When trying to run gdbserver compiled for x86 win32 under wine, I get:
$ wine ./gdbserver/gdbserver.exe --once :1234 ./test
/home/emaisin/src/binutils-gdb/gdb/gdbserver/regcache.c:177: A problem internal to GDBserver has been detected.
regcache* new_register_cache(const target_desc*): Assertion `tdesc->registers_size != 0' failed.
It seems like on that platform, init_target_desc is never called, so
registers_size is never computed.
My first thought was to call init_target_desc somewhere in win32-low.c,
but it turns out that when using win32 on arm, the target description is
already initialized by the generated code. My second thought was to
call it in {i386,amd64}_create_target_description, but those functions
are shared with GDB, and init_target_desc is gdbserver-specific. So I
ended up with the simplest fix, calling it in i386_arch_setup.
Now I hit some other problem:
$ wine ./gdbserver/gdbserver.exe --once :1234 ./test
Killing process(es): 39
No program to debug
Exiting
but still, I think fixing the tdesc issue this is a step forward.
gdb/gdbserver/ChangeLog:
* win32-i386-low.c (i386_arch_setup): Call init_target_desc.
I am getting this warning with clang:
/home/emaisin/src/binutils-gdb/gdb/guile/scm-value.c:439:11: error: variable 'address' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
if (res_val != NULL)
^~~~~~~~~~~~~~~
/home/emaisin/src/binutils-gdb/gdb/guile/scm-value.c:444:32: note: uninitialized use occurs here
if (gdbscm_is_exception (address))
^~~~~~~
/home/emaisin/src/binutils-gdb/gdb/guile/scm-value.c:439:7: note: remove the 'if' if its condition is always true
if (res_val != NULL)
^~~~~~~~~~~~~~~~~~~~
/home/emaisin/src/binutils-gdb/gdb/guile/scm-value.c:427:18: note: initialize the variable 'address' to silence this warning
SCM address;
^
= nullptr
We can get rid of it with a small refactoring. I think it's a bit
cleaner/safer to initialize address with a pessimistic value and assign
it on success. Then there's no chance of using it uninitialized. If I
understand correctly, the NULL check on res_val was to check whether
value_addr threw, and that if value_addr returns without throwing, the
result will never be NULL. If that's true, we can skip the res_val
variable.
Tested by running gdb.guile/*.exp locally.
gdb/ChangeLog:
* guile/scm-value.c (gdbscm_value_address): Initialize address,
get rid of res_val.
I just tried to compile gdb trunk on Solaris 11.4 (formerly 12), and
failed for a couple of reasons:
*
In file included from /usr/include/python2.7/Python.h:128:0,
from /vol/src/gnu/gdb/gdb/dist/gdb/python/python-internal.h:94,
from /vol/src/gnu/gdb/gdb/dist/gdb/python/py-instruction.h:23,
from /vol/src/gnu/gdb/gdb/dist/gdb/python/py-instruction.c:21:
/usr/include/python2.7/ceval.h:67:0: error: ignoring #pragma no_inline [-Werror=unknown-pragmas]
#pragma no_inline(PyEval_EvalFrameEx)
^
New in Solaris 11.4: <python2.7/ceval.h> uses a Studio-only #pragma.
I've disabled the warning in warnings.m4.
*
/vol/src/gnu/gdb/gdb/dist/gdb/ser-pipe.c: In function ‘int pipe_open(serial*, const char*)’:
/vol/src/gnu/gdb/gdb/dist/gdb/ser-pipe.c:77:9: error: ‘pid_t vfork()’ is deprecated (declared at /usr/include/unistd.h:659) [-Werror=deprecated-declarations]
pid = vfork ();
^
/vol/src/gnu/gdb/gdb/dist/gdb/ser-pipe.c:77:16: error: ‘pid_t vfork()’ is deprecated (declared at /usr/include/unistd.h:659) [-Werror=deprecated-declarations]
pid = vfork ();
^
Since Solaris 11, vfork () is marked deprecated in <unistd.h>.
cf. vfork(2):
The vfork() and vforkx() functions are deprecated. Their sole legiti-
mate use as a prelude to an immediate call to a function from the exec
family can be achieved safely by posix_spawn(3C) or posix_spawnp(3C).
Again, I've disabled the warning.
*
/vol/src/gnu/gdb/gdb/dist/gdb/cli/cli-cmds.c: In function ‘void shell_escape(const char*, int)’:
/vol/src/gnu/gdb/gdb/dist/gdb/cli/cli-cmds.c:750:14: error: ‘pid_t vfork()’ is deprecated (declared at /usr/include/unistd.h:659) [-Werror=deprecated-declarations]
if ((pid = vfork ()) == 0)
^
/vol/src/gnu/gdb/gdb/dist/gdb/cli/cli-cmds.c:750:21: error: ‘pid_t vfork()’ is deprecated (declared at /usr/include/unistd.h:659) [-Werror=deprecated-declarations]
if ((pid = vfork ()) == 0)
^
Same problem.
*
/vol/src/gnu/gdb/gdb/dist/gdb/procfs.c: In function ‘void procfs_init_inferior(target_ops*, int)’:
/vol/src/gnu/gdb/gdb/dist/gdb/procfs.c:4380:30: error: ‘START_INFERIOR_TRAPS_EXPECTED’ was not declared in this scope
gdb_startup_inferior (pid, START_INFERIOR_TRAPS_EXPECTED);
^
defined in nat/fork-inferior.h, need to include that header
/vol/src/gnu/gdb/gdb/dist/gdb/procfs.c: In function ‘void procfs_create_inferior(target_ops*, const char*, const string&, char**, int)’:
/vol/src/gnu/gdb/gdb/dist/gdb/procfs.c:4605:38: error: ‘fork_inferior’ was not declared in this scope
NULL, NULL, shell_file, NULL);
^
likewise
/vol/src/gnu/gdb/gdb/dist/gdb/procfs.c: In function ‘void procfs_info_proc(target_ops*, const char*, info_proc_what)’:
/vol/src/gnu/gdb/gdb/dist/gdb/procfs.c:5124:20: error: ‘argv’ was not declared in this scope
for (char *arg : argv)
^
Typo, should be built_argv instead!
*
Undefined first referenced
symbol in file
fork_inferior(char const*, std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, char**, void (*)(), void (*)(int), void (*)(), char const*, void (*)(char const*, char* const*, char* const*)) procfs.o
startup_inferior(int, int, target_waitstatus*, ptid_t*) fork-child.o
ld: fatal: symbol referencing errors
collect2: error: ld returned 1 exit status
make[2]: *** [Makefile:2249: gdb] Error 1
Need to add fork-inferior.o to NATDEPFILES.
With the changes below, I can build gdb on sparcv9-sun-solaris2.11 and
amd64-pc-solaris2.11 and a simple smoke test (gdb/gdb gdb/gdb) works.
gcc doesn't emit stack notes for ELFv1, since ELFv1 never needs an
executable stack. Note that ELFv1 is usually big-endian and ELFv2
little-endian, but the ABI is really orthogonal to endiannes.
* powerpc.cc (Target_powerpc<64,*>::powerpc_info): Set
is_default_stack_executable false.
We have multiple tests that report failure to assemble without saying
exactly what test is failing.
* testsuite/binutils-all/readelf.exp: Don't perror and exit on
bintest.s assembly failure. Report tests unresolved instead.
Likewise for version note test, pr18374, decompress, and dw5
tests.
(readelf_test): Set testname to include both option and binary
file name. Use for pass/fail.
Fix commit e407c74b5b ("Support for MIPS R5900 (Sony Playstation 2)"),
<https://sourceware.org/ml/binutils/2012-12/msg00240.html>, and add the
handling of E_MIPS_MACH_5900, correctly showing `5900' among `Flags:' in
the output of `-h' rather than `unknown CPU'.
binutils/
* readelf.c (get_machine_flags) <E_MIPS_MACH_5900>: New case.
gas/
* testsuite/gas/mips/elf_mach_5900.d: New test.
* testsuite/gas/mips/mips.exp: Run it.
If a .stabs directive was used before another .set directive in a MIPS
source file, s_mips_stab would call mips_mark_labels without having
initialized the mips_opts structure yet. Fix this by calling
file_mips_check_options which will initialize mips_opts if necessary.
gas/
PR gas/21762
* config/tc-mips.c (s_mips_stab): Insert call to
file_mips_check_options.
* testsuite/gas/mips/micromips@stabs-symbol-type.d: New test.
* testsuite/gas/mips/mips.exp: Run the new tests.
* testsuite/gas/mips/mips16@stabs-symbol-type.d: New test.
* testsuite/gas/mips/stabs-symbol-type.d: New test.
* testsuite/gas/mips/stabs-symbol-type.s: New test source.
As the title says, this is a test case for
Inferior.thread_from_thread_handle, a python method which will,
given a thread library dependent thread handle, find the GDB thread
which corresponds to that thread handle (in the inferior under
consideration).
The C file for this test case causes the thread handles for the
main thread and two child threads to be placed into an array. The
test case runs to one of the functions (do_something()) at which point,
it retrieves the thread handles from the array and attempts to find the
corresponding thread in GDB's internal thread list.
I use barriers to make sure that both threads have actually started;
execution will stop when one of the threads breaks at do_something.
Thanks to Simon Marchi for suggestions for forcing the thread
numbering to be stable.
gdb/testsuite/ChangeLog:
* gdb.python/py-thrhandle.c, gdb.python/py-thrhandle.exp: New
files.
This patch adds support to remote targets for converting a thread
handle to a thread_info struct pointer.
A thread handle is fetched via a "handle" attribute which has been
added to the qXfer:threads:read query packet. An implementation is
provided in gdbserver for targets using the Linux kernel.
gdb/gdbserver/ChangeLog:
* linux-low.h (struct lwp_info): Add new field, thread_handle.
(thread_db_thread_handle): Declare.
* linux-low.c (linux_target_ops): Initialize thread_handle.
* server.c (handle_qxfer_threads_worker): Add support for
"handle" attribute.
* target.h (struct target_ops): Add new function pointer,
thread_handle.
(target_thread_handle): Define.
* thread-db.c (find_one_thread, attach_thread): Set thread_handle
field in lwp.
(thread_db_thread_handle): New function.
gdb/ChangeLog:
* remote.c (vector): Include.
(struct private_thread_info): Add field, thread_handle.
(free_private_thread_info): Deallocate storage associated with
thread handle.
(get_private_info_thread): Initialize `thread_handle' field.
(struct thread_item): Add field, thread_handle.
(clear_threads_listing_context): Deallocate storage associated
with thread handle.
(start_thread): Add support for "handle" attribute.
(thread_attributes): Add "handle".
(remote_get_threads_with_qthreadinfo): Initialize thread_handle
field.
(remote_update_thread_list): Update thread_handle.
(remote_thread_handle_to_thread_info): New function.
(init_remote_ops): Initialize to_thread_handle_to_thread_info.
While working on a patch for fetching a thread handle in gdbserver, I
ran into a circumstance in which tests in gdb.mi/mi-nsmoribund.exp
would occasionally fail. Over a large enough number of runs, it would
fail roughly 2% of the time.
That thread handle patch caused find_one_thread() to be called on
every stop. find_one_thread() calls td_ta_map_lwp2thr() which, in
turn, can cause ps_get_thread_area() to be called.
ps_get_thread_area() makes a call to ptrace() for getting the thread
area address. If this should happen when the thread is not stopped,
the call to ptrace will return error which in turn propogates back to
find_one_thread(). find_one_thread() calls error() in this instance
which causes the program to die.
This patch causes find_one_thread() to be called upon reciept of a
clone event. Since the clone is stopped, the circumstances described
above cannot occur.
gdb/gdbserver/ChangeLog:
* linux-low.c (handle_extended_wait): Call thread_db_notice_clone().
* linux-low.h (thread_db_notice_clone): Declare.
* thread-db.c (thread_db_notice_clone): New function.
This patch adds a target method named `to_thread_handle_to_thread_info'.
It is intended to map a thread library specific thread handle (such as
pthread_t for the pthread library) to the corresponding GDB internal
thread_info struct (pointer).
An implementation is provided for Linux pthreads; see linux-thread-db.c.
gdb/ChangeLog:
* target.h (struct target_ops): Add to_thread_handle_to_thread_info.
(target_thread_handle_to_thread_info): Declare.
* target.c (target_thread_handle_to_thread_info): New function.
* target-delegates.c: Regenerate.
* gdbthread.h (find_thread_by_handle): Declare.
* thread.c (find_thread_by_handle): New function.
* linux-thread-db.c (thread_db_thread_handle_to_thread_info): New
function.
(init_thread_db_ops): Register thread_db_thread_handle_to_thread_info.
Back in commit f0db101d98 ("gdbserver: don't pick a random thread if
the current thread dies"), a couple years ago, the last references to
set_desired_thread(0) [select the Hc thread] were removed, and all the
remaining calls to set_desired_thread pass '1', meaning general
thread. This means we can simplify set_desired_thread.
gdb/gdbserver/ChangeLog:
2017-09-21 Pedro Alves <palves@redhat.com>
* server.c (gdb_read_memory, handle_status, process_serial_event)
(handle_serial_event, handle_target_event): Adjust to
set_desired_thread prototype change.
* target.c (set_desired_thread): Remove 'use_general' parameter
and adjust.
* target.h (set_desired_thread): Remove 'use_general' parameter.
Fix two typos that resulted in swapping the BFD names for the core note
register sections NT_S390_GS_CB and NT_S390_GS_BC.
bfd/ChangeLog:
* elf.c (elfcore_grok_note): For the cases NT_S390_GS_CB and
NT_S390_GS_BC, correct the previously swapped invocations of
elfcore_grok_s390_gs_bc and elfcore_grok_s390_gs_cb.
PowerPC64 .cfi directives use DW_EH_PE_sdata4 encoding for .eh_frame,
so there is no real reason why .eh_frame should be 8 byte aligned.
gas/
* config/tc-ppc.h (EH_FRAME_ALIGNMENT): Define.
ld/
* testsuite/ld-powerpc/tlsopt5.wf: Update for reduced alignment.
Clang gives this warning:
/home/emaisin/src/binutils-gdb/gdb/gdbserver/../nat/linux-waitpid.c:45:25: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
vfprintf (stderr, format, args);
^~~~~~
Get rid of it by adding ATTRIBUTE_PRINTF.
gdb/ChangeLog:
* nat/linux-waitpid.c (linux_debug): Add ATTRIBUTE_PRINTF.
I am getting this warning with clang:
/home/emaisin/src/binutils-gdb/gdb/microblaze-tdep.c:94:28: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
vprintf_unfiltered (fmt, args);
^~~
Adding ATTRIBUTE_PRINTF to microblaze_debug gets rid of it. Strangely,
gcc doesn't warn about non-literal format strings when calling vprintf
(or a vprintf-style function, like vprintf_unfiltered). I filed this
gcc bug:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82206
gdb/ChangeLog:
* microblaze-tdep.c (microblaze_debug): Add ATTRIBUTE_PRINTF.
I happen to see that fbsd-tdep.o is missing for target aarch64-freebsd,
and it causes the build failure,
aarch64-fbsd-tdep.o: In function `aarch64_fbsd_init_abi(gdbarch_info, gdbarch*)':
binutils-gdb/gdb/aarch64-fbsd-tdep.c:186: undefined reference to `fbsd_init_abi(gdbarch_info, gdbarch*)'
binutils-gdb/gdb/aarch64-fbsd-tdep.c:189: undefined reference to `svr4_lp64_fetch_link_map_offsets()'
binutils-gdb/gdb/aarch64-fbsd-tdep.c:189: undefined reference to `set_solib_svr4_fetch_link_map_offsets(gdbarch*, link_map_offsets* (*)())'
This patch fixed it.
gdb:
2017-09-21 Yao Qi <yao.qi@linaro.org>
* configure.tgt (aarch64*-*-freebsd*): Add fbsd-tdep.o solib-svr4.o
to gdb_target_obs.
This patch fixes the build failure by using disassembler to get
disassemble function pointer, and do the disassembly, because
print_insn_little_arm is no longer visible outside opcodes/
binutils-gdb/sim/arm/wrapper.c:98:10: error: implicit declaration of function 'print_insn_little_arm' [-Werror=implicit-function-declaration]
size = print_insn_little_arm (0, & info);
^
sim/arm:
2017-09-21 Yao Qi <yao.qi@linaro.org>
* wrapper.c (print_insn): Use disassembler instead of
print_insn_little_arm.
This changes counted_command_line to be a typedef for std::shared_ptr
and removes the associated cleanups. In the long run I believe that
cmd_list_element should also be changed to use a shared_ptr.
gdb/ChangeLog
2017-09-20 Tom Tromey <tom@tromey.com>
* breakpoint.c (struct counted_command_line): Remove.
(breakpoint_commands): Update.
(alloc_counted_command_line, incref_counted_command_line)
(decref_counted_command_line, do_cleanup_counted_command_line)
(make_cleanup_decref_counted_command_line): Remove.
(breakpoint_set_commands, commands_command_1, ~bpstats, bpstats)
(bpstat_clear_actions, bpstat_do_actions_1, watchpoint_check)
(bpstat_stop_status, print_one_breakpoint_location, ~breakpoint)
(save_breakpoints): Update.
* breakpoint.h (counted_command_line): Now a typedef to
shared_ptr.
(struct breakpoint) <commands>: Now a counted_command_line.
(struct bpstats) <command>: Likewise.
This changes iterate_over_related_breakpoints and
map_breakpoint_numbers to take a function_view. Then, it simplifies
the callers by using lambdas. This then allows the removal of some
bookkeeping types.
gdb/ChangeLog
2017-09-20 Tom Tromey <tom@tromey.com>
* breakpoint.c (struct commands_info, do_map_commands_command):
Remove.
(commands_command_1): Update.
(iterate_over_related_breakpoints): Take a function_view.
(do_delete_breakpoint, do_map_delete_breakpoint): Remove.
(delete_command): Update.
(map_breakpoint_numbers): Take a function_view.
(do_disable_breakpoint, do_map_delete_breakpoint): Remove.
(disable_command): Update.
(do_enable_breakpoint, do_map_enable_breakpoint): Remove.
(enable_command): Update.
(struct disp_data, do_enable_breakpoint_disp)
(do_map_enable_once_breakpoint, do_map_enable_count_breakpoint)
(do_map_enable_delete_breakpoint): Remove.
(enable_once_command, enable_count_command, enable_delete_command)
(delete_trace_variable_command): Update.
This changes struct bpstats to be allocated with new and freed with
delete, adding constructors and a destructor in the process. This
allows the removal of one cleanup and clears the way for more to
follow.
gdb/ChangeLog
2017-09-20 Tom Tromey <tom@tromey.com>
* breakpoint.c (~bpstats): Rename from bpstat_free. Update.
(bpstat_clear): Use delete.
(bpstats): New constructors.
(bpstat_copy, bpstat_stop_status): Use new.
(dprintf_after_condition_true): Update.
* breakpoint.h (bpstats::bpstats): Add constructors.
(bpstats::~bpstats): Add destructor.
While working on the no-debug-info debugging improvements series, I
noticed these bare xfree calls, which lead to leaks if
evaluate_subexp_standard throws.
Fix that by reworking make_params as a RAII class. Ends up
eliminating a couple heap allocations too.
gdb/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* eval.c (make_params): Delete, refactored as ...
(class fake_method): ... this new type's ctor.
(fake_method::~fake_method): New.
(evaluate_subexp_standard): Use 'fake_method'.
2017-09-20 Teresa Johnson <tejohnson@google.com>
* plugin.cc (is_visible_from_outside): Check for export dynamic symbol
option and list.
* testsuite/Makefile.am (plugin_test_12): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/export_dynamic_plugin.cc: New test source.
* testsuite/plugin_test_12.sh: New test script.
This changes catch_command_errors_const to be an overload of
catch_command_errors, which may mildly help future constification
efforts.
Tested by rebuilding.
gdb/ChangeLog
2017-09-20 Tom Tromey <tom@tromey.com>
* main.c (catch_command_errors): Rename from
catch_command_errors_const.
(captured_main_1): Update.
Currently, with an ambiguous "list first,last", we get:
(gdb) list bar,main
Specified first line 'bar' is ambiguous:
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98
This commit makes gdb's output above a bit clearer by printing the
symbol name as well:
(gdb) list bar,main
Specified first line 'bar' is ambiguous:
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97, symbol: "bar(A)"
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98, symbol: "bar(B)"
And while at it, makes gdb print the symbol name when actually listing
multiple locations too. I.e., before (with "set listsize 2"):
(gdb) list bar
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97
96
97 int bar (A) { return 11; }
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98
97 int bar (A) { return 11; }
98 int bar (B) { return 22; }
After:
(gdb) list bar
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97, symbol: "bar(A)"
96
97 int bar (A) { return 11; }
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98, symbol: "bar(B)"
97 int bar (A) { return 11; }
98 int bar (B) { return 22; }
Currently, the result of decoding a linespec loses information about
the original symbol that was found. All we end up with is an address.
This makes it difficult to find the original symbol again to get at
its print name. Fix that by storing a pointer to the symbol in the
sal. We already store the symtab and obj_section, so it feels like a
natural progression to me. This avoids having to do any extra symbol
lookup too.
gdb/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* cli/cli-cmds.c (list_command): Use print_sal_location.
(print_sal_location): New function.
(ambiguous_line_spec): Use print_sal_location.
* linespec.c (symbol_to_sal): Record the symbol in the sal.
* symtab.c (find_function_start_sal): Likewise.
* symtab.h (symtab_and_line::symbol): New field.
gdb/testsuite/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* gdb.base/list-ambiguous.exp (test_list_ambiguous_symbol): Expect
symbol names in gdb's output.
* gdb.cp/overload.exp ("list all overloads"): Likewise.
The "list" command allows specifying the name of variables as
argument, not just functions, so that users can type "list
a_global_variable".
That support is a broken when it comes to ambiguous locations though.
If there's more than one such global variable in the program, e.g.,
static globals in different compilation units, GDB ends up listing the
source of the first variable it finds, only.
linespec.c does find both symbol and minsym locations for all the
globals. The problem is that it ends up merging all the resulting
sals into one, because they all have address, zero. I.e., all sals
end up with sal.pc == 0, so maybe_add_address returns false for all
but the first.
The zero addresses appear because:
- in the minsyms case, linespec.c:minsym_found incorrectly treats all
minsyms as if they were function/text symbols. In list mode we can
end up with data symbols there, and we shouldn't be using
find_pc_sect_line on data symbols.
- in the debug symbols case, symbol_to_sal misses recording an address
(sal.pc) for non-block, non-label symbols.
gdb/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* linespec.c (minsym_found): Handle non-text minsyms.
(symbol_to_sal): Record a sal.pc for non-block, non-label symbols.
gdb/testsuite/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* gdb.base/list-ambiguous.exp (test_list_ambiguous_function):
Rename to ...
(test_list_ambiguous_symbol): ... this and add a symbol name
parameter. Adjust.
(test_list_ambiguous_function): Reimplement on top of
test_list_ambiguous_symbol and also test listing ambiguous
variables.
* gdb.base/list-ambiguous0.c (ambiguous): Rename to ...
(ambiguous_fun): ... this.
(ambiguous_var): New.
* gdb.base/list-ambiguous1.c (ambiguous): Rename to ...
(ambiguous_fun): ... this.
(ambiguous_var): New.
In an old commit the backslash of was missing in the rule for creating the
i386-avx-mpx-avx512-pku.dat file. No need to regenerate the files, this
was done by another commit from Yao.
2017-09-20 Walfred Tedeschi <walfred.tedeschi@intel.com>
* features/Makefile (i386-avx-mpx-avx512-pku.dat): Add backslash.
Test was running on a fault during code execution. Analysis have shown
that the wrong instruction had been used. An instruction that takes
not alligned memory is more appropriated for the task.
ChangeLog:
2017-09-20 Walfred Tedeschi <walfred.tedeschi@intel.com>
gdb/testesuite/ChangeLog:
* gdb.arch/i386-avx512.c (move_zmm_data_to_reg): Use
vmovups instead vmovaps.
(move_zmm_data_to_memory): Use vmovups instead vmovaps.
Change-Id: I4a95560861ef1792ed6ce86578fdd726162863f1
ppc32, like many targets, defines the address of a function as the PLT
call stub code for functions referenced but not defined in a non-PIC
executable. ppc32 gold, unlike other targets, inherits the ppc64
multiple stub capability for dealing with very large binaries where
one set of stubs can't be reached from all code locations. This means
there can be multiple choices of address for a function, which might
cause function pointer comparison failures. So for ppc32, make
non-branch references always use the first stub group.
(PowerPC64 ELFv1 is always PIC so doesn't need to define the address
of an external function as the PLT stub. PowerPC64 ELFv2 needs a
special set of global entry stubs to serve as the address of external
functions, so it too is not affected by this bug.)
* powerpc.cc (Target_powerpc::Branch_info::make_stub): Put
stubs for ppc32 non-branch relocs in first stub table.
(Target_powerpc::Relocate::relocate): Resolve similarly.
This works like 'start' but it stops at the first instruction rather
than the first line in main(). This is useful if one wants to single
step through runtime linker startup.
While here, introduce a RUN_ARGS_HELP macro for shared help text
between run, start, and starti. This includes expanding the help for
start and starti to include details from run's help text.
gdb/ChangeLog:
* NEWS (Changes since GDB 8.0): Add starti.
* infcmd.c (enum run_break): New.
(run_command_1): Queue pending event for RUN_STOP_AT_FIRST_INSN
case.
(run_command): Use enum run_how.
(start_command): Likewise.
(starti_command): New function.
(RUN_ARGS_HELP): New macro.
(_initialize_infcmd): Use RUN_ARGS_HELP for run and start
commands. Add starti command.
gdb/doc/ChangeLog:
* gdb.texinfo (Starting your Program): Add description of
starti command. Mention starti command as an alternative for
debugging the elaboration phase.
gdb/testsuite/ChangeLog:
* gdb.base/starti.c: New file.
* gdb.base/starti.exp: New file.
* lib/gdb.exp (gdb_starti_cmd): New procedure.
Complement commit 26eebcf553 ("Update OpenBSD/mips64 support"),
<https://sourceware.org/ml/binutils/2005-04/msg00382.html>, which added
OpenBSD/mips64 support to GAS, and also add it to LD, avoiding a build
failure at the configuration stage, like:
*** ld does not support target mips64-unknown-openbsd
*** see ld/configure.tgt for supported targets
make[1]: *** [configure-ld] Error 1
As per OS support only include n64 MIPS emulations, and use the
traditional ones, matching the choice already made with the addition of
GAS support.
ld/
* configure.tgt <mips64el-*-openbsd*, mips64-*-openbsd*>: New
targets.
gdb/monitor.c was removed by 40e0b27 (Delete the remaining ROM monitor
targets).
gdb:
2017-09-19 Yao Qi <yao.qi@linaro.org>
* Makefile.in (monitor.o): Remove the rule.
After the PR 21411 fix, the linker generated .eh_frame for ppc64 glink
can be edited by the generic code. The sequence of events goes
something like:
1) Some object file adds .eh_frame aligned to 8, making the output
.eh_frame aligned to at least 8, so linker generated .eh_frame FDE
is padded to an 8 byte boundary.
2) All .eh_frame past the glink .eh_frame is garbage collected.
3) Generic code detects that last FDE (the glink .eh_frame) doesn't
need to be padded to an 8 byte boundary, reducing size from 88 to
84.
4) elf64-ppc.c check fails.
PR 21441
* elf64-ppc.c (ppc64_elf_build_stubs): Don't check glink_eh_frame
size.
elf_gc_sweep_symbol should run after verdefs are calculated, since
the verdef code creates symbols for the versions. However,
elf_gc_sweep_symbol needs to run before verrefs so as to not emit
useless verrefs for symbols that are gc'd.
I've also removed a _bfd_elf_link_renumber_dynsyms calls added by
Maciej after I fussed about it when reviewing. On further examination
the call appears to be unnecessary. Looking at renumber_dynsyms also
made me realize that the test to exclude .gnu.version has been wrong
since 2016-04-26 (git commit d5486c4372), so fix that too.
PR 22150
* elflink.c (bfd_elf_size_dynamic_sections): Garbage collect
symbols before calculating verrefs. Don't renumber dynsyms
after gc. Exclude .gnu.version when zero or one dynsym.
Localize some vars and reindent.
The trouble with stubs per output section is that ppc32 uses a plt
stub as the address of a global function. This needs to be unique,
otherwise we'll get multiple addresses for a function.
Obviously this is only a partial solution, since ppc32 will get
multiple stubs when code is larger than 33M. A proper fix will
involve selecting a unique stub to use for non-branch relocs.
* options.h (stub-group-multi): Default to true. Add
--no-stub-group-multi.
Simon Marchi
H.J. Lu
Pedro Alves
Rainer Orth
Alan Modra
GDB Administrator
Maciej W. Rozycki
James Cowgill
Kevin Buettner
Andreas Arnez
Yao Qi
Nick Clifton
Tom Tromey
Teresa Johnson
Walfred Tedeschi
John Baldwin