdb.fortran/types.exp (test_float_literal_types_accepted):
Remove duplicate tests.

5 years agoAllow the use of the ORIGIN and LENGTH attributes in expressions inside MEMORY regions.
Nick Bowler [Wed, 6 Mar 2019 15:20:29 +0000 (15:20 +0000)]
Allow the use of the ORIGIN and LENGTH attributes in expressions inside MEMORY regions.

PR 24289
* ldexp.c (fold_name): Allow lookups of the LENGTH and ORIGIN
attributes during the first phase.

5 years agoFix remote-sim.c build after warn-unused-result change
Tom Tromey [Wed, 6 Mar 2019 14:55:42 +0000 (07:55 -0700)]
Fix remote-sim.c build after warn-unused-result change

John Darrington pointed out that commit 18cb7c9f3 ("Introduce
ATTRIBUTE_UNUSED_RESULT and use it") broke the build:

/home/john/binutils-gdb/gdb/remote-sim.c: In function 'void gdbsim_target_open(const char*, int)':
/home/john/binutils-gdb/gdb/remote-sim.c:765:18: error: ignoring return value of 'char** gdb_argv::release()', declared with attribute warn_unused_result [-Werror=unused-result]

This patch fixes the problem by arranging to use the result of the
release method.

Tested by rebuilding with a simulator enabled.

gdb/ChangeLog
2019-03-06  Tom Tromey  <tromey@adacore.com>

* remote-sim.c (gdbsim_target_open): Use result of
gdb_argv::release.

5 years agoTestsuite: Ensure changing directory does not break the log file
Alan Hayward [Wed, 6 Mar 2019 09:52:08 +0000 (09:52 +0000)]
Testsuite: Ensure changing directory does not break the log file

get_compiler_info switches to a new log file before checking the compiler
to ensure the checks are not logged. Afterwards it restores back to using
the original log file. However, the logfile uses a relative path name -
if the current test has changed the current directory then all further
output for the test will be lost.  This can confuse the code that collates
the main gdb.log file at the end of a FORCE_PARALLEL run.

fullpath-expand.exp calls gdb_compile after changing the current directory.

The "Ensure stack protection is off for GCC" patch added a call to
get_compiler_info from inside of gdb_compile, causing log file collection
to break for FORCE_PARALLEL runs.

The ideal solution would be to ensure the log file is always created using
an absolute path name. However, this is set at multiple points in
Makefile.in and in some instances just relies on dejagnu common code to set
the log file directory to "."

The simpler and safer solution is to override the builtin cd function. The
new function checks the current log file and if the path is relative, then
it resets the logging using an absolute path. Finally it calls the builtin
cd.  This ensures get_compiler_info (and any other code) can correctly
backup and restore the current log file.

gdb/testsuite/ChangeLog:

* lib/gdb.exp (builtin_cd): rename of cd.
(cd): Override builtin.

5 years agoFix a segmentation fault triggered by disassembling an EFi file with source included.
Nick Clifton [Wed, 6 Mar 2019 09:43:15 +0000 (09:43 +0000)]
Fix a segmentation fault triggered by disassembling an EFi file with source included.

* dwarf2.c (_bfd_dwarf2_find_symbol_bias): Check for a NULL symbol
table pointer.
* coffgen.c (coff_find_nearest_line_with_names): Do not call
_bfd_dwarf2_find_symbol_bias if there is no symbol table available.

https://bugzilla.redhat.com/show_bug.cgi?id=1685727

5 years agoFortran function calls with arguments
Richard Bunt [Wed, 6 Mar 2019 08:23:00 +0000 (08:23 +0000)]
Fortran function calls with arguments

Prior to this patch, calling functions on the inferior with arguments and
then using these arguments within a function resulted in an invalid
memory access. This is because Fortran arguments are typically passed as
pointers to values.

It is possible to call Fortran functions, but memory must be allocated in
the inferior, so a pointer can be passed to the function, and the
language must be set to C to enable C-style casting. This is cumbersome
and not a pleasant debug experience.

This patch implements the GNU Fortran argument passing conventions with
caveats. Firstly, it does not handle the VALUE attribute as there is
insufficient DWARF information to determine when this is the case.
Secondly, functions with optional parameters can only be called with all
parameters present. Both these cases are marked as KFAILS in the test.

Since the GNU Fortran argument passing convention has been implemented,
there is no guarantee that this patch will work correctly, in all cases,
with other compilers.

Despite these limitations, this patch improves the ease with which
functions can be called in many cases, without taking away the existing
approach of calling with the language set to C.

Regression tested on x86_64, aarch64 and POWER9 with GCC 7.3.0.
Regression tested with Ada on x86_64.
Regression tested with native-extended-gdbserver target board.

gdb/ChangeLog:

* eval.c (evaluate_subexp_standard): Call Fortran argument
wrapping logic.
* f-lang.c (struct value): A value which can be passed into a
Fortran function call.
(fortran_argument_convert): Wrap Fortran arguments in a pointer
where appropriate.
(struct type): Value ready for a Fortran function call.
(fortran_preserve_arg_pointer): Undo check_typedef, the pointer
is needed.
* f-lang.h (fortran_argument_convert): Declaration.
(fortran_preserve_arg_pointer): Declaration.
* infcall.c (value_arg_coerce): Call Fortran argument logic.

gdb/testsuite/ChangeLog:

* gdb.fortran/function-calls.exp: New file.
* gdb.fortran/function-calls.f90: New test.

5 years agoAutomatic date update in version.in
GDB Administrator [Wed, 6 Mar 2019 00:00:46 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoRemove some Python 3 #ifs
Tom Tromey [Tue, 5 Mar 2019 21:58:24 +0000 (14:58 -0700)]
Remove some Python 3 #ifs

A recent patch from Kevin Buettner taught me that the PyBytes API is
available on Python 2.  This patch removes a couple of related #ifs in
the Python code.

Tested on x86-64 Fedora 29, using both Python 3.7 and Python 2.7.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* python/py-prettyprint.c (print_string_repr): Remove #if.
* python/py-utils.c (unicode_to_encoded_string): Remove #if.

5 years agoChange the_dummy_target to be a global
Tom Tromey [Mon, 4 Mar 2019 20:37:59 +0000 (13:37 -0700)]
Change the_dummy_target to be a global

While debugging gdb, I printed the target stack and got:

    (top-gdb) p g_target_stack
    $10 = {
      m_top = thread_stratum,
      m_stack = {0x142b0b0, 0x13da600 <exec_ops>, 0x1c70690, 0x13d63b0 <ravenscar_ops>, 0x0, 0x0, 0x0}
    }

(This is clearly from before the change to make ravenscar
multi-target-capable.)

Here, 0x142b0b0 is the singleton dummy target.  It seems to me that
since this is always a singleton, it would be a bit nicer if it were a
global, so that it would be noted in the above.

This patch implements this idea, and now I get:

    (top-gdb) p g_target_stack
    $2 = {
      m_top = dummy_stratum,
      m_stack = {0x1f1b040 <the_dummy_target>, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}
    }

I did not do the same for the debug target.  It didn't seem as useful
to me.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* target.c (the_dummy_target): Move later.  Change type to
"dummy_target".
(initialize_targets): Don't initialize the_dummy_target.

5 years agoSmall fix to gdb.Value constructor doc
Simon Marchi [Tue, 5 Mar 2019 19:00:42 +0000 (14:00 -0500)]
Small fix to gdb.Value constructor doc

The synopsis of the two-parameters form of the gdb.Value constructor is
currently shown as

    Value.__init__ (val, [, type ])

in the documentation.

First, there is an extra comma, which I think we can remove in any
case.

Then, since the type parameter is not optional, I would not put in
between square brackets.  Those usually indicate that something is
optional.

With this patch, it appears as:

    Value.__init__ (val, type)

gdb/doc/ChangeLog:

* python.texi (Values From Inferior): Change synopsys of the
second form of Value.__init__.

5 years agoFix a spelling mistake: "complaint" instead of "compliant".
Nick Clifton [Tue, 5 Mar 2019 17:32:35 +0000 (17:32 +0000)]
Fix a spelling mistake: "complaint" instead of "compliant".

PR 24295
* doc/binutils.texi (ar cmdline): Fix spelling mistake.

5 years agoRemove gdb_bfd_fdopenr
Tom Tromey [Tue, 5 Mar 2019 16:03:55 +0000 (09:03 -0700)]
Remove gdb_bfd_fdopenr

I noticed that gdb_bfd_fdopenr is no longer used, so this patch
removes it.  Tested by rebuilding and by grep.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* gdb_bfd.c (gdb_bfd_fdopenr): Remove.
* gdb_bfd.h (gdb_bfd_fdopenr): Don't declare.

5 years agoUpdated Swedish translation for the binutils sub-directory.
Nick Clifton [Tue, 5 Mar 2019 16:04:18 +0000 (16:04 +0000)]
Updated Swedish translation for the binutils sub-directory.

5 years agoRemove excess calls to gdb_flush
Tom Tromey [Tue, 19 Feb 2019 20:36:17 +0000 (13:36 -0700)]
Remove excess calls to gdb_flush

A customer noticed some mildly odd MI output, where CLI output was
split into multiple MI strings at unusual boundaries, like this:

    ~"$1 = (b => true"
    ~", p => 0x407260"

This is technically correct according to the MI spec, but still
unusual, in that there's no particular reason for the string to be
split where it is.

I tracked this down to a call to gdb_flush in generic_val_print.
Then, I went through all calls to gdb_flush and removed the ones I
thought were superfluous.  In particular:

* Any call in the value-printing code;
* Likewise the type-printing code (just a single call); and
* Any call that immediately followed a printf that obviously
  ended with a newline, my belief being that gdb's standard output
  streams are line buffered (by inheriting the behavior from stdio)

Regression tested on x86-64 Fedora 29.

I didn't add a new test case.  I tend to think we don't necessarily
want to specify this behavior in the tests.  Let me know what you
think of this.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* windows-nat.c (windows_nat_target::attach)
(windows_nat_target::detach): Don't call gdb_flush.
* valprint.c (generic_val_print, val_print, val_print_string):
Don't call gdb_flush.
* utils.c (defaulted_query): Don't call gdb_flush.
* typeprint.c (print_type_scalar): Don't call gdb_flush.
* target.c (target_announce_detach): Don't call gdb_flush.
* sparc64-tdep.c (adi_print_versions): Don't call gdb_flush.
* remote.c (extended_remote_target::attach): Don't call
gdb_flush.
* procfs.c (procfs_target::detach): Don't call gdb_flush.
* printcmd.c (do_examine): Don't call gdb_flush.
(info_display_command): Don't call gdb_flush.
* p-valprint.c (pascal_val_print): Don't call gdb_flush.
* nto-procfs.c (nto_procfs_target::attach): Don't call gdb_flush.
* memattr.c (info_mem_command): Don't call gdb_flush.
* mdebugread.c (mdebug_build_psymtabs): Don't call gdb_flush.
* m2-valprint.c (m2_val_print): Don't call gdb_flush.
* infrun.c (follow_exec, handle_command): Don't call gdb_flush.
* inf-ptrace.c (inf_ptrace_target::attach): Don't call gdb_flush.
* hppa-tdep.c (unwind_command): Don't call gdb_flush.
* gnu-nat.c (gnu_nat_target::attach): Don't call gdb_flush.
(gnu_nat_target::detach): Don't call gdb_flush.
* f-valprint.c (f_val_print): Don't call gdb_flush.
* darwin-nat.c (darwin_nat_target::attach): Don't call gdb_flush.
* cli/cli-script.c (read_command_lines): Don't call gdb_flush.
* cli/cli-cmds.c (shell_escape, print_disassembly): Don't call
gdb_flush.
* c-valprint.c (c_val_print): Don't call gdb_flush.
* ada-valprint.c (ada_print_scalar): Don't call gdb_flush.

5 years agoAdd ATTRIBUTE_UNUSED_RESULT to ref_ptr::release
Tom Tromey [Tue, 26 Feb 2019 21:59:47 +0000 (14:59 -0700)]
Add ATTRIBUTE_UNUSED_RESULT to ref_ptr::release

This applies ATTRIBUTE_UNUSED_RESULT to ref_ptr::release and updates a
few spots to comply.  I believe one use in install_default_visualizer
was in error, fixed by this patch.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* varobj.c (update_dynamic_varobj_children): Update.
(install_default_visualizer): Use reset, not release.
* value.c (set_internalvar): Update.
* dwarf2loc.c (value_of_dwarf_reg_entry): Update.
* common/gdb_ref_ptr.h (class ref_ptr) <release>: Add
ATTRIBUTE_UNUSED_RESULT.

5 years agoAdd ATTRIBUTE_UNUSED_RESULT to scoped_remote_fd::release
Tom Tromey [Tue, 26 Feb 2019 21:52:47 +0000 (14:52 -0700)]
Add ATTRIBUTE_UNUSED_RESULT to scoped_remote_fd::release

This applies ATTRIBUTE_UNUSED_RESULT to scoped_remote_fd::release.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* remote.c (class scoped_remote_fd) <release>: Add
ATTRIBUTE_UNUSED_RESULT.

5 years agoAdd ATTRIBUTE_UNUSED_RESULT to macro_buffer
Tom Tromey [Tue, 26 Feb 2019 21:52:05 +0000 (14:52 -0700)]
Add ATTRIBUTE_UNUSED_RESULT to macro_buffer

This applies ATTRIBUTE_UNUSED_RESULT to macro_buffer::release.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* macroexp.c (struct macro_buffer) <release>: Add
ATTRIBUTE_UNUSED_RESULT.

5 years agoAdd ATTRIBUTE_UNUSED_RESULT to scoped_mmap
Tom Tromey [Tue, 26 Feb 2019 21:46:40 +0000 (14:46 -0700)]
Add ATTRIBUTE_UNUSED_RESULT to scoped_mmap

This applies ATTRIBUTE_UNUSED_RESULT to scoped_mmap::release and fixes
a couple of spots to comply.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* nat/linux-btrace.c (linux_enable_bts, linux_enable_pt): Update.
* common/scoped_mmap.h (class scoped_mmap) <release>: Add
ATTRIBUTE_UNUSED_RESULT.

5 years agoAdd ATTRIBUTE_UNUSED_RESULT to scoped_fd
Tom Tromey [Tue, 26 Feb 2019 21:44:29 +0000 (14:44 -0700)]
Add ATTRIBUTE_UNUSED_RESULT to scoped_fd

This applies ATTRIBUTE_UNUSED_RESULT to scoped_fd::release.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* common/scoped_fd.h (class scoped_fd) <release>: Add
ATTRIBUTE_UNUSED_RESULT.

5 years agoAdd ATTRIBUTE_UNUSED_RESULT to parser_state
Tom Tromey [Tue, 26 Feb 2019 21:39:55 +0000 (14:39 -0700)]
Add ATTRIBUTE_UNUSED_RESULT to parser_state

This applies ATTRIBUTE_UNUSED_RESULT to parser_state::release.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* parser-defs.h (struct parser_state) <release>: Add
ATTRIBUTE_UNUSED_RESULT.

5 years agoIntroduce ATTRIBUTE_UNUSED_RESULT and use it
Tom Tromey [Tue, 26 Feb 2019 21:39:01 +0000 (14:39 -0700)]
Introduce ATTRIBUTE_UNUSED_RESULT and use it

This introduces the new ATTRIBUTE_UNUSED_RESULT define, and applies it
to gdb_argv::release.

gdb/ChangeLog
2019-03-05  Tom Tromey  <tromey@adacore.com>

* utils.h (class gdb_argv) <release>: Add
ATTRIBUTE_UNUSED_RESULT.
* common/common-defs.h (ATTRIBUTE_UNUSED_RESULT): Define.

5 years agoAllow multiple tests to be run on systems with older versions of TCL installed.
Matthew Malcomson [Tue, 5 Mar 2019 10:09:50 +0000 (10:09 +0000)]
Allow multiple tests to be run on systems with older versions of TCL installed.

PR 24287
* testsuite/lib/binutils-common.exp (run_dump_test): Replace a use
of "lmap" with a "foreach" loop.

5 years agoAutomatic date update in version.in
GDB Administrator [Tue, 5 Mar 2019 00:00:26 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoUse '$enable_unittest' instead of '$development' on gdbserver/configure.srv (for...
Sergio Durigan Junior [Mon, 4 Mar 2019 17:44:20 +0000 (12:44 -0500)]
Use '$enable_unittest' instead of '$development' on gdbserver/configure.srv (for 'aarch64*-*-linux*' case)

On commit 8ecfd7bd4acd69213c06fac6de9af38299123547 ("Add parameter to
allow enabling/disabling selftests via configure") it seems that I
forgot to use the proper '$enable_unittest' variable when checking to
see whether to add selftest-related objects to 'srv_regobj'.  This
causes a build failure on Aarch64 when 'development=false' (which is
the case for the 8.3 branch) and 'enable_unittest=true'.

This patch fixes the problem by using '$enable_unittest' instead of
'$development' when performing the check.  As a reminder, it's
important to notice that '$enable_unittest's default value (i.e., when
the option '--enable-unit-tests' is not passed to configure) is the
same as '$development', so this patch doesn't affect the current
build.

I'd like to install this patch both on master and on the 8.3 branch.

OK?

gdb/gdbserver/ChangeLog:
2019-03-04  Sergio Durigan Junior  <sergiodj@redhat.com>

* configure.srv: Use '$enable_unittest' instead of '$development'
when checking whether to fill 'srv_regobj' on 'aarch64*-*-linux*'
case.

5 years agogdbserver short-circuit-argument-list failures
Richard Bunt [Mon, 4 Mar 2019 15:08:51 +0000 (15:08 +0000)]
gdbserver short-circuit-argument-list failures

This patch fixes test case failures observed when running
short-circuit-argument-list.exp with gdb server boards. Thanks to Sergio
Durigan Junior for pointing this out.

Assertions failed with the native{,-extended}-gdbserver boards as the
standard output from the test program appears in a different location
than observed on non-gdbserver boards. This standard output was used to
determine whether a function, which had been logically short-circuited,
was called or not. Since the location of the standard out cannot be
relied upon to verify this, a new mechanism was needed.

The test program now records function calls in variables named the same
as the function with a "_called" suffix. These variables can then be
queried from the test case to verify the occurrence of a call.

A method to reset the call counts has been included in the test case, so
that any future assertions added to this test can ensure a fresh set of
initial values before proceeding. Not resetting values between groups of
assertions creates a dependency between them, which increases the
likelihood that a single failure causes subsequent assertions to fail.

Regression tested on x86_64, aarch64 and ppc64le.
Regression tested with Ada on x86_64.
Regression tested with the native{,-extended}-gdbserver boards on x86_64.

5 years agoStop objcopy from attempting to copy thin archives.
Nick Clifton [Mon, 4 Mar 2019 13:11:08 +0000 (13:11 +0000)]
Stop objcopy from attempting to copy thin archives.

PR 24281
* objcopy.c (copy_archive): Do not copy thin  archives.

5 years agoAutomatic date update in version.in
GDB Administrator [Mon, 4 Mar 2019 00:00:30 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoAutomatic date update in version.in
GDB Administrator [Sun, 3 Mar 2019 00:00:38 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoFix documentation of styles.
Eli Zaretskii [Sat, 2 Mar 2019 19:42:54 +0000 (21:42 +0200)]
Fix documentation of styles.

gdb/doc/ChangeLog:
2019-03-02  Eli Zaretskii  <eliz@gnu.org>

* gdb.texinfo (Output Styling): Fix typos.  Document the default
foreground colors of the available styles.

5 years agoAvoid compilation warnings on MinGW.
Eli Zaretskii [Sat, 2 Mar 2019 13:22:11 +0000 (15:22 +0200)]
Avoid compilation warnings on MinGW.

gdb/ChangeLog:
2019-03-02  Eli Zaretskii  <eliz@gnu.org>

* xml-syscall.c (xml_list_syscalls_by_group): Drop 'struct' from
for-loop range, to avoid compiler warnings.

* tui/tui.c (tui_enable) [__MINGW32__]: Don't declare 'cap', to
avoid compiler warnings about unused variables.

5 years agoGDB no longer supports Windows before XP.
Eli Zaretskii [Sat, 2 Mar 2019 13:18:32 +0000 (15:18 +0200)]
GDB no longer supports Windows before XP.

gdb/ChangeLog:
2019-03-02  Eli Zaretskii  <eliz@gnu.org>

* NEWS: Mention end of support for native debugging on MS-Windows
before XP.

5 years agoFix GDB compilation on MinGW (PR gdb/24292)
Eli Zaretskii [Sat, 2 Mar 2019 13:13:54 +0000 (15:13 +0200)]
Fix GDB compilation on MinGW (PR gdb/24292)

gdb/ChangeLog:

2019-03-02  Eli Zaretskii  <eliz@gnu.org>

PR gdb/24292
* common/netstuff.c:
* gdbserver/gdbreplay.c
* gdbserver/remote-utils.c:
* ser-tcp.c:
* unittests/parse-connection-spec-selftests.c [USE_WIN32API]:
Include ws2tcpip.h instead of wsiapi.h and winsock2.h.  Redefine
_WIN32_WINNT to 0x0501 if defined to a smaller value, as
'getaddrinfo' and 'freeaddrinfo' were not available before
Windows XP, and mingw.org's MinGW headers by default define
_WIN32_WINNT to 0x500.

5 years agoAutomatic date update in version.in
GDB Administrator [Sat, 2 Mar 2019 00:00:57 +0000 (00:00 +0000)]
Automatic date update in version.in

5 years agoRevert "Add support to GNU ld to separate got related plt entries"
Andreas Krebbel [Fri, 1 Mar 2019 14:23:42 +0000 (15:23 +0100)]
Revert "Add support to GNU ld to separate got related plt entries"

bfd/ChangeLog:

2019-03-01  Andreas Krebbel  <krebbel@linux.ibm.com>

This reverts commit 5a12586d44fa8d5dfc74cbca4f2f36a273a16335.
2019-01-14  Maamoun Tarsha  <maamountk@hotmail.com>

PR 20113
* elf32-s390.c (allocate_dynrelocs): Update comment.

ld/ChangeLog:

2019-03-01  Andreas Krebbel  <krebbel@linux.ibm.com>

This reverts commit 5a12586d44fa8d5dfc74cbca4f2f36a273a16335.
2019-01-14  Maamoun Tarsha  <maamountk@hotmail.com>

PR 20113
* emulparams/elf64_s390.sh (SEPARATE_GOTPLT): Define.
* emulparams/elf_s390.sh (SEPARATE_GOTPLT): Define.
* testsuite/ld-s390/gotreloc_31-1.dd: Update expected output.
* testsuite/ld-s390/tlsbin.dd: Likewise.
* testsuite/ld-s390/tlsbin.rd: Likewise.
* testsuite/ld-s390/tlsbin.sd: Likewise.
* testsuite/ld-s390/tlsbin_64.dd: Likewise.
* testsuite/ld-s390/tlsbin_64.rd: Likewise.
* testsuite/ld-s390/tlsbin_64.sd: Likewise.
* testsuite/ld-s390/tlspic.dd: Likewise.
* testsuite/ld-s390/tlspic.rd: Likewise.
* testsuite/ld-s390/tlspic.sd: Likewise.
* testsuite/ld-s390/tlspic_64.dd: Likewise.
* testsuite/ld-s390/tlspic_64.rd: Likewise.
* testsuite/ld-s390/tlspic_64.sd: Likewise.
* testsuite/ld-s390/s390.exp: Skip s390 tests for tpf targets.

5 years agoFix coff_start_symtab resource leak found by Coverity
Gary Benson [Fri, 1 Mar 2019 13:50:27 +0000 (13:50 +0000)]
Fix coff_start_symtab resource leak found by Coverity

This commit fixes a resource leak found by Coverity, where
coff_start_symtab performs an xstrdup that is now performed
within start_symtab by buildsym_compunit::buildsym_compunit.

gdb/ChangeLog:

* coffread.c (coff_start_symtab): Remove unnecessary xstrdup.

5 years agoBinutils: Always skip only 1 byte for CIE version 1's return address register.
Tamar Christina [Fri, 1 Mar 2019 11:37:51 +0000 (11:37 +0000)]
Binutils: Always skip only 1 byte for CIE version 1's return address register.

According to the specification for the CIE entries, when the CIE version is 1 then
the return address register field is always 1 byte.  Readelf does this correctly in
read_cie in dwarf.c but ld does this incorrectly and always tries to read a
skip_leb128.  If the value here has the top bit set then ld will incorrectly read
at least another byte, causing either an assert failure or an incorrect address to
be used in eh_frame.

I'm not sure how to generate a generic test for this as I'd need to write assembly,
and it's a bit hard to trigger. Essentially the relocated value needs to start with
something that & 0x70 != 0x10 while trying to write a personality.

bfd/ChangeLog:

* elf-eh-frame.c (_bfd_elf_write_section_eh_frame): Correct CIE parse.

5 years agoPR24272, out-of-bounds read in pex64_xdata_print_uwd_codes
Alan Modra [Thu, 28 Feb 2019 22:58:47 +0000 (09:28 +1030)]
PR24272, out-of-bounds read in pex64_xdata_print_uwd_codes

The fix here is to use an unsigned comparison for
    if (a->NumberOfRvaAndSizes > IMAGE_NUMBEROF_DIRECTORY_ENTRIES)

include/
PR 24272
* coff/internal.h (struct internal_extra_pe_aouthdr): Change type
of SizeOfCode, SizeOfInitializedData, and SizeOfUninitializedData
to bfd_vma.  Change type of SectionAlignment, FileAlignment,
Reserved1, SizeOfImage, SizeOfHeaders, CheckSum, LoaderFlags,
and NumberOfRvaAndSizes to uint32_t.
bfd/
PR 24272
* peXXigen.c (_bfd_XXi_swap_aouthdr_in): Use unsigned index.
(_bfd_XX_print_private_bfd_data_common): Adjust for type changes.

5 years agoAutomatic date update in version.in
GDB Administrator [Fri, 1 Mar 2019 00:00:52 +0000 (00:00 +0000)]
Automatic date update in version.in