Compilers can materialize renamings of arrays (or of accesses to arrays)
in Ada into variables whose types are references to the actual array
types. Before this change, trying to use such an array renaming yielded
an error in GDB:
(gdb) print my_array(1)
cannot subscript or call a record
(gdb) print my_array_ptr(1)
cannot subscript or call something of type `(null)'
This behavior comes from bad handling for array renamings, in particular
the OP_FUNCALL expression operator handling from ada-lang.c
(ada_evaluate_subexp): in one place we turn the reference into a
pointer, but the code that follows expect the value to be an array.
This patch fixes how we handle references in call/subscript evaluation
so that we turn these references into the actual array values instead of
pointers to them.
gdb/ChangeLog:
* ada-lang.c (ada_evaluate_subexp) <OP_FUNCALL>: When the input
value is a reference, actually dereference it in order to get
the underlying value.
gdb/testsuite/ChangeLog:
* gdb.ada/array_ptr_renaming.exp: New testcase.
* gdb.ada/array_ptr_renaming/foo.adb: New file.
* gdb.ada/array_ptr_renaming/pack.ads: New file.
Tested on x86_64-linux, no regression.
ret->args_u.text is const char *, probe_args is const char *, so no cast
is needed. Found while doing cxx-conversion stuff, since it wouldn't
build in C++.
gdb/ChangeLog:
* stap-probe.c (handle_stap_probe): Remove unnecessary cast.
Two missing consts, found while doing cxx-conversion work. We end up
with a char*, even though we pass a const char* to strstr. I am pushing
this as obvious.
gdb/ChangeLog:
* cli/cli-setshow.c (cmd_show_list): Constify a variable.
* linespec.c (linespec_lexer_lex_string): Same.
We will need to decode both ADR and ADRP instructions in GDBserver.
This patch makes common code handle both cases, even if GDB only needs
to decode the ADRP instruction.
gdb/ChangeLog:
* aarch64-tdep.c (aarch64_analyze_prologue): New is_adrp
variable. Call aarch64_decode_adr instead of
aarch64_decode_adrp.
* arch/aarch64-insn.h (aarch64_decode_adrp): Delete.
(aarch64_decode_adr): New function declaration.
* arch/aarch64-insn.c (aarch64_decode_adrp): Delete.
(aarch64_decode_adr): New function, factored out from
aarch64_decode_adrp to decode both adr and adrp instructions.
This patch moves the following functions into the arch/ common
directory, in new files arch/aarch64-insn.{h,c}. They are prefixed with
'aarch64_':
- aarch64_decode_adrp
- aarch64_decode_b
- aarch64_decode_cb
- aarch64_decode_tb
We will need them to implement fast tracepoints in GDBserver.
For consistency, this patch also adds the 'aarch64_' prefix to static
decoding functions that do not need to be shared right now.
V2:
make sure the formatting issues propagated
fix `gdbserver/configure.srv'.
gdb/ChangeLog:
* Makefile.in (ALL_64_TARGET_OBS): Add aarch64-insn.o.
(HFILES_NO_SRCDIR): Add arch/aarch64-insn.h.
(aarch64-insn.o): New rule.
* configure.tgt (aarch64*-*-elf): Add aarch64-insn.o.
(aarch64*-*-linux*): Likewise.
* arch/aarch64-insn.c: New file.
* arch/aarch64-insn.h: New file.
* aarch64-tdep.c: Include arch/aarch64-insn.h.
(aarch64_debug): Move to arch/aarch64-insn.c. Declare in
arch/aarch64-insn.h.
(decode_add_sub_imm): Rename to ...
(aarch64_decode_add_sub_imm): ... this.
(decode_adrp): Rename to ...
(aarch64_decode_adrp): ... this. Move to arch/aarch64-insn.c.
Declare in arch/aarch64-insn.h.
(decode_b): Rename to ...
(aarch64_decode_b): ... this. Move to arch/aarch64-insn.c.
Declare in arch/aarch64-insn.h.
(decode_bcond): Rename to ...
(aarch64_decode_bcond): ... this. Move to arch/aarch64-insn.c.
Declare in arch/aarch64-insn.h.
(decode_br): Rename to ...
(aarch64_decode_br): ... this.
(decode_cb): Rename to ...
(aarch64_decode_cb): ... this. Move to arch/aarch64-insn.c.
Declare in arch/aarch64-insn.h.
(decode_eret): Rename to ...
(aarch64_decode_eret): ... this.
(decode_movz): Rename to ...
(aarch64_decode_movz): ... this.
(decode_orr_shifted_register_x): Rename to ...
(aarch64_decode_orr_shifted_register_x): ... this.
(decode_ret): Rename to ...
(aarch64_decode_ret): ... this.
(decode_stp_offset): Rename to ...
(aarch64_decode_stp_offset): ... this.
(decode_stp_offset_wb): Rename to ...
(aarch64_decode_stp_offset_wb): ... this.
(decode_stur): Rename to ...
(aarch64_decode_stur): ... this.
(decode_tb): Rename to ...
(aarch64_decode_tb): ... this. Move to arch/aarch64-insn.c.
Declare in arch/aarch64-insn.h.
(aarch64_analyze_prologue): Adjust calls to renamed functions.
gdb/gdbserver/ChangeLog:
* Makefile.in (aarch64-insn.o): New rule.
* configure.srv (aarch64*-*-linux*): Add aarch64-insn.o.
We have noticed that GDB would sometimes crash trying to print
from a nested function the value of a variable declared in an
enclosing scope. This appears to be target dependent, although
that correlation might only be fortuitious. We noticed the issue
on x86_64-darwin, x86-vxworks6 and x86-solaris. The investigation
was done on Darwin.
This is a new feature that was introduced by:
commit 63e43d3aed
Date: Thu Feb 5 17:00:06 2015 +0100
DWARF: handle non-local references in nested functions
We can reproduce the problem with one of the testcases that was
added with the patch (gdb.base/nested-subp1.exp), where we have...
18 int
19 foo (int i1)
20 {
21 int
22 nested (int i2)
23 {
[...]
27 return i1 * i2; /* STOP */
28 }
... After building the example program, and running until line 27,
try printing the value of "i1":
% gdb gdb.base/nested-subp1
(gdb) break foo.c:27
(gdb) run
Breakpoint 1, nested (i2=2) at /[...]/nested-subp1.c:27
27 return i1 * i2; /* STOP */
(gdb) p i1
[1] 73090 segmentation fault ../gdb -q gdb.base/nested-subp1
Ooops!
What happens is that, because the reference is non-local, we are trying
to follow the function's static link, which does...
/* If we don't know how to compute FRAME's base address, don't give up:
maybe the frame we are looking for is upper in the stace frame. */
if (framefunc != NULL
&& SYMBOL_BLOCK_OPS (framefunc)->get_frame_base != NULL
&& (SYMBOL_BLOCK_OPS (framefunc)->get_frame_base (framefunc, frame)
== upper_frame_base))
... or, in other words, calls the get_frame_base "method" of
framefunc's struct symbol_block_ops data. This resolves to
the block_op_get_frame_base function.
Looking at the function's implementation, we see:
struct dwarf2_locexpr_baton *dlbaton;
[...]
dlbaton = SYMBOL_LOCATION_BATON (framefunc);
[...]
result = dwarf2_evaluate_loc_desc (type, frame, start, length,
dlbaton->per_cu);
^^^^^^^^^^^^^^^
Printing dlbaton->per_cu gives a value that seems fairly bogus for
a memory address (0x60). Because of it, dwarf2_evaluate_loc_desc
then crashes trying to dereference it.
What's different on Darwin compared to Linux is that the function's
frame base is encoded using the following form:
.byte 0x40 # uleb128 0x40; (DW_AT_frame_base)
.byte 0x6 # uleb128 0x6; (DW_FORM_data4)
... and so dwarf2_symbol_mark_computed ends up creating
a SYMBOL_LOCATION_BATON as a struct dwarf2_loclist_baton:
if (attr_form_is_section_offset (attr)
/* .debug_loc{,.dwo} may not exist at all, or the offset may be outside
the section. If so, fall through to the complaint in the
other branch. */
&& DW_UNSND (attr) < dwarf2_section_size (objfile, section))
{
struct dwarf2_loclist_baton *baton;
[...]
SYMBOL_LOCATION_BATON (sym) = baton;
However, if you look more closely at block_op_get_frame_base's
implementation, you'll notice that the function extracts the
symbol's SYMBOL_LOCATION_BATON as a dwarf2_locexpr_baton
(a DWARF _expression_ rather than a _location list_).
That's why we end up decoding the DLBATON improperly, and thus
pass a random dlbaton->per_cu when calling dwarf2_evaluate_loc_desc.
This works on x86_64-linux, because we indeed have the frame base
described using a different form:
.uleb128 0x40 # (DW_AT_frame_base)
.uleb128 0x18 # (DW_FORM_exprloc)
This patch fixes the issue by doing what we do for most (if not all)
other such methods: providing one implementation each for loc-list,
and loc-expr. Both implementations are nearly identical, so perhaps
we might later want to improve this. But this patch first tries to
fix the crash first, leaving the design issue for later.
gdb/ChangeLog:
* dwarf2loc.c (locexpr_get_frame_base): Renames
block_op_get_frame_base.
(dwarf2_block_frame_base_locexpr_funcs): Replace reference to
block_op_get_frame_base by reference to locexpr_get_frame_base.
(loclist_get_frame_base): New function, near identical copy of
locexpr_get_frame_base.
(dwarf2_block_frame_base_loclist_funcs): Replace reference to
block_op_get_frame_base by reference to loclist_get_frame_base.
Tested on x86_64-darwin (AdaCore testsuite), and x86_64-linux
(official testsuite).
bfd/ChangeLog:
* targets.c (enum bfd_flavour): Add comment.
(bfd_flavour_name): New function.
* bfd-in2.h: Regenerate.
gdb/ChangeLog:
* findvar.c (default_read_var_value) <LOC_UNRESOLVED>: Include the
kind of minimal symbol in the error message.
* objfiles.c (objfile_flavour_name): New function.
* objfiles.h (objfile_flavour_name): Declare.
gdb/testsuite/ChangeLog:
* gdb.dwarf2/dw2-bad-unresolved.c: New file.
* gdb.dwarf2/dw2-bad-unresolved.exp: New file.
With the kernle fix <http://lists.infradead.org/pipermail/linux-arm-kernel/2015-July/356511.html>,
aarch64 GDB is able to read the base of thread area of 32-bit arm
program through NT_ARM_TLS.
This patch is to teach both GDB and GDBserver to read the base of
thread area correctly in the multi-arch case. A new function
aarch64_ps_get_thread_area is added, and is shared between GDB and
GDBserver.
With this patch applied, the following fails in multi-arch testing
(GDB is aarch64 but the test cases are arm) are fixed,
-FAIL: gdb.threads/tls-nodebug.exp: thread local storage
-FAIL: gdb.threads/tls-shared.exp: print thread local storage variable
-FAIL: gdb.threads/tls-so_extern.exp: print thread local storage variable
-FAIL: gdb.threads/tls-var.exp: print tls_var
-FAIL: gdb.threads/tls.exp: first thread local storage
-FAIL: gdb.threads/tls.exp: first another thread local storage
-FAIL: gdb.threads/tls.exp: p a_thread_local
-FAIL: gdb.threads/tls.exp: p file2_thread_local
-FAIL: gdb.threads/tls.exp: p a_thread_local second time
gdb:
2015-09-18 Yao Qi <yao.qi@linaro.org>
* nat/aarch64-linux.c: Include elf/common.h,
nat/gdb_ptrace.h, asm/ptrace.h and sys/uio.h.
(aarch64_ps_get_thread_area): New function.
* nat/aarch64-linux.h: Include gdb_proc_service.h.
(aarch64_ps_get_thread_area): Declare.
* aarch64-linux-nat.c (ps_get_thread_area): Call
aarch64_ps_get_thread_area.
gdb/gdbserver:
2015-09-18 Yao Qi <yao.qi@linaro.org>
* linux-aarch64-low.c: Don't include sys/uio.h.
(ps_get_thread_area): Call aarch64_ps_get_thread_area.
In all-stop mode, record btrace maintains the old behaviour of an implicit
scheduler-locking on.
Now that we added a scheduler-locking mode to model this old behaviour, we
don't need the respective code in record btrace anymore. Remove it.
For all-stop targets, step inferior_ptid and continue other threads matching
the argument ptid. Assert that inferior_ptid matches the argument ptid.
This should make record btrace honour scheduler-locking.
gdb/
* record-btrace.c (record_btrace_resume): Honour scheduler-locking.
testsuite/
* gdb.btrace/multi-thread-step.exp: Test scheduler-locking on, step,
and replay.
Record targets behave as if scheduler-locking were on in replay mode. Add a
new scheduler-locking option "replay" to make this implicit behaviour explicit.
It behaves like "on" in replay mode and like "off" in record mode.
By making the current behaviour a scheduler-locking option, we allow the user
to change it. Since it is the current behaviour, this new option is also
the new default.
One caveat is that when resuming a thread that is at the end of its execution
history, record btrace implicitly stops replaying other threads and resumes
the entire process. This is a convenience feature to not require the user
to explicitly move all other threads to the end of their execution histories
before being able to resume the process.
We mimick this behaviour with scheduler-locking replay and move it from
record-btrace into infrun. With all-stop on top of non-stop, we can't do
this in record-btrace anymore.
Record full does not really support multi-threading and is therefore not
impacted. If it were extended to support multi-threading, it would 'benefit'
from this change. The good thing is that all record targets will behave the
same with respect to scheduler-locking.
I put the code for this into clear_proceed_status. It also sends the
about_to_proceed notification.
gdb/
* NEWS: Announce new scheduler-locking mode.
* infrun.c (schedlock_replay): New.
(scheduler_enums): Add schedlock_replay.
(scheduler_mode): Change default to schedlock_replay.
(user_visible_resume_ptid): Handle schedlock_replay.
(clear_proceed_status_thread): Stop replaying if resumed thread is
not replaying.
(schedlock_applies): Handle schedlock_replay.
(_initialize_infrun): Document new scheduler-locking mode.
* record-btrace.c (record_btrace_resume): Remove code to stop other
threads when not replaying the resumed thread.
doc/
* gdb.texinfo (All-Stop Mode): Describe new scheduler-locking mode.
Add a new target method to_record_will_replay to query if there is a record
target that will replay at least one thread matching the argument PTID if it
were executed in the argument execution direction.
gdb/
* record-btrace.c ((record_btrace_will_replay): New.
(init_record_btrace_ops): Initialize to_record_will_replay.
* record-full.c ((record_full_will_replay): New.
(init_record_full_ops): Initialize to_record_will_replay.
* target-delegates.c: Regenerated.
* target.c (target_record_will_replay): New.
* target.h (struct target_ops) <to_record_will_replay>: New.
(target_record_will_replay): New.
Signed-off-by: Markus Metzger <markus.t.metzger@intel.com>
The record btrace target does not allow accessing memory and storing registers
while replaying. For multi-threaded applications, this prevents those
accesses also for threads that are at the end of their execution history as
long as at least one thread is replaying.
Change this to only check if the selected thread is replaying. This allows
threads that are at the end of their execution history to read and write
memory and to store registers.
Also change the error message to reflect this change.
gdb/
* record-btrace.c (record_btrace_xfer_partial)
(record_btrace_store_registers, record_btrace_prepare_to_store):
Call record_btrace_is_replaying with inferior_ptid instead of
minus_one_ptid.
(record_btrace_store_registers): Change error message.
The to_record_is_replaying target method is used to query record targets if
they are replaying. This is currently interpreted as "is any thread being
replayed".
Add a PTID argument and change the interpretation to "is any thread matching
PTID being replayed".
Change all users to pass minus_one_ptid to preserve the old meaning.
The record full target does not really support multi-threading and ignores
the PTID argument.
gdb/
* record-btrace.c (record_btrace_is_replaying): Add ptid argument.
Update users to pass minus_one_ptid.
* record-full.c (record_full_is_replaying): Add ptid argument (ignored).
* record.c (cmd_record_delete): Pass inferior_ptid to
target_record_is_replaying.
* target-delegates.c: Regenerated.
* target.c (target_record_is_replaying): Add ptid argument.
* target.h (struct target_ops) <to_record_is_replaying>: Add ptid
argument.
(target_record_is_replaying): Add ptid argument.
A thread that runs out of its execution history is stopped. We already set
stop_pc and call stop_waiting. But we do not switch to the stopped thread.
In normal_stop, we call finish_thread_state_cleanup to set a thread's running
state. In all-stop mode, we call it with minus_one_ptid; in non-stop mode, we
only call it for inferior_ptid.
If in non-stop mode normal_stop is called on behalf of a thread that is not
inferior_ptid, that other thread will still be reported as running. If it is
actually stopped it can't be resumed again.
Record targets traditionally don't support non-stop and only resume
inferior_ptid. So this has not been a problem, so far.
Switch to the eventing thread for NO_HISTORY events as preparation to support
non-stop for the record btrace target.
gdb/
* infrun.c (handle_inferior_event_1): Switch to the eventing thread
in the TARKET_WAITKIND_NO_HISTORY case.
The record btrace target runs synchronous with GDB. That is, GDB steps
resumed threads in record btrace's to_wait method. Without GDB calling
to_wait, nothing happens 'on the target'.
Check for further expected events in to_wait before reporting the current
event and mark record btrace's async event handler in async mode.
gdb/
* record-btrace.c (record_btrace_maybe_mark_async_event): New.
(record_btrace_wait): Call record_btrace_maybe_mark_async_event.
Get_current_frame uses inferior_ptid. In record_btrace_start_replaying,
we need to get the current frame of the argument thread. So far, this
has always been inferior_ptid. With non-stop, this is not guaranteed.
Temporarily set inferior_ptid to the ptid of the argument thread.
We already temporarily set the argument thread's executing flag to false.
Move both into a new function get_thread_current_frame that does the temporary
adjustments, calls get_current_frame, and restores the previous values.
gdb/
* record-btrace.c (get_thread_current_frame): New.
(record_btrace_start_replaying): Call get_thread_current_frame.
The record targets are implicitly schedlocked. They only step the current
thread and keep other threads where they are.
Change record btrace to step all requested threads in to_resume.
For maintenance and debugging, we keep the old behaviour when the target below
is not non-stop. Enable with "maint set target-non-stop on".
gdb/
* record-btrace.c (record_btrace_resume_thread): A move request
overwrites a previous move request.
(record_btrace_find_resume_thread): Removed.
(record_btrace_resume): Resume all requested threads.
Record btrace's to_wait method picks a single thread to step. When passed
minus_one_ptid, it picks the current thread. All other threads remain where
they are.
Change this to step all resumed threads together, one step at a time, until
the first thread reports an event.
We do delay reporting NO_HISTORY events until there are no other events to
report to prevent threads at the end of their execution history from starving
other threads.
We keep threads at the end of their execution history moving and replaying
until we announce their stop in to_wait. This shouldn't really be user-visible
but its a detail worth mentioning.
Since record btrace's to_resume method also picks only a single thread to
resume, there shouldn't be a difference with the current all-stop.
With non-stop or all-stop on top of non-stop, we will see differences. The
behaviour should be more natural as we're moving all threads.
gdb/
* record-btrace.c: Include vec.h.
(record_btrace_find_thread_to_move): Removed.
(btrace_step_no_resumed, btrace_step_again)
(record_btrace_stop_replaying_at_end): New.
(record_btrace_cancel_resume): Call record_btrace_stop_replaying_at_end.
(record_btrace_single_step_forward): Remove calls to
record_btrace_stop_replaying.
(record_btrace_step_thread): Do only one step for BTHR_CONT and
BTHR_RCONT. Keep threads at the end of their history moving.
(record_btrace_wait): Call record_btrace_step_thread for all threads
until one reports an event. Call record_btrace_stop_replaying_at_end
for the eventing thread.
If a single-step ended right at the end of the execution history, we forgot
to announce that. Fix it.
gdb/
* record-btrace.c (record_btrace_single_step_forward): Return
NO_HISTORY if a step brings us to the end of the execution history.
Breakpoints are only checked for BTHR_CONT and BTHR_RCONT stepping requests.
A BTHR_STEP and BTHR_RSTEP request will always report stopped without reason.
Since breakpoints are reported correctly, I assume infrun is handling this.
Move the breakpoint check into the btrace single stepping functions. This
will cause us to report breakpoint hits now also for single-step requests.
One thing to notice is that
- when executing forwards, the breakpoint is checked before 'executing'
the instruction, i.e. before moving the PC to the next instruction.
- when executing backwards, the breakpoint is checked after 'executing'
the instruction, i.e. after moving the PC to the preceding instruction
in the recorded execution.
There is code in infrun (see, for example proceed and adjust_pc_after_break)
that handles this and also depends on this behaviour.
gdb/
* record-btrace.c (record_btrace_step_thread): Move breakpoint check
to ...
(record_btrace_single_step_forward): ... here and
(record_btrace_single_step_backward): ... here.
The code for BTHR_STEP and BTHR_CONT is fairly similar. Extract the common
parts into a new function record_btrace_single_step_forward. The function
returns TARGET_WAITKIND_SPURIOUS to indicate that the single-step completed
without triggering a trap.
Same for BTHR_RSTEP and BTHR_RCONT.
gdb/
* record-btrace.c (btrace_step_spurious)
(record_btrace_single_step_forward)
(record_btrace_single_step_backward): New.
(record_btrace_step_thread): Call record_btrace_single_step_forward
and record_btrace_single_step_backward.
There are two places where record_btrace_step_thread checks for a breakpoint
at the current replay position. Move this code into its own function.
gdb/
* record-btrace.c (record_btrace_replay_at_breakpoint): New.
(record_btrace_step_thread): Call record_btrace_replay_at_breakpoint.
Add support for the to_stop target method to the btrace record target.
gdb/
* btrace.h (enum btrace_thread_flag) <BTHR_STOP>: New.
* record-btrace (record_btrace_resume_thread): Clear BTHR_STOP.
(record_btrace_find_thread_to_move): Also accept threads that have
BTHR_STOP set.
(btrace_step_stopped_on_request, record_btrace_stop): New.
(record_btrace_step_thread): Support BTHR_STOP.
(record_btrace_wait): Also clear BTHR_STOP when stopping other threads.
(init_record_btrace_ops): Initialize to_stop.
The record btrace target stops other threads in non-stop mode after stepping
the to-be-resumed thread.
The check is done on the non_stop variable. It should rather be done on
target_is_non_stop_p (). With all-stop on top of non-stop, infrun will
take care of stopping other threads.
gdb/
* record-btrace.c (record_btrace_wait): Replace non_stop check with
target_is_non_stop_p ().
This change is relevant only for standard DWARF (as opposed to the GNAT
encodings extensions): at the time of writing it only makes a difference
with GCC patches that are to be integrated: see the patch series
submission at
<https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01353.html>.
Given the following Ada declarations:
subtype Small_Int is Natural range 0 .. 100;
type R_Type (L : Small_Int := 0) is record
S : String (1 .. L);
end record;
type A_Type is array (Natural range <>) of R_Type;
A : A_Type := (1 => (L => 0, S => ""),
2 => (L => 2, S => "ab"));
Before this change, we would get the following GDB session:
(gdb) ptype a
type = array (1 .. 2) of foo.r_type <packed: 838-bit elements>
This is wrong: "a" is not a packed array. This output comes from the
fact that, because R_Type has a dynamic size (with a maximum), the
compiler has to describe in the debugging information the size allocated
for each array element (i.e. the stride, in DWARF parlance: see
DW_AT_byte_stride). Ada type printing currently assumes that arrays
with a stride are packed, hence the above output.
In practice, GNAT never performs bit-packing for arrays that contain
variable-sized elements. Leveraging this fact, this patch enhances type
printing so that ptype does not pretend that arrays are packed when they
have a stride and they contain dynamic elements. After this change, we
get the following expected output:
(gdb) ptype a
type = array (1 .. 2) of foo.r_type
gdb/ChangeLog:
* ada-typeprint.c (print_array_type): Do not describe arrays as
packed when they embed dynamic elements.
gdb/testsuite/ChangeLog:
* gdb.ada/array_of_variable_length.exp: New testcase.
* gdb.ada/array_of_variable_length/foo.adb: New file.
* gdb.ada/array_of_variable_length/pck.adb: New file.
* gdb.ada/array_of_variable_length/pck.ads: New file.
Tested on x86_64-linux, no regression.
ppc64le loses control when stepping between two PLT-called functions inside
a shared library:
29 shlib_second (); /* first-hit */^M
(gdb) PASS: gdb.base/solib-intra-step.exp: first-hit
step^M
^M
Program received signal SIGABRT, Aborted.^M
0x00003fffb7cbe578 in __GI_raise (sig=<optimized out>) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56^M
56 return INLINE_SYSCALL (tgkill, 3, pid, selftid, sig);^M
(gdb) FAIL: gdb.base/solib-intra-step.exp: second-hit
->
29 shlib_second (); /* first-hit */^M
(gdb) PASS: gdb.base/solib-intra-step.exp: first-hit
step^M
shlib_second () at ./gdb.base/solib-intra-step-lib.c:23^M
23 abort (); /* second-hit */^M
(gdb) PASS: gdb.base/solib-intra-step.exp: second-hit
This is because gdbarch_skip_trampoline_code() will resolve the final function
as shlib_second+0 and place there the breakpoint, but ld.so will jump after
the breakpoint - at shlib_second+8 - as it is ELFv2 local symbol optimization:
Dump of assembler code for function shlib_second:
0x0000000000000804 <+0>: addis r2,r12,2
0x0000000000000808 <+4>: addi r2,r2,30668
0x000000000000080c <+8>: mflr r0
Currently gdbarch_skip_entrypoint() has been called in skip_prologue_sal() and
fill_in_stop_func() but that is not enough. I believe
gdbarch_skip_entrypoint() should be called after every
gdbarch_skip_trampoline_code().
gdb/ChangeLog
2015-09-15 Jan Kratochvil <jan.kratochvil@redhat.com>
* linespec.c (minsym_found): Call gdbarch_skip_entrypoint.
* ppc64-tdep.c (ppc64_skip_trampoline_code): Rename to ...
(ppc64_skip_trampoline_code_1): ... here.
(ppc64_skip_trampoline_code): New wrapper function.
* symtab.c (find_function_start_sal): Call gdbarch_skip_entrypoint.
gdb/testsuite/ChangeLog
2015-09-15 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.opt/solib-intra-step-lib.c: New file.
* gdb.opt/solib-intra-step-main.c: New file.
* gdb.opt/solib-intra-step.exp: New file.
gdb/ChangeLog -> gdb/gdbserver/ChangeLog
2015-09-15 Pedro Alves <palves@redhat.com>
PR remote/18965
* remote-utils.c (prepare_resume_reply): Merge
TARGET_WAITKIND_VFORK_DONE switch case with the
TARGET_WAITKIND_FORKED case.
The vforkdone stop reply misses indicating the thread ID of the vfork
parent which the event relates to:
@cindex vfork events, remote reply
@item vfork
The packet indicates that @code{vfork} was called, and @var{r}
is the thread ID of the new child process. Refer to
@ref{thread-id syntax} for the format of the @var{thread-id}
field. This packet is only applicable to targets that support
vfork events.
@cindex vforkdone events, remote reply
@item vforkdone
The packet indicates that a child process created by a vfork
has either called @code{exec} or terminated, so that the
address spaces of the parent and child process are no longer
shared. The @var{r} part is ignored. This packet is only
applicable to targets that support vforkdone events.
Unfortunately, this is not just a documentation issue. GDBserver
is really not specifying the thread ID. I noticed because
in non-stop mode, gdb complains:
[Thread 6089.6089] #1 stopped.
#0 0x0000003615a011f0 in ?? ()
0x0000003615a011f0 in ?? ()
(gdb) set debug remote 1
(gdb) c
Continuing.
Sending packet: $QPassSignals:e;10;14;17;1a;1b;1c;21;24;25;2c;4c;#5f...Packet received: OK
Sending packet: $vCont;c:p17c9.17c9#88...Packet received: OK
Notification received: Stop:T05vfork:p17ce.17ce;06:40d7ffffff7f0000;07:30d7ffffff7f0000;10:e4c9eb1536000000;thread:p17c9.17c9;core:2;
Sending packet: $vStopped#55...Packet received: OK
Sending packet: $D;17ce#af...Packet received: OK
Sending packet: $vCont;c:p17c9.17c9#88...Packet received: OK
Notification received: Stop:T05vforkdone:;
No process or thread specified in stop reply: T05vforkdone:;
(gdb)
This is not non-stop-mode-specific, however. Consider e.g., that in
all-stop, you may be debugging more than one process at the same time.
You continue, and both processes vfork. So when you next get a
T05vforkdone, there's no way to tell which of the parent processes is
done with the vfork.
Tests will be added later.
Tested on x86_64 Fedora 20.
gdb/ChangeLog:
2015-09-15 Pedro Alves <palves@redhat.com>
PR remote/18965
* remote-utils.c (prepare_resume_reply): Merge
TARGET_WAITKIND_VFORK_DONE switch case with the
TARGET_WAITKIND_FORKED case.
gdb/doc/ChangeLog:
2015-09-15 Pedro Alves <palves@redhat.com>
PR remote/18965
* gdb.texinfo (Stop Reply Packets): Explain that vforkdone's 'r'
part indicates the thread ID of the parent process.
Nowadays, GDB only knows whether architecture supports hardware single
step or software single step (through gdbarch hook software_single_step),
and for a given instruction or instruction sequence, GDB knows how to
do single step (hardware or software). However, GDB doesn't know whether
the target supports hardware single step. It is possible that the
architecture doesn't support hardware single step, such as arm, but
the target supports, such as simulator. This was discussed in this
thread https://www.sourceware.org/ml/gdb/2009-12/msg00033.html before.
I encounter this problem for aarch64 multi-arch support. When aarch64
debugs arm program, gdbarch is arm, so software single step is still
used. However, the underneath linux kernel does support hardware
single step, so IWBN to use it.
This patch is to add a new target_ops hook to_can_do_single_step, and
only use it in arm_linux_software_single_step to decide whether or not
to use hardware single step. On the native aarch64 linux target, 1 is
returned. On other targets, -1 is returned. On the remote target, if
the target supports s and S actions in the vCont? reply, then target
can do single step. However, old GDBserver will send s and S in the
reply to vCont?, which will confuse new GDB. For example, old GDBserver
on arm-linux will send s and S in the reply to vCont?, but it doesn't
support hardware single step. On the other hand, new GDBserver, on
arm-linux for example, will not send s and S in the reply to vCont?,
but old GDB thinks it doesn't support vCont packet at all. In order
to address this problem, I add a new qSupported feature vContSupported,
which indicates GDB wants to know the supported actions in the reply
to vCont?, and qSupported response contains vContSupported if the
stub is able tell supported vCont actions in the reply of vCont?.
If the patched GDB talks with patched GDBserver on x86, the RSP traffic
is like this:
-> $qSupported:...+;vContSupported+
<- ...+;vContSupported+
...
-> $vCont?
<- vCont;c;C;t;s;S;r
then, GDB knows the stub can do single step, and may stop using software
single step even the architecture doesn't support hardware single step.
If the patched GDB talks with patched GDBserver on arm, the last vCont?
reply will become:
<- vCont;c;C;t
GDB thinks the target doesn't support single step, so it will use software
single step.
If the patched GDB talks with unpatched GDBserver, the RSP traffic is like
this:
-> $qSupported:...+;vContSupported+
<- ...+
...
-> $vCont?
<- vCont;c;C;t;s;S;r
although GDBserver returns s and S, GDB still thinks GDBserver may not
support single step because it doesn't support vContSupported.
If the unpatched GDB talks with patched GDBserver on x86, the RSP traffic
is like:
-> $qSupported:...+;
<- ...+;vContSupported+
...
-> $vCont?
<- vCont;c;C;t;s;S;r
Since GDB doesn't sent vContSupported in the qSupported feature, GDBserver
sends s and S regardless of the support of hardware single step.
gdb:
2015-09-15 Yao Qi <yao.qi@linaro.org>
* aarch64-linux-nat.c (aarch64_linux_can_do_single_step): New
function.
(_initialize_aarch64_linux_nat): Install it to to_can_do_single_step.
* arm-linux-tdep.c (arm_linux_software_single_step): Return 0
if target_can_do_single_step returns 1.
* remote.c (struct vCont_action_support) <s, S>: New fields.
(PACKET_vContSupported): New enum.
(remote_protocol_features): New element for vContSupported.
(remote_query_supported): Append "vContSupported+".
(remote_vcont_probe): Remove support_s and support_S, use
rs->supports_vCont.s and rs->supports_vCont.S instead. Disable
vCont packet if c and C actions are not supported.
(remote_can_do_single_step): New function.
(init_remote_ops): Install it to to_can_do_single_step.
(_initialize_remote): Call add_packet_config_cmd.
* target.h (struct target_ops) <to_can_do_single_step>: New field.
(target_can_do_single_step): New macro.
* target-delegates.c: Re-generated.
gdb/gdbserver:
2015-09-15 Yao Qi <yao.qi@linaro.org>
* server.c (vCont_supported): New global variable.
(handle_query): Set vCont_supported to 1 if "vContSupported+"
matches. Append ";vContSupported+" to own_buf.
(handle_v_requests): Append ";s;S" to own_buf if target supports
hardware single step or vCont_supported is false.
(capture_main): Set vCont_supported to zero.
gdb/doc:
2015-09-15 Yao Qi <yao.qi@linaro.org>
* gdb.texinfo (General Query Packets): Add vContSupported to
tables of 'gdbfeatures' and 'stub features' supported in the
qSupported packet, as well as to the list containing stub
feature details.
This patch is to fixup the siginfo_t when aarch64 gdb or gdbserver
read from or write to the arm inferior. It is to convert the
"struct siginfo_t" between aarch64 and arm, which is quite mechanical.
gdb/gdbserver:
2015-09-15 Yao Qi <yao.qi@linaro.org>
* linux-aarch64-low.c (aarch64_linux_siginfo_fixup): New
function.
(struct linux_target_ops the_low_target): Install
aarch64_linux_siginfo_fixup.
gdb:
2015-09-15 Yao Qi <yao.qi@linaro.org>
* aarch64-linux-nat.c (aarch64_linux_siginfo_fixup): New function.
(_initialize_aarch64_linux_nat): Call linux_nat_set_siginfo_fixup.
* nat/aarch64-linux.c (aarch64_compat_siginfo_from_siginfo):
New function.
(aarch64_siginfo_from_compat_siginfo): New function.
* nat/aarch64-linux.h: Include signal.h.
(compat_int_t, compat_uptr_t, compat_time_t): Typedef.
(compat_timer_t, compat_clock_t): Likewise.
(struct compat_timeval): New.
(union compat_sigval): New.
(struct compat_siginfo): New.
(cpt_si_pid, cpt_si_uid, cpt_si_timerid): New macros.
(cpt_si_overrun, cpt_si_status, cpt_si_utime): Likewise.
(cpt_si_stime, cpt_si_ptr, cpt_si_addr): Likewise.
(cpt_si_band, cpt_si_fd): Likewise.
This patch, relative to a tree with
https://sourceware.org/ml/gdb-patches/2015-08/msg00295.html, fixes
issues/crashes that trigger if something unexpected happens during a
hook-stop.
E.g., if the inferior disappears while running the hook-stop, we hit
failed assertions:
(gdb) define hook-stop
Type commands for definition of "hook-stop".
End with a line saying just "end".
>kill
>end
(gdb) si
Kill the program being debugged? (y or n) [answered Y; input not from terminal]
/home/pedro/gdb/mygit/build/../src/gdb/thread.c:88: internal-error: inferior_thread: Assertion `tp' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n)
I noticed that if a hook-stop issues a synchronous execution command,
we print the same stop event twice:
(gdb) define hook-stop
Type commands for definition of "hook-stop".
End with a line saying just "end".
>si
>end
(gdb) si
0x000000000040074a 42 args[i] = 1; /* Init value. */ <<<<<<< once
0x000000000040074a 42 args[i] = 1; /* Init value. */ <<<<<<< twice
(gdb)
In MI:
*stopped,reason="end-stepping-range",frame={addr="0x000000000040074a",func="main",args=[],file="threads.c",fullname="/home/pedro/gdb/tests/threads.c",line="42"},thread-id="1",stopped-threads="all",core="0"
*stopped,reason="end-stepping-range",frame={addr="0x000000000040074a",func="main",args=[],file="threads.c",fullname="/home/pedro/gdb/tests/threads.c",line="42"},thread-id="1",stopped-threads="all",core="0"
(gdb)
The fix has GDB stop processing the event if the context changed. I
don't expect people to be doing crazy things from the hook-stop.
E.g., it gives me headaches to try to come up a proper behavior for
handling a thread change from a hook-stop... (E.g., imagine the
hook-stop does thread N; step, with scheduler-locing on). I think the
most important bit here is preventing crashes.
The patch adds a new hook-stop.exp test that covers the above and also
merges in the old hook-stop-continue.exp and hook-stop-frame.exp into
the same framework.
gdb/ChangeLog:
2015-09-14 Pedro Alves <palves@redhat.com>
* infrun.c (current_stop_id): New global.
(get_stop_id, new_stop_id): New functions.
(fetch_inferior_event): Handle normal_stop proceeding the target.
(struct stop_context): New.
(save_stop_context, release_stop_context_cleanup)
(stop_context_changed): New functions.
(normal_stop): Return true if the hook-stop changes the stop
context.
* infrun.h (get_stop_id): Declare.
(normal_stop): Now returns int. Add documentation.
gdb/testsuite/ChangeLog:
2015-09-14 Pedro Alves <palves@redhat.com>
* gdb.base/hook-stop-continue.c: Delete.
* gdb.base/hook-stop-continue.exp: Delete.
* gdb.base/hook-stop-frame.c: Delete.
* gdb.base/hook-stop-frame.exp: Delete.
* gdb.base/hook-stop.c: New file.
* gdb.base/hook-stop.exp: New file.
This change is relevant only for standard DWARF (as opposed to the GNAT
encodings extensions): at the time of writing it only makes a difference
with GCC patches that are to be integrated: see in particular
<https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01364.html>.
Given the following Ada declarations:
type Small is mod 2 ** 6;
type Array_Type is array (0 .. 9) of Small
with Pack;
type Array_Access is access all Array_Type;
A : aliased Array_Type := (1, 2, 3, 4, 5, 6, 7, 8, 9, 10);
AA : constant Array_Type := A'Access;
Before this change, we would get the following GDB session:
(gdb) print aa.all(2)
$1 = 3
(gdb) print aa(2)
$2 = 16
This is wrong: both expression should yield the same value: 3. The
problem is simply that the routine which handles accesses to arrays lack
general handling for packed arrays. After this patch, we have the
expected output:
(gdb) print aa.all(2)
$1 = 3
(gdb) print aa(2)
$2 = 3
gdb/ChangeLog:
* ada-lang.c (ada_value_ptr_subscript): Update the heading
comment. Handle packed arrays.
gdb/testsuite/ChangeLog:
* gdb.ada/access_to_packed_array.exp: New testcase.
* gdb.ada/access_to_packed_array/foo.adb: New file.
* gdb.ada/access_to_packed_array/pack.adb: New file.
* gdb.ada/access_to_packed_array/pack.ads: New file.
Tested on x86_64-linux, no regression.
Commit fbea99ea8a added this to both the "Changes in GDB 7.10" and
"Changes since GDB 7.10" sections by mistake.
gdb/ChangeLog:
2015-09-14 Pedro Alves <palves@redhat.com>
* NEWS (Changes in GDB 7.10, New commands>: Remove duplicate
mention of maint set/show target-non-stop.
This patch adds documentation of support for exec events on
extended-remote Linux targets.
gdb/ChangeLog:
* NEWS: Announce new remote packets for the exec-events
feature and the exec-events feature and associated commands.
gdb/doc/ChangeLog:
* gdb.texinfo (Remote Configuration): Add exec event
feature to table of packet settings.
(Stop Reply Packets): Add exec events to the list of stop
reasons.
(General Query Packets): Add exec events to tables of
'gdbfeatures' and 'stub features' supported in the qSupported
packet, as well as to the list containing stub feature
details.
This patch implements exec catchpoints for extended-remote Linux
targets. The implementation follows the same approach used for
fork catchpoints, implementing extended-remote target routines for
inserting and removing the catchpoints by just checking if exec events
are supported. Existing host-side code and previous support for
extended-remote exec events takes care of the rest.
gdb/ChangeLog:
* remote.c (remote_exec_event_p): New function.
(remote_insert_exec_catchpoint): New function.
(remote_remove_exec_catchpoint): New function.
(init_extended_remote_ops): Initialize extended_remote_ops
members to_insert_exec_catchpoint and
to_remove_exec_catchpoint.
This patch implements support for exec events on extended-remote Linux
targets. Follow-exec-mode and rerun behave as expected. Catchpoints and
test updates are implemented in subsequent patches.
This patch was derived from a patch posted last October:
https://sourceware.org/ml/gdb-patches/2014-10/msg00877.html.
It was originally based on some work done by Luis Machado in 2013.
IMPLEMENTATION
----------------
Exec events are enabled via ptrace options.
When an exec event is detected by gdbserver, the existing process
data, along with all its associated lwp and thread data, is deleted
and replaced by data for a new single-threaded process. The new
process data is initialized with the appropriate parts of the state
of the execing process. This approach takes care of several potential
pitfalls, including:
* deleting the data for an execing non-leader thread before any
wait/sigsuspend occurs
* correctly initializing the architecture of the execed process
We then report the exec event using a new RSP stop reason, "exec".
When GDB receives an "exec" event, it saves the status in the event
structure's target_waitstatus field, like what is done for remote fork
events. Because the original and execed programs may have different
architectures, we skip parsing the section of the stop reply packet
that contains register data. The register data will be retrieved
later after the inferior's architecture has been set up by
infrun.c:follow_exec.
At that point the exec event is handled by the existing event handling
in GDB. However, a few changes were necessary so that
infrun.c:follow_exec could accommodate the remote target.
* Where follow-exec-mode "new" is handled, we now call
add_inferior_with_spaces instead of add_inferior with separate calls
to set up the program and address spaces. The motivation for this
is that add_inferior_with_spaces also sets up the initial architecture
for the inferior, which is needed later by target_find_description
when it calls target_gdbarch.
* We call a new target function, target_follow_exec. This function
allows us to store the execd_pathname in the inferior, instead of
using the static string remote_exec_file from remote.c. The static
string didn't work for follow-exec-mode "new", since once you switched
to the execed program, the original remote exec-file was lost. The
execd_pathname is now stored in the inferior's program space as a
REGISTRY field. All of the requisite mechanisms for this are
defined in remote.c.
gdb/gdbserver/ChangeLog:
* linux-low.c (linux_mourn): Static declaration.
(linux_arch_setup): Move in front of
handle_extended_wait.
(linux_arch_setup_thread): New function.
(handle_extended_wait): Handle exec events. Call
linux_arch_setup_thread. Make event_lwp argument a
pointer-to-a-pointer.
(check_zombie_leaders): Do not check stopped threads.
(linux_low_ptrace_options): Add PTRACE_O_TRACEEXEC.
(linux_low_filter_event): Add lwp and thread for exec'ing
non-leader thread if leader thread has been deleted.
Refactor code into linux_arch_setup_thread and call it.
Pass child lwp pointer by reference to handle_extended_wait.
(linux_wait_for_event_filtered): Update comment.
(linux_wait_1): Prevent clobbering exec event status.
(linux_supports_exec_events): New function.
(linux_target_ops) <supports_exec_events>: Initialize new member.
* lynx-low.c (lynx_target_ops) <supports_exec_events>: Initialize
new member.
* remote-utils.c (prepare_resume_reply): New stop reason 'exec'.
* server.c (report_exec_events): New global variable.
(handle_query): Handle qSupported query for exec-events feature.
(captured_main): Initialize report_exec_events.
* server.h (report_exec_events): Declare new global variable.
* target.h (struct target_ops) <supports_exec_events>: New
member.
(target_supports_exec_events): New macro.
* win32-low.c (win32_target_ops) <supports_exec_events>:
Initialize new member.
gdb/ChangeLog:
* infrun.c (follow_exec): Use process-style ptid for
exec message. Call add_inferior_with_spaces and
target_follow_exec.
* nat/linux-ptrace.c (linux_supports_traceexec): New function.
* nat/linux-ptrace.h (linux_supports_traceexec): Declare.
* remote.c (remote_pspace_data): New static variable.
(remote_pspace_data_cleanup): New function.
(get_remote_exec_file): New function.
(set_remote_exec_file_1): New function.
(set_remote_exec_file): New function.
(show_remote_exec_file): New function.
(remote_exec_file): Delete static variable.
(anonymous enum) <PACKET_exec_event_feature> New
enumeration constant.
(remote_protocol_features): Add entry for exec-events feature.
(remote_query_supported): Add client side of qSupported query
for exec-events feature.
(remote_follow_exec): New function.
(remote_parse_stop_reply): Handle 'exec' stop reason.
(extended_remote_run, extended_remote_create_inferior): Call
get_remote_exec_file and set_remote_exec_file_1.
(init_extended_remote_ops) <to_follow_exec>: Initialize new
member.
(_initialize_remote): Call
register_program_space_data_with_cleanup. Call
add_packet_config_cmd for remote exec-events feature.
Modify call to add_setshow_string_noescape_cmd for exec-file
to use new functions set_remote_exec_file and
show_remote_exec_file.
* target-debug.h, target-delegates.c: Regenerated.
* target.c (target_follow_exec): New function.
* target.h (struct target_ops) <to_follow_exec>: New member.
(target_follow_exec): Declare new function.
gdb/ChangeLog:
* aarch64-tdep.c (decode_cb): Move up comment describing the
encoding.
(decode_tb): Fix a typo in comment above the function. Move up
comment describing the encoding.
The encoding of the b.cond instruction is described in the architecture
reference manual as:
b.cond 0101 0100 iiii iiii iiii iiii iii0 cccc
So the mask should be 0xff000010.
gdb/ChangeLog:
* aarch64-tdep.c (decode_bcond): Fix incorrect mask.
