Pedro Alves [Fri, 15 Sep 2017 16:40:33 +0000 (17:40 +0100)]
Sync libiberty/ & include/ with GCC
Note this brings in the interface files for libcc1/G++ as well, which
we will be needing in GDB soon anyway. That commit renamed a method
in the C interface and that required a small update to GDB's compile/
code, which I've included that in this patch to keep the tree
building.
include/ChangeLog:
2017-09-15 Pedro Alves <palves@redhat.com>
* ansidecl.h (DISABLE_COPY_AND_ASSIGN): New macro.
2017-09-12 Jiong Wang <jiong.wang@arm.com>
* dwarf2.def (DW_CFA_AARCH64_negate_ra_state): New DW_CFA_DUP.
* dwarf2.h (DW_CFA_DUP): New define.
2017-08-21 Richard Biener <rguenther@suse.de>
* simple-object.h (simple_object_copy_lto_debug_sections): New
function.
2017-05-18 Martin Liska <mliska@suse.cz>
* ansidecl.h: Define CONSTEXPR macro.
2017-05-24 Nathan Sidwell <nathan@acm.org>
* libiberty.h (ASTRDUP): Adjust cast to avoid warning.
2017-01-30 Alexandre Oliva <aoliva@redhat.com>
Introduce C++ support in libcc1.
* gcc-c-fe.def (int_type_v0): Rename from...
(int_type): ... this. Introduce new version.
(float_type_v0): Rename from...
(float_type): ... this. Introduce new version.
(char_type): New.
* gcc-c-interface.h (gcc_c_api_version): Add GCC_C_FE_VERSION_1.
(gcc_type_array): Move...
* gcc-interface.h: ... here.
* gcc-cp-fe.def: New.
* gcc-cp-interface.h: New.
* cp-demangle.c (is_fnqual_component_type): Reimplement using
FNQUAL_COMPONENT_CASE.
(d_encoding): Hold bare_function_type in local var.
(d_local_name): Build name in both cases and build result once.
Collapse switch-if to single conditional.
(d_local_name):
* testsuite/demangle-expected: Realign blank lines with tests.
2017-09-12 Jiong Wang <jiong.wang@arm.com>
* dwarfnames.c (DW_CFA_DUP): New define.
gdb/ChangeLog:
2017-09-15 Pedro Alves <palves@redhat.com>
* compile/compile-c-types.c (convert_enum, convert_int)
(convert_float): Adjust to refer to int_type_v0 and float_type_v0.
Simon Marchi [Fri, 15 Sep 2017 16:02:51 +0000 (18:02 +0200)]
gdbserver: Remove thread_to_gdb_id
As explained in the previous patch, the gdb_id concept is no longer
relevant. The function thread_to_gdb_id is trivial, it returns the
thread's ptid. Remove it and replace its usage with ptid_of.
The changes in nto-low.c and lynx-low.c are fairly straightforward, but
I was not able to build test them.
There was problem when using a 32-bits gdb with a 64-bits gdbserver.
For some reason that I don't fully understand, the thread ids
exchanged between gdb and gdbserver could overflow a 32 bits data
type. My guess is that they were the thread address (e.g. the
0x7ffff7f20b40 in "Thread 0x7ffff7f20b40 (LWP 1058)" today). This
patch changed that so gdb/gdbserver would talk in terms of the OS
assigned numerical id (as shown in ps). It therefore added a way to
convert between this gdb_id (the numerical id) and the thread id (the
address).
- 95954743cb Implement the multiprocess extensions, and add linux
multiprocess supportNon-stop mode support.
This patch made gdbserver deal with threads using their numerical ids
and not the address-like id. Starting from there, the gdb_id <->
thread id conversion was not needed anymore, since the remote protocol
and gdbserver were using the same kind of ids again. The gdb_id field
in the thread_info structure was also unused starting there.
- d50171e4 Teach linux gdbserver to step-over-breakpoints.
This patch moved the thread_info structure around, and got rid of the
gdb_id field (which was unused).
Looking at the implementation of gdb_id_to_thread_id, it is not doing
anything useful. It is looking up a thread by ptid using
find_thread_ptid, which basically loops over all threads looking at
their entry.id field. If a thread with that ptid is found, it returns
its entry.id field. So it will always return the same thing as it input
(with the exception of if no thread exist with that ptid, then it will
return null_ptid).
gdb/gdbserver/ChangeLog:
* inferiors.h (gdb_id_to_thread_id): Remove.
* inferiors.c (gdb_id_to_thread_id): Remove.
* server.c (process_serial_event): Adjust to gdb_id_to_thread_id
removal. Move pid declaration closer to where it's used.
Simon Marchi [Fri, 15 Sep 2017 15:59:36 +0000 (17:59 +0200)]
Deduplicate require_running macros and move them up
I find it very confusing to define the require_running in the middle of
the file, and re-define it to something else later in the middle of the
same file. I think it would be better if those macros had different
names so that we know exactly what they do.
gdb/gdbserver/ChangeLog:
* server.c (require_running): Rename to ...
(require_running_or_return): ... this ...
(require_running_or_break): ... and this.
(handle_query, process_serial_event): Adjust.
Nick Clifton [Fri, 15 Sep 2017 15:18:20 +0000 (16:18 +0100)]
Add -l option to src-release script.
Since we are updating src-release.sh I thought I would check in this
small patch. It adds the option to create lzip compressed tarballs
via the -l command line option.
* src-release.sh (LZIPPROG): New define. Provides the name of the
lzip program.
(do_lz): New function. Compresses a tarball using the lzip
program.
(do_compress): Add support for lzip compression.
(usage): Mention -l option.
(build_release): Support -l option to invoke lzip compression.
Simon Marchi [Fri, 15 Sep 2017 12:53:07 +0000 (14:53 +0200)]
gdbserver: Remove duplicate functions to find any thread of process
We have about 6 functions/callbacks to find_inferior meant to find a
thread that belongs to a given pid. Remove all but
find_any_thread_of_pid and replace their uses with
find_any_thread_of_pid.
gdb/gdbserver/ChangeLog:
* server.c (first_thread_of): Remove.
(process_serial_event): Replace usage of first_thread_of with
find_any_thread_of_pid.
* tracepoint.c (same_process_p): Remove.
(gdb_agent_about_to_close): Replace usage of same_process_p with
find_any_thread_of_pid.
* linux-x86-low.c (same_process_callback): Remove.
(x86_arch_setup_process_callback): Replace usage of
same_process_callback with find_any_thread_of_pid.
* thread-db.c (any_thread_of): Remove.
(switch_to_process): Replace usage of any_thread_of with
find_any_thread_of_pid.
* inferiors.c (thread_pid_matches_callback): Remove.
(find_thread_process): Adjust to use find_any_thread_of_pid.
Yao Qi [Fri, 15 Sep 2017 12:05:39 +0000 (13:05 +0100)]
Fix x86 build failures
Christophe Lyon told me that GDB build failed on i386-linux with
--enable-64-bit-bfd=yes, so I audit the gdb/configure.tgt again. I
find that i386-darwin has the same issue too. Additionally, GDB
for solaris target fails to build too. This patch fixes all of them.
Alan Modra [Fri, 15 Sep 2017 05:03:58 +0000 (14:33 +0930)]
Enable/disable various dirs in src-release.sh
gold needs to be enabled for a binutils release, but it's rather odd
to do so for gas, gdb or sim. This patch passes various --enable and
--disable options depending on the directories being released.
* src-release.sh (do_proto_toplev): Revert last patch. Enable or
disable binutils, gas, gdb, gold, gprof, ld, libdecnumber, readline,
and sim depending on $tool and $support_files. Echo configure line.
Alan Modra [Fri, 15 Sep 2017 00:52:13 +0000 (10:22 +0930)]
Error when 32-bit ar tries to handle 4G or larger files
We used to silently truncate the size returned by stat() to 32 bits.
While it is possible to make binutils handle a 64-bit off_t on a
32-bit host, to me the effort needed doesn't seem worth the benefit.
Instead, error if we truncate the size. I've written the test the way
I have to avoid a signed/unsigned warning.
PR 22116
* archive.c (bfd_ar_hdr_from_filesystem): Detect when status.st_size
overflows bfd_size_type.
Simon Marchi [Thu, 14 Sep 2017 20:36:57 +0000 (22:36 +0200)]
Make dwarf_expr_context::stack an std::vector
Replace the manually managed array with a vector. It is mostly
straightforward, except maybe one thing in execute_stack_op, in the
handling of DW_OP_fbreg. When the code stumbles on that opcode while
evaluating an expression, it needs to evaluate a subexpression to find
where the fb reg has been saved. Rather than creating a new context, it
reuses the current context. It saves the size of the stack before and
restores the stack to that size after.
I think we can do a little bit better by saving the current stack
locally and installing a new empty stack. This way, if the
subexpression is malformed and underflows, we'll get an exception.
Before, it would have overwritten the top elements of the top-level
expression. The evaluation of the top-level expression would have then
resumed with the same stack size, but possibly some corrupted elements.
H.J. Lu [Thu, 14 Sep 2017 18:41:58 +0000 (11:41 -0700)]
x86: Cache section contents and relocations
bfd/
PR ld/22135
* elf32-i386.c (elf_i386_convert_load_reloc): Add an argument
to indicate if conversion is performed.
(elf_i386_check_relocs): Cache section contents and relocations
if conversion is performed.
* elf64-x86-64.c (elf_x86_64_check_relocs): Cache section
contents and relocations if conversion is performed.
Simon Marchi [Thu, 14 Sep 2017 14:13:33 +0000 (16:13 +0200)]
Make dwarf_stack_value::in_stack_memory a bool
Replace int with bool, because that's what it is.
gdb/ChangeLog:
* dwarf2expr.h (dwarf_expr_piece) <v.mem.in_stack_memory>:
Change type to bool.
(dwarf_stack_value) <in_stack_memory>: Likewise.
(dwarf_expr_context) <push_address>: Change parameter type to
bool.
<fetch_in_stack_memory>: Change return type to bool.
<push>: Change parameter type to bool.
* dwarf2expr.c (dwarf_expr_context::push): Change parameter type
to bool.
(dwarf_expr_context::push_address): Likewise.
(dwarf_expr_context::fetch_in_stack_memory): Change return type
to bool.
(dwarf_expr_context::execute_stack_op): Adjust.
* dwarf2loc.c (dwarf2_evaluate_loc_desc_full): Adjust.
Simon Marchi [Thu, 14 Sep 2017 13:57:01 +0000 (15:57 +0200)]
Make dwarf_expr_piece::pieces an std::vector
Change the manually managed array dwarf_expr_piece::piece with an
std::vector. After passing the pieces array to allocate_piece_closure,
dwarf2_evaluate_loc_desc_full doesn't need that data anymore. We can
therefore move the content of the vector to avoid copying it.
Reg-tested on the buildbot.
gdb/ChangeLog:
* dwarf2expr.h (struct dwarf_expr_piece): Move up.
(struct dwarf_expr_context) <n_pieces>: Remove.
<pieces>: Change type to std::vector.
* dwarf2expr.c (dwarf_expr_context::dwarf_expr_context): Adjust.
(dwarf_expr_context::~dwarf_expr_context): Don't manually free
pieces.
(dwarf_expr_context::add_piece): Adjust.
* dwarf2loc.c (struct piece_closure): Initialize fields.
<n_pieces>: Remove.
<pieces>: Change type to std::vector.
(allocate_piece_closure): Adjust, change parameter to
std::vector rvalue and std::move it to piece_closure.
(rw_pieced_value): Adjust.
(check_pieced_synthetic_pointer): Adjust.
(indirect_synthetic_pointer): Adjust.
(coerce_pieced_ref): Adjust.
(free_pieced_value_closure): Adjust. Use delete to free
piece_closure.
(dwarf2_evaluate_loc_desc_full): Adjust. std::move ctx.pieces
to allocate_piece_closure.
(dwarf2_loc_desc_get_symbol_read_needs): Adjust.
Alan Modra [Thu, 14 Sep 2017 00:49:31 +0000 (10:19 +0930)]
PR22127, as segfaults assembling invalid .reloc
"sec" gets set to NULL on errors in the offset expression. This patch
disables part of the reloc expression processing that needs "sec"
valid. I didn't disable the entire reloc expression handling so that
errors in the reloc expression are reported even when the offset
expression has an error.
PR 22127
* write.c (resolve_reloc_expr_symbols): Don't segfault when
sec has been set to NULL.
Simon Marchi [Tue, 12 Sep 2017 12:15:23 +0000 (14:15 +0200)]
probe: Replace VEC(probe_ops_cp) with std::vector
This patch replaces the usage of VEC to store pointers to probe_ops with
an std::vector. The sole usage of that vector type is one global
variable that holds the ops for the various kinds of probes, so this is
pretty straightforward (no allocation/deallocation issues).
gdb/ChangeLog:
* probe.h (probe_ops_cp): Remove typedef.
(DEF_VEC_P (probe_ops_cp)): Remove.
(all_probe_ops): Change type to std::vector.
* probe.c (info_probes_for_ops): Adjust to vector change.
(probe_linespec_to_ops): Likewise.
(all_probe_ops): Change type to std::vector.
(_initialize_probe): Adjust to vector change.
* dtrace-probe.c (_initialize_dtrace_probe): Likewise.
* elfread.c (elf_get_probes): Likewise.
* stap-probe.c (_initialize_stap_probe): Likewise.
Simon Marchi [Tue, 12 Sep 2017 11:55:32 +0000 (13:55 +0200)]
Make collect_probes return an std::vector
Change collect_probes so it returns an std::vector<bound_probe> instead
of a VEC(bound_probe_s). This allows removing some cleanups. It also
seems like enable_probes_command and disable_probes_command were not
freeing that vector.
The comparison function compare_probes needs to be updated to return a
bool indicating whether the first parameter is "less than" the second
parameter.
I defined two constructors to bound_probe. The default constructor is
needed, for example, so the instance in struct bp_location can be
constructed without parameters. The constructor with parameters is
useful so we can use emplace_back and pass the values directly.
The s390 builder on the buildbot shows a weird failure that I can't
explain:
../../binutils-gdb/gdb/elfread.c: In function void probe_key_free(bfd*, void*):
../../binutils-gdb/gdb/elfread.c:1346:8: error: types may not be defined in a for-range-declaration [-Werror]
for (struct probe *probe : *probes)
^~~~~~
I guess it's a bug with that specific version< of the compiler, since no
other gcc gives me that error. It is using:
Simon Marchi [Tue, 12 Sep 2017 11:37:00 +0000 (13:37 +0200)]
Make probe_ops::get_probes fill an std::vector
This patch changes one usage of VEC to std::vector. It is a relatively
straightforward 1:1 change. The implementations of
sym_probe_fns::sym_get_probes return a borrowed reference to their probe
vectors, meaning that the caller should not free it. In the new code, I
made them return a const reference to the vector.
This patch and the following one were tested by the buildbot. I didn't
see any failures that looked related to this one.
gdb/ChangeLog:
* probe.h (struct probe_ops) <get_probes>: Change parameter from
vec to std::vector.
* probe.c (parse_probes_in_pspace): Update.
(find_probes_in_objfile): Update.
(find_probe_by_pc): Update.
(collect_probes): Update.
(probe_any_get_probes): Update.
* symfile.h (struct sym_probe_fns) <sym_get_probes> Change
return type to reference to std::vector.
* dtrace-probe.c (dtrace_process_dof_probe): Change parameter to
std::vector and update.
(dtrace_process_dof): Likewise.
(dtrace_get_probes): Likewise.
* elfread.c (elf_get_probes): Change return type to std::vector,
store an std::vector in bfd_data.
(probe_key_free): Update to std::vector.
* stap-probe.c (handle_stap_probe): Change parameter to
std::vector and update.
(stap_get_probes): Likewise.
* symfile-debug.c (debug_sym_get_probes): Change return type to
std::vector and update.
Tom Tromey [Sun, 10 Sep 2017 20:48:30 +0000 (14:48 -0600)]
Make extract_arg return a std::string
Change extract_arg to return a std::string and fix up all the users.
I think string is mildly better than unique_xmalloc_ptr<char>, when
possible, because it provides a more robust API.
I changed the error messages emitted from find_location_by_number to
avoid either writing to a string or an extra allocation; this can be
changed but I thought that the new message was not any less clear.
You can see an example in the testsuite patch.
Tom Tromey [Sun, 10 Sep 2017 20:19:19 +0000 (14:19 -0600)]
Rename _const functions to use overloading instead
This renames a few functions -- skip_spaces_const,
skip_to_space_const, get_number_const, extract_arg_const -- to drop
the "_const" suffix and instead rely on overloading.
This makes future const fixes simpler by reducing the number of lines
that must be changed. I think it is also not any less clear, as all
these functions have the same interface as their non-const versions by
design. Furthermore there's an example of using an overload in-tree
already, namely check_for_argument.
This patch was largely created using some perl one-liners; then a few
fixes were applied by hand.
ChangeLog
2017-09-11 Tom Tromey <tom@tromey.com>
* common/common-utils.h (skip_to_space): Remove macro, redeclare
as function.
(skip_to_space): Rename from skip_to_space_const.
* common/common-utils.c (skip_to_space): New function.
(skip_to_space): Rename from skip_to_space_const.
* cli/cli-utils.h (get_number): Rename from get_number_const.
(extract_arg): Rename from extract_arg_const.
* cli/cli-utils.c (get_number): Rename from get_number_const.
(extract_arg): Rename from extract_arg_const.
(number_or_range_parser::get_number): Use ::get_number.
* aarch64-linux-tdep.c, ada-lang.c, arm-linux-tdep.c, ax-gdb.c,
break-catch-throw.c, breakpoint.c, cli/cli-cmds.c, cli/cli-dump.c,
cli/cli-script.c, cli/cli-setshow.c, compile/compile.c,
completer.c, demangle.c, disasm.c, findcmd.c, linespec.c,
linux-tdep.c, linux-thread-db.c, location.c, mi/mi-parse.c,
minsyms.c, nat/linux-procfs.c, printcmd.c, probe.c,
python/py-breakpoint.c, record.c, rust-exp.y, serial.c, stack.c,
stap-probe.c, tid-parse.c, tracepoint.c: Update all callers.
Tom Tromey [Fri, 8 Sep 2017 21:38:12 +0000 (15:38 -0600)]
Make it simpler to add events to Python
The first patch in this series went through several iterations as I'd
forgotten how many places had to be touched to add a new event and a
new event type.
This patch simplifies the process using two new ".def" files. Now, a
new event type can be added by adding a line to "py-event-types.def",
and a new event registry can be added by adding a line to
"py-all-events.def".
ChangeLog
2017-09-11 Tom Tromey <tom@tromey.com>
* python/python.c (do_start_initialization): Use
py-event-types.def to initialize types.
Define all object type structures.
* python/python-internal.h: Don't declare event initialization
functions.
* python/py-threadevent.c (thread_event_object_type): Don't
define.
* python/py-stopevent.c (stop_event_object_type): Don't define.
* python/py-signalevent.c (signal_event_object_type): Don't
declare or define.
* python/py-newobjfileevent.c (new_objfile_event_object_type)
(clear_objfiles_event_object_type): Don't declare or define.
* python/py-infevents.c (inferior_call_pre_event_object_type)
(inferior_call_post_event_object_type)
(register_changed_event_object_type)
(memory_changed_event_object_type): Don't declare or define.
* python/py-inferior.c (new_thread_event_object_type)
(new_inferior_event_object_type)
(inferior_deleted_event_object_type): Don't declare or define.
* python/py-exitedevent.c (exited_event_object_type): Don't
declare or define.
* python/py-evts.c (gdbpy_initialize_py_events): Use
py-all-events.def.
* python/py-events.h (thread_event_object_type): Don't declare.
(events_object): Use py-all-events.def.
* python/py-event.h (GDBPY_NEW_EVENT_TYPE): Remove. Use
py-event-types.def.
* python/py-event-types.def: New file.
* python/py-continueevent.c (create_continue_event_object): Don't
declare or define.
* python/py-bpevent.c (breakpoint_event_object_type): Don't
declare or define.
* python/py-all-events.def: New file.
Tom Tromey [Fri, 8 Sep 2017 20:26:43 +0000 (14:26 -0600)]
Small event ownership clean up in Python layer
It seems cleaner to me for functions like create_thread_event_object,
which pass object ownership to their callers, to directly return a
gdb_ref<>. This way the ownership transfer is part of the API. This
patch makes this change.
Tom Tromey [Tue, 5 Sep 2017 18:07:00 +0000 (12:07 -0600)]
Add new_inferior, inferior_deleted, and new_thread events
This adds a few new events to gdb's Python layer: new_inferior,
inferior_deleted, and new_thread. I wanted to be able to add a
combined inferior/thread display window to my GUI, and I needed a few
events to make this work. This is PR python/15622.
The last commit unfortunately was not enough to fix the build breakage
on AArch64. I made a mistake and did not test it alone on BuildBot,
but along with another patch that was responsible for fixing the
breakage.
The failure is:
In file included from /usr/include/string.h:640:0,
from build-gnulib-gdbserver/import/string.h:41,
from ../../../binutils-gdb/gdb/gdbserver/../common/common-defs.h:56,
from ../../../binutils-gdb/gdb/gdbserver/server.h:22,
from ../../../binutils-gdb/gdb/gdbserver/regcache.c:19:
In function ‘void* memset(void*, int, size_t)’,
inlined from ‘regcache* init_register_cache(regcache*, const target_desc*, unsigned char*)’ at ../../../binutils-gdb/gdb/gdbserver/regcache.c:150:50:
/usr/include/aarch64-linux-gnu/bits/string3.h:81:32: error: call to ‘__warn_memset_zero_len’ declared with attribute warning: memset used with constant zero length parameter; this could be due to transposed parameters [-Werror]
__warn_memset_zero_len ();
^
In function ‘void* memset(void*, int, size_t)’,
inlined from ‘regcache* get_thread_regcache(thread_info*, int)’ at ../../../binutils-gdb/gdb/gdbserver/regcache.c:57:60:
/usr/include/aarch64-linux-gnu/bits/string3.h:81:32: error: call to ‘__warn_memset_zero_len’ declared with attribute warning: memset used with constant zero length parameter; this could be due to transposed parameters [-Werror]
__warn_memset_zero_len ();
This is likely due to a GCC bug, because for some reason the compiler
assumes that the third argument to the memset:
And the underlying problem is that the code is not calling the new
function "allocate_target_description" to allocate the "struct
target_desc" using "new" instead of XNEW, which end up not properly
initializing the fields of the structure.
The problem is that this is undefined if current_ui is NULL, which can
happen early on during gdb start up.
If we run into an error during early gdb start up then we write the
error message to gdb_stderr. However, if we are too early during the
start up then current_ui is NULL, and using the gdb_stderr macro
triggers undefined behaviour.
We try to avoid this using a check 'gdb_stderr == NULL' which was fine
before the recent changes, but now, still triggers undefined behaviour.
A better check is instead 'current_ui == NULL' which is what I use in
this patch.
Triggering this failure is pretty hard, most of the really early errors
are only triggered if pretty basic things are not as expected, for
example, if the default signal handlers are not as expected. Seeing one
of these errors trigger usually means that someone working on gdb has
made an incorrect change. Still, the errors are present in gdb, and
should we ever trigger one it would be nice if gdb didn't crash.
For testing this change I've been applying this patch which adds an
unconditional error into a function called early during gdb start up.
Later in the same function is a real error call which, in some
circumstances could be triggered:
Tom Tromey [Mon, 14 Aug 2017 06:15:33 +0000 (00:15 -0600)]
Use gdb::byte_vector in pascal_object_print_value
This changes pascal_object_print_value to use a gdb::byte_vector.
This removes a cleanup. This change also points out how the previous
code had a possible use-after-free bug.
ChangeLog
2017-09-09 Tom Tromey <tom@tromey.com>
* p-valprint.c (pascal_object_print_value): Use gdb::byte_vector.
Tom Tromey [Thu, 7 Sep 2017 03:41:40 +0000 (21:41 -0600)]
Use ui_out_emit_list and ui_out_emit_tuple with gdb::optional
This changes a few spots to use ui_out_emit_list and/or
ui_out_emit_tuple with gdb::optional, to preserve existing behavior.
This allows for the removal of a few more cleanups.
ChangeLog
2017-09-09 Tom Tromey <tom@tromey.com>
* mi/mi-cmd-var.c (mi_cmd_var_list_children): Use gdb::optional,
ui_out_emit_list, ui_out_emit_tuple.
(mi_cmd_var_update): Likewise.
Tom Tromey [Sat, 9 Sep 2017 15:15:23 +0000 (09:15 -0600)]
Use ui_out_emit_table and ui_out_emit_list in print_thread_info_1
This changes print_thread_info_1 to use ui_out_emit_table and
ui_out_emit_list. Which one is used depends on whether the ui-out is
mi-like; so the emitters are wrapped in gdb::optional.
ChangeLog
2017-09-09 Tom Tromey <tom@tromey.com>
* thread.c (print_thread_info_1): Use ui_out_emit_table,
ui_out_emit_list, gdb::optional.
Alan Modra [Sat, 9 Sep 2017 12:25:22 +0000 (21:55 +0930)]
PowerPC64 --plt-align
This changes the PowerPC64 --plt-align option to perform the usual
alignment of code as suggested by its name, as well as the previous
behaviour of padding so as to reduce boundary crossing. The old
behaviour is had by using a negative parameter.
The default is also changed to align plt stub code by default to 32
byte boundaries, the point being to get better bctr branch prediction
on power8 and power9 hardware.
bfd/
* elf64-ppp.c (plt_stub_pad): Handle positive and negative
plt_stub_align.
ld/
* ld.texinfo (--plt-align): Describe new behaviour of option.
* emultempl/ppc64elf.em (params): Default plt_stub_align to 5.
* testsuite/ld-powerpc/powerpc.exp: Pass --no-plt-align for
selected tests.
* testsuite/ld-powerpc/relbrlt.d: Pass --no-plt-align.
* testsuite/ld-powerpc/elfv2so.d: Adjust expected output.
H.J. Lu [Sat, 9 Sep 2017 14:29:15 +0000 (07:29 -0700)]
x86: Update UNDEFINED_WEAK_RESOLVED_TO_ZERO
Since the only information which SYMBOL_REFERENCES_LOCAL_P doesn't check
is relocations, UNDEFINED_WEAK_RESOLVED_TO_ZERO only needs to check for
relocations with SYMBOL_REFERENCES_LOCAL_P.
H.J. Lu [Sat, 9 Sep 2017 12:31:30 +0000 (05:31 -0700)]
x86: Remove restriction on NOTRACK prefix position
Since the NOTRACK prefix is no longer required to be the last prefix
before the REX prefix, restriction on the NOTRACK prefix position is
removed from assembler as well as disassembler. Assembler encodes the
NOTRACK prefix the same way as the DS segment register, which places
it before other prefixes. Disassembler displays prefixes in the order
they appear.
gas/
* config/tc-i386.c (NOTRACK_PREFIX): Removed.
(REX_PREFIX): Updated.
(MAX_PREFIXES): Likewise.
(parse_insn): Remove restriction on NOTRACK prefix position.
* testsuite/gas/i386/notrack.s: Add tests with NOTRACK prefix
before other prefixes.
* testsuite/gas/i386/x86-64-notrack.s: Likewise.
* testsuite/gas/i386/notrackbad.s: Remove tests with NOTRACK
prefix before other prefixes.
* testsuite/gas/i386/x86-64-notrackbad.s: Likewise.
* testsuite/gas/i386/notrack-intel.d: Updated.
* testsuite/gas/i386/notrack.d: Likewise.
* testsuite/gas/i386/notrackbad.l: Likewise.
* testsuite/gas/i386/x86-64-notrack-intel.d: Likewise.
* testsuite/gas/i386/x86-64-notrack.d: Likewise.
* testsuite/gas/i386/x86-64-notrackbad.l: Likewise.
H.J. Lu [Sat, 9 Sep 2017 12:05:16 +0000 (05:05 -0700)]
x86: Properly handle __ehdr_start
After _bfd_i386_elf_convert_load and _bfd_x86_64_elf_convert_load are
removed, elf_i386_convert_load_reloc and elf_x86_64_convert_load_reloc
see __ehdr_start as an undefined symbol when they are called from
check_relocs to convert GOT relocations against local symbols. But
__ehdr_start will be defined as a hidden symbol by linker at the later
stage if it is referenced. This patch marks __ehdr_start as a defined
local symbol at the start of check_relocs if it is referenced and not
defined.
bfd/
PR ld/22115
* elf32-i386.c (elf_i386_convert_load_reloc): Check linker_def.
Don't use UNDEFINED_WEAK_RESOLVED_TO_ZERO.
* elf64-x86-64.c (elf_x86_64_convert_load_reloc): Check
linker_def. Don't use UNDEFINED_WEAK_RESOLVED_TO_ZERO.
* elfxx-x86.c (_bfd_x86_elf_link_check_relocs): Set local_ref
and linker_def on __ehdr_start if it is referenced and not
defined.
(_bfd_x86_elf_link_symbol_references_local): Also set local_ref
and return TRUE when building executable, if a symbol has
non-GOT/non-PLT relocations in text section or there is no
dynamic linker.
* elfxx-x86.h (elf_x86_link_hash_entry): Add linker_def.
Remove C/C++ relevant code in Fortran specific file.
Remove code relevant for printing C/C++ Integer values in a
Fortran specific file to unify printing of Fortran values.
This does not change the output.
Frank Penczek [Fri, 8 Sep 2017 13:11:47 +0000 (15:11 +0200)]
Fix indentation for printing Fortran types with pointers
Printing the prefix "PTR TO -> (" resp. "REF TO ->(" ignored the active
indentation level. This caused inconsistent appearance of user-defined
Fortran types containing pointers. Fix by using "fprintfi_filtered" with the
current indentation level for outputting the prefix string. Add test case
ptr-indentation.
Example using 'ptype' on object of type:
type TypeWithPointer
integer i
integer, pointer:: p
end type TypeWithPointer
Before:
type = Type typewithpointer
integer(kind=4) :: i
PTR TO -> ( integer(kind=4) :: p)
End Type typewithpointer
After:
type = Type typewithpointer
integer(kind=4) :: i
PTR TO -> ( integer(kind=4) :: p)
End Type typewithpointer
H.J. Lu [Fri, 8 Sep 2017 00:19:10 +0000 (17:19 -0700)]
x86; Don't add elf64-x86-64.lo nor elf64.lo together with elfxx-x86.lo
Don't set r_info and r_sym fields in _bfd_x86_elf_link_hash_table_create.
Instead, set them in _bfd_x86_elf_link_setup_gnu_properties. We can
avoid adding elf64-x86-64.lo and elf64.lo together with elfxx-x86.lo to
bfd_backends.
* configure.ac (bfd_backends): Don't add elf64-x86-64.lo nor
elf64.lo together with elfxx-x86.lo for 64-bit BFD.
* configure: Regenerated.
* elf32-i386.c (elf_i386_link_setup_gnu_properties): Set r_info
and r_sym fields of plt_layout.
* elf64-x86-64.c (elf_x86_64_link_setup_gnu_properties):
Likewise.
* elfxx-x86.c (elf_x86_64_is_reloc_section): Remove BFD64 check.
(_bfd_x86_elf_link_hash_table_create): Likewise. Don't set
r_info nor r_sym fields.
(_bfd_x86_elf_link_setup_gnu_properties): Set r_info and r_sym
fields of htab.
* elfxx-x86.h (elf_x86_plt_layout_table): Add r_info and r_sym.
This ends up manifesting due to the two-byte compressed NOP that's
pessimisticly emitted by the ".align 2", which results in "rvc_boundry"
being 2-byte aligned. frag_align_code() then goes and outputs a 2-byte
NOP (which is invalid in no-RVC mode) to align the code back to a 4-byte
boundry, which can't be relaxed away by the linker as it's not part of
the R_RISCV_RELAX relocation.
The fix is to just always emit the worst case possible alignment into
the output as a single R_RISCV_RELAX, which the linker will then fix up.
With this patch I get the expected code generation
Palmer Dabbelt [Wed, 16 Aug 2017 17:41:56 +0000 (10:41 -0700)]
RISC-V: Support PCREL_* relocations agaist weak undefined symbols
I recently modified our Linux port's base address such the absolute
address 0 is no longer addressable as a 32-bit PC-relative offset.
Since Linux links a weak undefined symbol in an intermediate binary, it
needs to be able to reference absolute address 0.
This patch changes R_RISCV_PCREL_* relocations to absolute relocations
while resolving them in order to allow these symbols to be referenced in
PC-relative programs linked at high addresses. Note that this doesn't
apply to PIC, which also uses PC-relative relocations, just to
position-dependent objects, which we use to allow programs to be linked
at high addresses.
In case some of our embedded users are using R_RISCV_PCREL_* as a hacked
up method of getting position-independent binaries (which can work if
you have very simple programs), we only convert the relocations when the
PC-relative version would overflow.
bfd/ChangeLog:
2017-09-07 Palmer Dabbelt <palmer@dabbelt.com>
* elfnn-riscv.c (riscv_zero_pcrel_hi_reloc): New function.
(riscv_record_pcrel_hi_reloc): Add absolute argument.
(riscv_elf_relocate_section): Call riscv_zero_pcrel_hi_reloc for
R_RISCV_PCREL_HI20 relocs, and pass the result to
riscv_record_pcrel_hi_reloc.
I think the second entry is just a rebase/merge oversight, and it wasn't
meant to be added there, particularly since the 7.11 branch was no longer
active at that time anymore.
This patch just removes the entry.
gdb/ChangeLog:
* NEWS (Changes in GDB 7.11): Remove entry for QStartupWithShell.
H.J. Lu [Thu, 7 Sep 2017 11:03:15 +0000 (04:03 -0700)]
x86: Remove _bfd_{i386,x86_64}_elf_convert_load
Instead of converting GOT relocations when sizing dynamic sections, we
convert GOT relocations during relocation check. Add a field, local_ref,
to elf_x86_link_hash_entry to indicate if symbol references are always
local with a new function to check if symbol references are always local,
which works in check_relocs.
* elf32-i386.c (elf_i386_convert_load_reloc): Add an argument,
r_type_p. Remove the converted argument. Replace
SYMBOL_REFERENCES_LOCAL with SYMBOL_REFERENCES_LOCAL_P. Return
the new relocation type via r_type_p.
(elf_i386_relocate_section): Likewise.
(elf_i386_finish_dynamic_symbol): Likewise.
(need_convert_load): Removed.
(check_relocs_failed): Updated.
(elf_i386_check_relocs): Call elf_i386_convert_load_reloc,
instead of setting need_convert_load.
(_bfd_i386_elf_convert_load): Removed.
* elf64-x86-64.c (need_convert_load): Removed.
(check_relocs_failed): Updated.
(elf_x86_64_convert_load_reloc): Add an argument, r_type_p.
Replace SYMBOL_REFERENCES_LOCAL with SYMBOL_REFERENCES_LOCAL_P.
Return the new relocation type via r_type_p.
(elf_x86_64_check_relocs): Call elf_x86_64_convert_load_reloc,
instead of setting need_convert_load.
(elf_x86_64_check_relocs): Don't check PIC if relocation has
been converted.
(_bfd_x86_64_elf_convert_load): Removed.
(elf_x86_64_relocate_section): Replace SYMBOL_REFERENCES_LOCAL
with SYMBOL_REFERENCES_LOCAL_P.
(elf_x86_64_finish_dynamic_symbol): Likewise.
* elfxx-x86.c (_bfd_x86_elf_link_hash_table_create): Don't
set convert_load.
(_bfd_x86_elf_size_dynamic_sections): Don't call convert_load.
(_bfd_x86_elf_link_symbol_references_local): New function.
* elfxx-x86.h (SYMBOL_REFERENCES_LOCAL_P): New.
(UNDEFINED_WEAK_RESOLVED_TO_ZERO): Replace elf.forced_local with
SYMBOL_REFERENCES_LOCAL_P.
(elf_x86_link_hash_entry): Add local_ref.
(elf_x86_link_hash_table): Remove convert_load.
(_bfd_i386_elf_convert_load): Removed.
(_bfd_x86_64_elf_convert_load): Likewise.
(_bfd_x86_elf_link_symbol_references_local): New.
Tom Tromey [Mon, 4 Sep 2017 03:50:47 +0000 (21:50 -0600)]
Change funcall_chain to be a std::vector
This simplifies the handling of funcall_chain, by changing it to be a
std::vector<int> and then fixing the users. This allows the removal
of a cleanup.
It would be even cleaner to replace this with better logic in the
parsers; but a baby step seemed ok.
gdb/ChangeLog
2017-09-05 Tom Tromey <tom@tromey.com>
* parse.c (funcall_chain): Now a std::vector.
(start_arglist, end_arglist): Simplify.
(free_funcalls): Remove.
(parse_exp_in_context_1): Remove cleanup.
This patch introduces functions to simplify this to:
mangled = dw2_linkage_name (die, cu);
or
attr = dw2_linkage_name_attr (die, cu);
gdb/ChangeLog:
* dwarf2read.c (dw2_linkage_name_attr): New function.
(dw2_linkage_name): New function.
(dwarf2_compute_name, dwarf2_physname, read_call_site_scope)
(guess_full_die_structure_name, dwarf2_name): Use dw2_linkage_name.
(anonymous_struct_prefix, dwarf2_name): Use dw2_linkage_name_attr.
Tests in gdb.arch/thumb2-it.exp call functions defined in assembly
without type debugging information. Since 7022349d5c86bae74b49225515f42d2e221bd368 ("Stop assuming no-debug-info
functions return int") this triggers an error which leads to many tests
to FAIL. This patch cast the call to indicate the return type of the
functions when calling them.
2017-09-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
gdb/testsuite/
* gdb.arch/thumb2-it.exp: Cast call to assembly defined function.
H.J. Lu [Wed, 6 Sep 2017 12:06:35 +0000 (05:06 -0700)]
x86-64: Add R_X86_64_converted_reloc_bit
Add R_X86_64_converted_reloc_bit to relocation type to indicate if a
relocation is converted from a GOTPCREL relocation. Linker now generates
failed to convert GOTPCREL relocation; relink with --no-relax
for all cases, including relocations against local symbols.
bfd/
* elf64-x86-64.c (R_X86_64_converted_reloc_bit): New.
(elf_x86_64_info_to_howto): Get the real relocation type by
masking out R_X86_64_converted_reloc_bit.
(elf_x86_64_check_tls_transition): Get the real relocation type
by masking out R_X86_64_converted_reloc_bit.
(elf_x86_64_convert_load_reloc): Set R_X86_64_converted_reloc_bit
instead of setting converted_reloc.
(elf_x86_64_relocate_section): Check R_X86_64_converted_reloc_bit
instead of converted_reloc. Get the real relocation type by
masking out R_X86_64_converted_reloc_bit.
(elf_x86_64_link_setup_gnu_properties): Verify that the value of
R_X86_64_converted_reloc_bit is valid.
* elfxx-x86.h (converted_reloc): Removed.