In C99 the sizeof operator computes the size of a variable length array
at runtime (6.5.3.4 The sizeof operator). This patch reflects the semantic
change in the debugger.
We now are able to get the size of a vla:
1| void foo (size_t n) {
2| int vla[n];
3| }
(gdb) p sizeof(vla)
yields N * sizeof(int).
* eval.c (evaluate_subexp_for_sizeof) <OP_VAR_VALUE>: If the type
passed to sizeof is dynamic evaluate the argument to compute the length.
The dwarf standard allow certain attributes to be expressed as dwarf
expressions rather than constants. For instance upper-/lowerbound attributes.
In case of a c99 variable length array the upperbound is a dynamic attribute.
With this change c99 vla behave the same as with static arrays.
1| void foo (size_t n) {
2| int ary[n];
3| memset(ary, 0, sizeof(ary));
4| }
(gdb) print ary
$1 = {0 <repeats 42 times>}
* dwarf2loc.c (dwarf2_locexpr_baton_eval): New function.
(dwarf2_evaluate_property): New function.
* dwarf2loc.h (dwarf2_evaluate_property): New function prototype.
* dwarf2read.c (attr_to_dynamic_prop): New function.
(read_subrange_type): Use attr_to_dynamic_prop to read high bound
attribute.
* gdbtypes.c: Include dwarf2loc.h.
(is_dynamic_type): New function.
(resolve_dynamic_type): New function.
(resolve_dynamic_bounds): New function.
(get_type_length): New function.
(check_typedef): Use get_type_length to compute type length.
* gdbtypes.h (TYPE_HIGH_BOUND_KIND): New macro.
(TYPE_LOW_BOUND_KIND): New macro.
(is_dynamic_type): New function prototype.
* value.c (value_from_contents_and_address): Call resolve_dynamic_type
to resolve dynamic properties of the type. Update comment.
* valops.c (get_value_at, value_at, value_at_lazy): Update comment.
The rational behind this patch is to get started to implement the feature
described in dwarf4 standard (2.19) Static and Dynamic Values of Attributes.
It adds new BOUND_PROP to store either a constant, exprloc, or reference to
describe an upper-/lower bound of a subrange. Other than that no new features
are introduced.
* dwarf2read.c (read_subrange_type): Use struct bound_prop for
declaring high/low bounds and change uses accordingly. Call
create_range_type instead of create_static_range_type.
* gdbtypes.c (create_range_type): New function.
(create_range_type): Convert bounds into struct bound_prop and pass
them to create_range_type.
* gdbtypes.h (struct bound_prop): New struct.
(create_range_type): New function prototype.
(struct range_bounds): Use struct bound_prop instead of LONGEST for
high/low bounds. Remove low_undefined/high_undefined and adapt all uses.
(TYPE_LOW_BOUND,TYPE_HIGH_BOUND): Adapt macros to refer to the static
part of the bound.
* parse.c (follow_types): Set high bound kind to BOUND_UNDEFINED.
* gdbtypes.c (create_static_range_type): Renamed from create_range_type.
* gdbtypes.h (create_static_range_type): Renamed from create_range_type.
* ada-lang.c: All uses of create_range_type updated.
* coffread.c: All uses of create_range_type updated.
* dwarf2read.c: All uses of create_range_type updated.
* f-exp.y: All uses of create_range_type updated.
* m2-valprint.c: All uses of create_range_type updated.
* mdebugread.c: All uses of create_range_type updated.
* stabsread.c: All uses of create_range_type updated.
* valops.c: All uses of create_range_type updated.
* valprint.c: All uses of create_range_type updated.
Breakpoints are supposed to be transparent to memory accesses. For
all kinds of breakpoints breakpoint_xfer_memory hides the breakpoint
instructions. However, sss breakpoints aren't tracked like all other
breakpoints, and nothing is taking care of hiding them from memory
reads.
Say, as is, a background step + disassemble will see breakpoints
instructions on software step targets. E.g., stepping over this line:
while (1);
with s&
and then "disassemble" would show sss breakpoints.
Actually, that's still not be possible to see today, because:
- in native Linux, you can't read memory while the program
is running.
- with Linux gdbserver, you can, but in the all-stop RSP you
can't talk to the server while the program is running...
- and with non-stop, on software step targets, we presently
force the use of displaced-stepping for all single-steps,
so no single-step breakpoints are used...
I've been working towards making non-stop not force displaced stepping
on sss targets, and I noticed the issue then. With that, I indeed see
this:
(gdb) set remote Z-packet off
(gdb) s&
(gdb) disassemble main
Dump of assembler code for function main:
0x000000000040049c <+0>: push %rbp
0x000000000040049d <+1>: mov %rsp,%rbp
0x00000000004004a0 <+4>: int3
0x00000000004004a1 <+5>: (bad)
End of assembler dump.
Instead of the correct:
(gdb) disassemble main
Dump of assembler code for function main:
0x000000000040049c <+0>: push %rbp
0x000000000040049d <+1>: mov %rsp,%rbp
0x00000000004004a0 <+4>: jmp 0x4004a0 <main+4>
This is actually one thing that my v1 of the recent "fix a bunch of
run control bugs" series was fixing, because it made sss breakpoints
be regular breakpoints in the breakpoint chain. But dropped it in the
version that landed in the tree, due to some problems.
So instead of making sss breakpoints regular breakpoints, go with a
simpler fix (at least for now) -- make breakpoint_xfer_memory take
software single-step breakpoints into account. After the patch, I get
the correct disassemble output.
Tested on x86_64 Fedora 17, and also on top of my "use software
single-step on x86" series.
Also fixes the issue pointed out by Yao at
https://sourceware.org/ml/gdb-patches/2014-04/msg00045.html, where the
prologue analysis/frame sniffing manages to see software step
breakpoint instructions.
gdb/
2014-04-10 Pedro Alves <palves@redhat.com>
* breakpoint.c (single_step_breakpoints)
(single_step_gdbarch): Move up in the file.
(one_breakpoint_xfer_memory): New function, factored out from ...
(breakpoint_xfer_memory): ... here. Also process single-step
breakpoints.
Due to a thinko, a message could be not understood and ignored. The result
was a dead-lock (gdb is waiting for an event that never happen). The port
of the thread was deallocated before new threads are discovered. As a
consequence, the origin of the message was unknown (instead of being
linked to the newly created thread).
gdb/
* darwin-nat.c (darwin_check_new_threads): Fix port leak, add
comments.
(darwin_decode_exception_message): Free port only after use.
* avr-tdep.c (struct gdbarch_tdep): Mention avrxmega in the comment.
(avr_gdbarch_init): Add xmega architectures given by bfd_architecture
when setting the size of call_length.
gdb.Value.dynamic_type is supposed to work for reference and pointer
values. However, the value object in the function 'valpy_get_dynamic_type'
was being dereferenced using 'value_ind' irrespective of the value type
being TYPE_CODE_PTR or TYPE_CODE_REF. This patch fixes that to use
'coerce_ref' for TYPE_CODE_REF values.
ChangeLog:
* python/py-value.c (valpy_get_dynamic_type): Use coerce_ref to
dereference TYPE_CODE_REF values.
testsuite/
* gdb.python/py-value.c: Improve test case.
* gdb.python/py-value.exp: Add new test.
When a VDSO gets large enough that it doesn't entirely fit in one page,
but not so large that the part described by the program header exceeds
one page, then gdb/BFD doesn't read the section headers and symbol
table information. This patch cures that by passing the size of the
vdso to BFD, and fixes a number of other issues in the BFD code.
bfd/
* elfcode.h (bfd_from_remote_memory): Add "size" parameter.
Consolidate code handling possible section headers past end of
segment. Don't use p_align for page size guess, instead use
minpagesize. Take note of ld.so clearing section headers when
p_memsz > p_filesz. Handle file header specifying no section
headers. Handle zero p_align throughout. Default loadbase to
zero. Add comments. Rename contents_size to high_offset, and
make it a bfd_vma. Delete unnecessary bfd_set_error calls.
* bfd-in.h (bfd_elf_bfd_from_remote_memory): Update prototpe.
* elf-bfd.h (struct elf_backend_data <elf_backend_from_remote_memory>):
Likewise.
(_bfd_elf32_bfd_from_remote_memory): Likewise.
(_bfd_elf64_bfd_from_remote_memory): Likewise.
* elf.c (bfd_elf_bfd_from_remote_memory): Adjust.
* bfd-in2.h: Regnerate.
gdb/
* symfile-mem.c (symbol_file_add_from_memory): Add size parameter.
Pass to bfd_elf_bfd_from_remote_memory. Adjust all callers.
(struct symbol_file_add_from_memory_args): Add size field.
(find_vdso_size): New function.
(add_vsyscall_page): Attempt to find vdso size.
it is possible that gdb gets mach exceptions from an unknown inferior. This
happens when an inferior creates a child and that child gets a signal.
So instead of reporting messages with unknown origins, simply reply to these
notifications. The kernel will then post the unix signal.
gdb/
* darwin-nat.c (darwin_encode_reply): Add prototype.
(darwin_decode_exception_message): Reply to unknown inferiors.
(darwin_decode_message): Handle message by id. Ignore message
to unknown inferior.
(darwin_wait): Discard unknown messages, add debug trace.
* NEWS: Mention it.
* solib.c (solib_read_symbols): Only print symbol loading messages
if requested.
(solib_add): If symbol loading is in "brief" mode, notify user
symbols are being loaded.
(reload_shared_libraries_1): Ditto.
* symfile.c (print_symbol_loading_off): New static global.
(print_symbol_loading_brief): New static global.
(print_symbol_loading_full): New static global.
(print_symbol_loading_enums): New static global.
(print_symbol_loading): New static global.
(print_symbol_loading_p): New function.
(symbol_file_add_with_addrs): Only print symbol loading messages
if requested.
(_initialize_symfile): Register "print symbol-loading" set/show
command.
* symfile.h (print_symbol_loading_p): Declare.
doc/
* gdb.texinfo (Symbols): Document set/show print symbol-loading.
testsuite/
* gdb.base/print-symbol-loading-lib.c: New file.
* gdb.base/print-symbol-loading-main.c: New file.
* gdb.base/print-symbol-loading.exp: New file.
Given the following variable...
BT : Bounded := New_Bounded (Low => 1, High => 3);
... where type Bounded is defined as a simple unconstrained array:
type Bounded is array (Integer range <>) of Integer;
Creating a varobj for that variable, and immediately asking for
varobj updates, GDB says that our varobj changed types!
(gdb)
-var-create bt * bt
^done,name="bt",numchild="3",value="[3]",type="<ref> array (1 .. 3) of integer",has_more="0"
(gdb)
-var-update 1 *
^done,changelist=[{name="bt",value="[3]",in_scope="true",type_changed="true",new_type="<ref> array (1 .. 3) of integer",new_num_children="3",has_more="0"}]
The expected output for the -var-update command is, in this case:
(gdb)
-var-update 1 *
^done,changelist=[]
The problem occurs because the ada-varobj module does not handle
references, and while the references gets stripped when the varobj
gets created, it doesn't when computing varobj updates.
More specifically, when creating the varobj, varobj_create creates
a new value which is a reference to a TYPE_CODE_ARRAY. It then calls
install_new_value which calls coerce_ref with the following comment:
/* We are not interested in the address of references, and given
that in C++ a reference is not rebindable, it cannot
meaningfully change. So, get hold of the real value. */
if (value)
value = coerce_ref (value);
This leaves the varobj's type component still a ref, while
the varobj's value is now our array, without the ref. This explains
why the "value" field in the varobj indicates an array with 3 elements
"[3]" while the "type" field shows a ref to an array. Generally
speaking, most users have said that showing the ref was a useful
piece of information, so this patch is not touching this part.
Next, when the user issues the -var-update request, varobj_update
calls value_of_root to compute the varobj's new value as well as
determine whether the value's type has changed or not. What happens
in a nutshell is that it calls value_of_root_1 (which re-evaluates
the expression and returns the corresponding new value), finds that
the new value is not NULL, and thus asks whether it has mutated:
else if (varobj_value_has_mutated (var, value, value_type (value)))
This then indirectly delegates the determination to the language-specific
callback, which fails, because it does not handle references.
This patch fixes the issue by adjusting varobj_value_has_mutated to
expect references, and strip them when seen. This allows the various
language-specific implementations to remain unaware of references.
gdb/ChangeLog:
* varobj.c (varobj_value_has_mutated): If NEW_VALUE is
a reference, strip the reference layer before calling
the lang_ops value_has_mutated callback.
gdb/testsuite/ChangeLog:
* gdb.ada/mi_dyn_arr: New testcase.
Sourcing a GDB script that runs foreground execution commands in
succession fails if the target can async:
Breakpoint 1, main () at ../../../src/gdb/testsuite/gdb.base/source-execution.c:36
36 func1 ();
(gdb) source ../../../src/gdb/testsuite/gdb.base/source-execution.gdb
../../../src/gdb/testsuite/gdb.base/source-execution.gdb:21: Error in sourced command file:
Cannot execute this command while the selected thread is running.
(gdb) FAIL: gdb.base/source-execution.exp: source source-execution.gdb
That is, after a foreground execution command, GDB moves on to the
following command immediately before waiting for the previous command
to complete.
https://sourceware.org/ml/gdb-patches/2011-09/msg00037.html (b4a14fd0)
addressed this for command lists, Python's gdb.execute, etc., but
missed "source". Fixed now in the same way.
gdb/
2014-03-25 Pedro Alves <palves@redhat.com>
* cli/cli-script.c (script_from_file): Force the interpreter to
sync mode.
gdb/testsuite/
2014-03-25 Pedro Alves <palves@redhat.com>
Doug Evans <dje@google.com>
* gdb.base/source-execution.c: New file.
* gdb.base/source-execution.exp: New file.
* gdb.base/source-execution.gdb: New file.
As we move code on reading unavailable memory to target side, GDB core
side doesn't need the "switching momentarily out of tfind mode" dance.
The target remote knows how to read live memory (through remote_ops).
Remove set_traceframe_number and
make_cleanup_restore_traceframe_number, since they are no longer used.
gdb:
2014-03-22 Yao Qi <yao@codesourcery.com>
* remote.c (target_read_live_memory): Remove.
(memory_xfer_live_readonly_partial): Rename it to
remote_xfer_live_readonly_partial. Remove argument 'object'.
All callers updated. Call remote_read_bytes_1
instead of target_read_live_memory.
* tracepoint.c (set_traceframe_number): Remove.
(make_cleanup_restore_traceframe_number): Likewise .
* tracepoint.h (set_traceframe_number): Remove declaration.
(make_cleanup_restore_traceframe_number): Likewise.
This patch moves code in remote_read_bytes on reading from the remote
stub to a new function remote_read_bytes_1.
gdb:
2014-03-22 Yao Qi <yao@codesourcery.com>
* remote.c (remote_read_bytes): Move code on reading from the
remote stub to ...
(remote_read_bytes_1): ... here. New function.
As a follow-up to
[PATCH 7/8] Adjust read_value_memory to use to_xfer_partial
https://sourceware.org/ml/gdb-patches/2014-02/msg00384.html
this patch moves traceframe_available_memory down to the target side.
After this patch, the gdb core code is cleaner, and code on handling
unavailable memory is moved to remote/tfile/ctf targets.
In details, this patch moves traceframe_available_memory code from
memory_xfer_partial_1 to remote target only, so remote target still
uses traceframe_info mechanism to check unavailable memory, and use
remote_ops to read them from read-only sections. We don't use
traceframe_info mechanism for tfile and ctf target, because it is
fast to iterate all traceframes from trace file, so the summary
information got from traceframe_info is not necessary.
This patch also moves two functions to remote.c from target.c,
because they are only used in remote.c. I'll clean them up in another
patch.
gdb:
2014-03-22 Yao Qi <yao@codesourcery.com>
* ctf.c (ctf_xfer_partial): Check the return value of
exec_read_partial_read_only, if it is not TARGET_XFER_OK,
return TARGET_XFER_UNAVAILABLE.
* tracefile-tfile.c (tfile_xfer_partial): Likewise.
* target.c (target_read_live_memory): Move it to remote.c.
(memory_xfer_live_readonly_partial): Likewise.
(memory_xfer_partial_1): Move some code to remote_read_bytes.
* remote.c (target_read_live_memory): Moved from target.c.
(memory_xfer_live_readonly_partial): Likewise.
(remote_read_bytes): New, factored out from
memory_xfer_partial_1.
ps -e | grep a.out
28886 pts/12 00:00:00 a.out
gdb -p 28886
Loaded symbols for /lib64/ld-linux-x86-64.so.2
0x0000003b0ccbc970 in __nanosleep_nocancel () from /lib64/libc.so.6
../../binutils-gdb/gdb/cleanups.c:265: internal-warning: restore_my_cleanups has found a stale cleanup
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n)
The backtrace of this issue:
(gdb) bt
file=0x8b0c10 "s' failed.", line=265, fmt=0x8b0c38 "nutils-gdb/gdb/cleanups.c",
ap=0x7fff803e3ed8) at ../../binutils-gdb/gdb/utils.c:748
fmt=0x8b0c38 "nutils-gdb/gdb/cleanups.c", ap=0x7fff803e3ed8)
at ../../binutils-gdb/gdb/utils.c:799
string=0x8b0c38 "nutils-gdb/gdb/cleanups.c") at ../../binutils-gdb/gdb/utils.c:809
at ../../binutils-gdb/gdb/cleanups.c:265
at ../../binutils-gdb/gdb/cleanups.c:276
at ../../binutils-gdb/gdb/exceptions.c:142
at ../../binutils-gdb/gdb/exceptions.c:203
command=0x5d5fb8 <attach_command_continuation_free_args+18>, arg=0x7fff803e525b "2914",
from_tty=1, mask=RETURN_MASK_ALL) at ../../binutils-gdb/gdb/exceptions.c:549
---Type <return> to continue, or q <return> to quit---
func_args=0x7fff803e4280, errstring=0x8cf2e4 "/local/bin", mask=RETURN_MASK_ALL)
at ../../binutils-gdb/gdb/exceptions.c:522
This is a new issue. It is introduced by commit https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=8bc2fe488957946d2cdccda3ce8d4f39e4003ea0
It removed the discard_cleanups (back_to) inside attach_command.
Then restore_my_cleanups will throw a internal_warning.
https://sourceware.org/ml/gdb-patches/2014-03/msg00374.html
2014-03-21 Hui Zhu <hui@codesourcery.com>
Pedro Alves <palves@redhat.com>
* darwin-nat.c (darwin_pid_to_exec_file): Change xmalloc to
static buffer.
* fbsd-nat.c (fbsd_pid_to_exec_file): Ditto.
* linux-nat.c (linux_child_pid_to_exec_file): Ditto.
* nbsd-nat.c (nbsd_pid_to_exec_file): Ditto.
-- Initial message by Tom Tromey:
While testing on AIX, I happened to notice an internal error coming
from parse_probes. This happens because there are no probes defined
on this platform. This patch fixes the problem by changing an assert
into an ordinary error, and then changing the relevant caller to cope.
This fixes a few tests on AIX; also regtested on x86-64 Fedora 18.
-- Followup by Sergio Durigan Junior:
By reading the patch (and the original code), I found it a little bit
obscure, so I took the liberty to try to improve it. Here's the patch.
Could you please take a look and see if it works on AIX (and also if you
like the approach)?
gdb/
2014-03-20 Tom Tromey <tromey@redhat.com>
Sergio Durigan Junior <sergiodj@redhat.com>
* probe.c (parse_probes): Turn assert into an ordinary error.
* break-catch-throw.c (re_set_exception_catchpoint): Ignore
exceptions when parsing probes. Rearrange the code for clarity.
With target async enabled, py-finish-breakpoint.exp triggers an
assertion failure.
The failure occurs because execute_command re-enters the event loop in
some circumstances, and in this case resets the sync_execution flag.
Then later GDB reaches this assertion in normal_stop:
gdb_assert (sync_execution || !target_can_async_p ());
In detail:
#1 - A synchronous execution command is run. sync_execution is set.
#2 - A python breakpoint is hit (TARGET_WAITKIND_STOPPED), and the
corresponding Python breakpoint's stop method is executed. When
and while python commands are executed, interpreter_async is
forced to 0.
#3 - The Python stop method happens to execute a not-execution-related
gdb command. In this case, "where 1".
#4 - Seeing that sync_execution is set, execute_command nests a new
event loop (although that wasn't necessary; this is the problem).
#5 - The linux-nat target's pipe in the event loop happens to be
marked. That's normal, due to this in linux_nat_wait:
/* If we requested any event, and something came out, assume there
may be more. If we requested a specific lwp or process, also
assume there may be more. */
The nested event loop thus immediately wakes up and calls
target_wait. No thread is actually executing in the inferior, so
the target returns TARGET_WAITKIND_NO_RESUMED.
#6 - normal_stop is reached. GDB prints "No unwaited-for children
left.", and resets the sync_execution flag (IOW, there are no
resumed threads left, so the synchronous command is considered
completed.) This is already bogus. We were handling a
breakpoint!
#7 - the nested event loop unwinds/ends. GDB is now back to handling
the python stop method (TARGET_WAITKIND_STOPPED), which decides
the breakpoint should stop. normal_stop is called for this
event. However, normal_stop actually works with the _last_
reported target status:
void
normal_stop (void)
{
struct target_waitstatus last;
ptid_t last_ptid;
struct cleanup *old_chain = make_cleanup (null_cleanup, NULL);
...
get_last_target_status (&last_ptid, &last);
...
if (last.kind == TARGET_WAITKIND_NO_RESUMED)
{
gdb_assert (sync_execution || !target_can_async_p ());
target_terminal_ours_for_output ();
printf_filtered (_("No unwaited-for children left.\n"));
}
And due to the nesting in execute command, the last event is now
TARGET_WAITKIND_NO_RESUMED, not the actual breakpoint event being
handled. This could be seen to be broken in itself, but we can
leave fixing that for another pass. The assertion is reached, and
fails.
execute_command has a comment explaining when it should synchronously
wait for events:
/* If the interpreter is in sync mode (we're running a user
command's list, running command hooks or similars), and we
just ran a synchronous command that started the target, wait
for that command to end. */
However, the code did not follow this comment -- it didn't check to
see if the command actually started the target, just whether the
target was executing a sync command at this point.
This patch fixes the problem by noting whether the target was
executing in sync_execution mode before running the command, and then
augmenting the condition to test this as well.
2014-03-20 Tom Tromey <tromey@redhat.com>
PR gdb/14135
* top.c (execute_command): Only dispatch events if the command
started the target.
When target-async is enabled, dprintf.exp fails:
Running ../../../src/gdb/testsuite/gdb.base/dprintf.exp ...
FAIL: gdb.base/dprintf.exp: 1st dprintf, call
FAIL: gdb.base/dprintf.exp: 2nd dprintf, call
FAIL: gdb.base/dprintf.exp: Set dprintf function
FAIL: gdb.base/dprintf.exp: 1st dprintf, fprintf
FAIL: gdb.base/dprintf.exp: 2nd dprintf, fprintf
Breakpoint 2, main (argc=1, argv=0x7fffffffd3f8) at ../../../src/gdb/testsuite/gdb.base/dprintf.c:33
33 int loc = 1234;
(gdb) continue
Continuing.
kickoff 1234
also to stderr 1234
At foo entry
(gdb) FAIL: gdb.base/dprintf.exp: 1st dprintf, call
The problem is that GDB gave the prompt back to the user too early.
This happens when calling functions while handling an event that
doesn't cause a user visible stop. dprintf with "set dprintf-style
gdb" is one such case. This patch adds a test case that has a
breakpoint with a condition that calls a function that returns false,
so that regression testing isn't dependent on the implementation of
dprintf.
The problem happens because run_inferior_call causes GDB to forget
that it is running in sync_execution mode, so any event that runs an
inferior call causes fetch_inferior_event to display the prompt, even
if the event should not result in a user visible stop (that is, gdb
resumes the inferior and waits for the next event).
This patch fixes the issue by noticing when GDB was in sync_execution
mode in run_inferior_call, and taking care to restore this state
afterward.
gdb/
2014-03-20 Tom Tromey <tromey@redhat.com>
PR cli/15718
* infcall.c: Include event-top.h.
(run_inferior_call): Call async_disable_stdin if needed.
gdb/testsuite/
2014-03-20 Tom Tromey <tromey@redhat.com>
Pedro Alves <palves@redhat.com>
PR cli/15718
* gdb.base/condbreak-call-false.c: New file.
* gdb.base/condbreak-call-false.exp: New file.
This test fails with current mainline.
If the program stopped for a breakpoint in thread 1, and then the user
switches to thread 2, and resumes the program, GDB first switches back
to thread 1 to step it over the breakpoint, in order to make progress.
However, that logic only considers the last reported event, assuming
only one thread needs that stepping over dance.
That's actually not true when we play with scheduler-locking. The
patch adds an example to the testsuite of multiple threads needing a
step-over before the stepping thread can be resumed. With current
mainline, the program re-traps the same breakpoint it had already
trapped before.
E.g.:
Breakpoint 2, main () at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:99
99 wait_threads (); /* set wait-threads breakpoint here */
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: continue to breakpoint: run to breakpoint
info threads
Id Target Id Frame
3 Thread 0x7ffff77c9700 (LWP 4310) "multiple-step-o" 0x00000000004007ca in child_function_3 (arg=0x1) at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:43
2 Thread 0x7ffff7fca700 (LWP 4309) "multiple-step-o" 0x0000000000400827 in child_function_2 (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:60
* 1 Thread 0x7ffff7fcb740 (LWP 4305) "multiple-step-o" main () at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:99
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: info threads shows all threads
set scheduler-locking on
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: set scheduler-locking on
break 44
Breakpoint 3 at 0x4007d3: file ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c, line 44.
(gdb) break 61
Breakpoint 4 at 0x40082d: file ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c, line 61.
(gdb) thread 3
[Switching to thread 3 (Thread 0x7ffff77c9700 (LWP 4310))]
#0 0x00000000004007ca in child_function_3 (arg=0x1) at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:43
43 (*myp) ++;
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: thread 3
continue
Continuing.
Breakpoint 3, child_function_3 (arg=0x1) at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:44
44 callme (); /* set breakpoint thread 3 here */
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: continue to breakpoint: run to breakpoint in thread 3
p *myp = 0
$1 = 0
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: unbreak loop in thread 3
thread 2
[Switching to thread 2 (Thread 0x7ffff7fca700 (LWP 4309))]
#0 0x0000000000400827 in child_function_2 (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:60
60 (*myp) ++;
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: thread 2
continue
Continuing.
Breakpoint 4, child_function_2 (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:61
61 callme (); /* set breakpoint thread 2 here */
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: continue to breakpoint: run to breakpoint in thread 2
p *myp = 0
$2 = 0
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: unbreak loop in thread 2
thread 1
[Switching to thread 1 (Thread 0x7ffff7fcb740 (LWP 4305))]
#0 main () at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:99
99 wait_threads (); /* set wait-threads breakpoint here */
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: thread 1
set scheduler-locking off
(gdb) PASS: gdb.threads/multiple-step-overs.exp: step: set scheduler-locking off
At this point all thread are stopped for a breakpoint that needs stepping over.
(gdb) step
Breakpoint 2, main () at ../../../src/gdb/testsuite/gdb.threads/multiple-step-overs.c:99
99 wait_threads (); /* set wait-threads breakpoint here */
(gdb) FAIL: gdb.threads/multiple-step-overs.exp: step
But that "step" retriggers the same breakpoint instead of making
progress.
The patch teaches GDB to step over all breakpoints of all threads
before resuming the stepping thread.
Tested on x86_64 Fedora 17, against pristine mainline, and also my
branch that implements software single-stepping on x86.
gdb/
2014-03-20 Pedro Alves <palves@redhat.com>
* infrun.c (prepare_to_proceed): Delete.
(thread_still_needs_step_over): New function.
(find_thread_needs_step_over): New function.
(proceed): If the current thread needs a step-over, set its
steping_over_breakpoint flag. Adjust to use
find_thread_needs_step_over instead of prepare_to_proceed.
(process_event_stop_test): For BPSTAT_WHAT_STOP_NOISY and
BPSTAT_WHAT_STOP_SILENT, assume the thread stopped for a
breakpoint.
(switch_back_to_stepped_thread): Step over breakpoints of all
threads not the stepping thread, before switching back to the
stepping thread.
gdb/testsuite/
2014-03-20 Pedro Alves <palves@redhat.com>
* gdb.threads/multiple-step-overs.c: New file.
* gdb.threads/multiple-step-overs.exp: New file.
* gdb.threads/signal-while-stepping-over-bp-other-thread.exp:
Adjust expected infrun debug output.
Even with deferred_step_ptid out of the way, GDB can still lose
watchpoints.
If a watchpoint triggers and the PC points to an address where a
thread-specific breakpoint for another thread is set, the thread-hop
code triggers, and we lose the watchpoint:
if (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP)
{
int thread_hop_needed = 0;
struct address_space *aspace =
get_regcache_aspace (get_thread_regcache (ecs->ptid));
/* Check if a regular breakpoint has been hit before checking
for a potential single step breakpoint. Otherwise, GDB will
not see this breakpoint hit when stepping onto breakpoints. */
if (regular_breakpoint_inserted_here_p (aspace, stop_pc))
{
if (!breakpoint_thread_match (aspace, stop_pc, ecs->ptid))
thread_hop_needed = 1;
^^^^^^^^^^^^^^^^^^^^^
}
And on software single-step targets, even without a thread-specific
breakpoint in the way, here in the thread-hop code:
else if (singlestep_breakpoints_inserted_p)
{
...
if (!ptid_equal (singlestep_ptid, ecs->ptid)
&& in_thread_list (singlestep_ptid))
{
/* If the PC of the thread we were trying to single-step
has changed, discard this event (which we were going
to ignore anyway), and pretend we saw that thread
trap. This prevents us continuously moving the
single-step breakpoint forward, one instruction at a
time. If the PC has changed, then the thread we were
trying to single-step has trapped or been signalled,
but the event has not been reported to GDB yet.
There might be some cases where this loses signal
information, if a signal has arrived at exactly the
same time that the PC changed, but this is the best
we can do with the information available. Perhaps we
should arrange to report all events for all threads
when they stop, or to re-poll the remote looking for
this particular thread (i.e. temporarily enable
schedlock). */
CORE_ADDR new_singlestep_pc
= regcache_read_pc (get_thread_regcache (singlestep_ptid));
if (new_singlestep_pc != singlestep_pc)
{
enum gdb_signal stop_signal;
if (debug_infrun)
fprintf_unfiltered (gdb_stdlog, "infrun: unexpected thread,"
" but expected thread advanced also\n");
/* The current context still belongs to
singlestep_ptid. Don't swap here, since that's
the context we want to use. Just fudge our
state and continue. */
stop_signal = ecs->event_thread->suspend.stop_signal;
ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;
ecs->ptid = singlestep_ptid;
ecs->event_thread = find_thread_ptid (ecs->ptid);
ecs->event_thread->suspend.stop_signal = stop_signal;
stop_pc = new_singlestep_pc;
}
else
{
if (debug_infrun)
fprintf_unfiltered (gdb_stdlog,
"infrun: unexpected thread\n");
thread_hop_needed = 1;
stepping_past_singlestep_breakpoint = 1;
saved_singlestep_ptid = singlestep_ptid;
}
}
}
we either end up with thread_hop_needed, ignoring the watchpoint
SIGTRAP, or switch to the stepping thread, again ignoring that the
SIGTRAP could be for some other event.
The new test added by this patch exercises both paths.
So the fix is similar to the deferred_step_ptid fix -- defer the
thread hop to _after_ the SIGTRAP had a change of passing through the
regular bpstat handling. If the wrong thread hits a breakpoint, we'll
just end up with BPSTAT_WHAT_SINGLE, and if nothing causes a stop,
keep_going starts a step-over.
Most of the stepping_past_singlestep_breakpoint mechanism is really
not necessary -- setting the thread to step over a breakpoint with
thread->trap_expected is sufficient to keep all other threads locked.
It's best to still keep the flag in some form though, because when we
get to keep_going, the software single-step breakpoint we need to step
over is already gone -- an optimization done by a follow up patch will
check whether a step-over is still be necessary by looking to see
whether the breakpoint is still there, and would find the thread no
longer needs a step-over, while we still want it.
Special care is still needed to handle the case of PC of the thread we
were trying to single-step having changed, like in the old code. We
can't just keep_going and re-step it, as in that case we can over-step
the thread (if it was already done with the step, but hasn't reported
it yet, we'd ask it to step even further). That's now handled in
switch_back_to_stepped_thread. As bonus, we're now using a technique
that doesn't lose signals, unlike the old code -- we now insert a
breakpoint at PC, and resume, which either reports the breakpoint
immediately, or any pending signal.
Tested on x86_64 Fedora 17, against pristine mainline, and against a
branch that implements software single-step on x86.
gdb/
2014-03-20 Pedro Alves <palves@redhat.com>
* breakpoint.c (single_step_breakpoint_inserted_here_p): Make
extern.
* breakpoint.h (single_step_breakpoint_inserted_here_p): Declare.
* infrun.c (saved_singlestep_ptid)
(stepping_past_singlestep_breakpoint): Delete.
(resume): Remove stepping_past_singlestep_breakpoint handling.
(proceed): Store the prev_pc of the stepping thread too.
(init_wait_for_inferior): Adjust. Clear singlestep_ptid and
singlestep_pc.
(enum infwait_states): Delete infwait_thread_hop_state.
(struct execution_control_state) <hit_singlestep_breakpoint>: New
field.
(handle_inferior_event): Adjust.
(handle_signal_stop): Delete stepping_past_singlestep_breakpoint
handling and the thread-hop code. Before removing single-step
breakpoints, check whether the thread hit a single-step breakpoint
of another thread. If it did, the trap is not a random signal.
(switch_back_to_stepped_thread): If the event thread hit a
single-step breakpoint, unblock it before switching to the
stepping thread. Handle the case of the stepped thread having
advanced already.
(keep_going): Handle the case of the current thread moving past a
single-step breakpoint.
gdb/testsuite/
2014-03-20 Pedro Alves <palves@redhat.com>
* gdb.threads/step-over-trips-on-watchpoint.c: New file.
* gdb.threads/step-over-trips-on-watchpoint.exp: New file.
Say the program is stopped at a breakpoint, and the user sets a
watchpoint. When the program is next resumed, GDB will first step
over the breakpoint, as explained in the manual:
@value {GDBN} normally ignores breakpoints when it resumes
execution, until at least one instruction has been executed. If it
it did not do this, you would be unable to proceed past a breakpoint
without first disabling the breakpoint. This rule applies whether
or not the breakpoint already existed when your program stopped.
However, GDB currently also removes watchpoints, catchpoints, etc.,
and that means that the first instruction off the breakpoint does not
trigger the watchpoint, catchpoint, etc.
testsuite/gdb.base/watchpoint.exp has a kfail for this.
The PR proposes installing watchpoints only when stepping over a
breakpoint, but that misses catchpoints, etc.
A better fix would instead work from the opposite direction -- remove
only real breakpoints, leaving all other kinds of breakpoints
inserted.
But, going further, it's really a waste to constantly remove/insert
all breakpoints when stepping over a single breakpoint (generating a
pair of RSP z/Z packets for each breakpoint), so the fix goes a step
further and makes GDB remove _only_ the breakpoint being stepped over,
leaving all others installed. This then has the added benefit of
reducing breakpoint-related RSP traffic substancialy when there are
many breakpoints set.
gdb/
2014-03-20 Pedro Alves <palves@redhat.com>
PR breakpoints/7143
* breakpoint.c (should_be_inserted): Don't insert breakpoints that
are being stepped over.
(breakpoint_address_match): Make extern.
* breakpoint.h (breakpoint_address_match): New declaration.
* inferior.h (stepping_past_instruction_at): New declaration.
* infrun.c (struct step_over_info): New type.
(step_over_info): New global.
(set_step_over_info, clear_step_over_info)
(stepping_past_instruction_at): New functions.
(handle_inferior_event): Clear the step-over info when
trap_expected is cleared.
(resume): Remove now stale comment.
(clear_proceed_status): Clear step-over info.
(proceed): Adjust step-over handling to set or clear the step-over
info instead of removing all breakpoints.
(handle_signal_stop): When setting up a thread-hop, don't remove
breakpoints here.
(stop_stepping): Clear step-over info.
(keep_going): Adjust step-over handling to set or clear step-over
info and then always inserting breakpoints, instead of removing
all breakpoints when stepping over one.
gdb/testsuite/
2014-03-20 Pedro Alves <palves@redhat.com>
PR breakpoints/7143
* gdb.base/watchpoint.exp: Mention bugzilla bug number instead of
old gnats gdb/38. Remove kfail. Adjust to use gdb_test instead
of gdb_test_multiple.
* gdb.cp/annota2.exp: Remove kfail for gdb/38.
* gdb.cp/annota3.exp: Remove kfail for gdb/38.
Consider the case of the user doing "step" in thread 2, while thread 1
had previously stopped for a breakpoint. In order to make progress,
GDB makes thread 1 step over its breakpoint first (with all other
threads stopped), and once that is over, thread 2 then starts stepping
(with thread 1 and all others running free, by default). If GDB
didn't do that, thread 1 would just trip on the same breakpoint
immediately again. This is what the prepare_to_proceed /
deferred_step_ptid code is all about.
However, deferred_step_ptid code resumes the target with:
resume (1, GDB_SIGNAL_0);
prepare_to_wait (ecs);
return;
Recall we were just stepping over a breakpoint when we get here. That
means that _nothing_ had installed breakpoints yet! If there's
another breakpoint just after the breakpoint that was just stepped,
we'll miss it. The fix for that would be to use keep_going instead.
However, there are more problems. What if the instruction that was
just single-stepped triggers a watchpoint? Currently, GDB just
happily resumes the thread, losing that too...
Missed watchpoints will need yet further fixes, but we should keep
those in mind.
So the fix must be to let the trap fall through the regular bpstat
handling, and only if no breakpoint, watchpoint, etc. claims the trap,
shall we switch back to the stepped thread.
Now, nowadays, we have code at the tail end of trap handling that does
exactly that -- switch back to the stepped thread
(switch_back_to_the_stepped_thread).
So the deferred_step_ptid code is just standing in the way, and can
simply be eliminated, fixing bugs in the process. Sweet.
The comment about spurious "Switching to ..." made me pause, but is
actually stale nowadays. That isn't needed anymore.
previous_inferior_ptid used to be re-set at each (internal) event, but
now it's only touched in proceed and normal stop.
The two tests added by this patch fail without the fix.
Tested on x86_64 Fedora 17 (also against my software single-stepping
on x86 branch).
gdb/
2014-03-20 Pedro Alves <palves@redhat.com>
* infrun.c (previous_inferior_ptid): Adjust comment.
(deferred_step_ptid): Delete.
(infrun_thread_ptid_changed, prepare_to_proceed)
(init_wait_for_inferior): Adjust.
(handle_signal_stop): Delete deferred_step_ptid handling.
gdb/testsuite/
2014-03-20 Pedro Alves <palves@redhat.com>
* gdb.threads/step-over-lands-on-breakpoint.c: New file.
* gdb.threads/step-over-lands-on-breakpoint.exp: New file.
gdb/
2014-03-18 Jan Kratochvil <jan.kratochvil@redhat.com>
PR gdb/15358
* defs.h (sync_quit_force_run): New declaration.
(QUIT): Check also SYNC_QUIT_FORCE_RUN.
* event-top.c (async_sigterm_handler): New declaration.
(async_sigterm_token): New variable.
(async_init_signals): Create also async_sigterm_token.
(async_sigterm_handler): New function.
(sync_quit_force_run): New variable.
(handle_sigterm): Replace quit_force call by other calls.
* utils.c (quit): Call quit_force if SYNC_QUIT_FORCE_RUN.
gdb/testsuite/
2014-03-18 Jan Kratochvil <jan.kratochvil@redhat.com>
PR gdb/15358
* gdb.base/gdb-sigterm.c: New file.
* gdb.base/gdb-sigterm.exp: New file.
Message-ID: <20140316135334.GA30698@host2.jankratochvil.net>
This change corrects GPR frame offset calculation for the e500v2
processor. On this target, featuring the SPE APU, GPRs are 64-bit and
are held in stack frames whole with the use of `evstdd' and `evldd'
instructions. Their integer 32-bit part occupies the low-order word and
therefore its offset varies between the two endiannesses possible.
* rs6000-tdep.c (rs6000_frame_cache): Correct little-endian GPR
offset into SPE pseudo registers.
Part of PR gdb/13860 is about the mi-solib.exp test's output being
different in sync vs async modes.
sync:
>./gdb -nx -q ./testsuite/gdb.mi/solib-main -ex "set stop-on-solib-events 1" -ex "set target-async off" -i=mi
=thread-group-added,id="i1"
~"Reading symbols from /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.mi/solib-main..."
~"done.\n"
(gdb)
&"start\n"
~"Temporary breakpoint 1 at 0x400608: file ../../../src/gdb/testsuite/gdb.mi/solib-main.c, line 21.\n"
=breakpoint-created,bkpt={number="1",type="breakpoint",disp="del",enabled="y",addr="0x0000000000400608",func="main",file="../../../src/gdb/testsuite/gdb.mi/solib-main.c",fullname="/home/pedro/gdb/mygit/src/gdb/testsuite/gdb.mi/solib-main.c",line="21",times="0",original-location="main"}
~"Starting program: /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.mi/solib-main \n"
=thread-group-started,id="i1",pid="17724"
=thread-created,id="1",group-id="i1"
^running
*running,thread-id="all"
(gdb)
=library-loaded,id="/lib64/ld-linux-x86-64.so.2",target-name="/lib64/ld-linux-x86-64.so.2",host-name="/lib64/ld-linux-x86-64.so.2",symbols-loaded="0",thread-group="i1"
~"Stopped due to shared library event (no libraries added or removed)\n"
*stopped,reason="solib-event",frame={addr="0x000000379180f990",func="_dl_debug_state",args=[],from="/lib64/ld-linux-x86-64.so.2"},thread-id="1",stopped-threads="all",core="3"
(gdb)
async:
>./gdb -nx -q ./testsuite/gdb.mi/solib-main -ex "set stop-on-solib-events 1" -ex "set target-async on" -i=mi
=thread-group-added,id="i1"
~"Reading symbols from /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.mi/solib-main..."
~"done.\n"
(gdb)
start
&"start\n"
~"Temporary breakpoint 1 at 0x400608: file ../../../src/gdb/testsuite/gdb.mi/solib-main.c, line 21.\n"
=breakpoint-created,bkpt={number="1",type="breakpoint",disp="del",enabled="y",addr="0x0000000000400608",func="main",file="../../../src/gdb/testsuite/gdb.mi/solib-main.c",fullname="/home/pedro/gdb/mygit/src/gdb/testsuite/gdb.mi/solib-main.c",line="21",times="0",original-location="main"}
~"Starting program: /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.mi/solib-main \n"
=thread-group-started,id="i1",pid="17729"
=thread-created,id="1",group-id="i1"
^running
*running,thread-id="all"
=library-loaded,id="/lib64/ld-linux-x86-64.so.2",target-name="/lib64/ld-linux-x86-64.so.2",host-name="/lib64/ld-linux-x86-64.so.2",symbols-loaded="0",thread-group="i1"
(gdb)
*stopped,reason="solib-event",thread-id="1",stopped-threads="all",core="1"
For now, let's focus only on the *stopped event. We see that the
async output is missing frame info. And this causes a test failure in
async mode, as "mi_expect_stop solib-event" wants to see the frame
info.
However, if we compare the event output when a real MI execution
command is used, compared to a CLI command (e.g., run vs -exec-run,
next vs -exec-next, etc.), we see:
>./gdb -nx -q ./testsuite/gdb.mi/solib-main -ex "set stop-on-solib-events 1" -ex "set target-async off" -i=mi
=thread-group-added,id="i1"
~"Reading symbols from /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.mi/solib-main..."
~"done.\n"
(gdb)
r
&"r\n"
~"Starting program: /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.mi/solib-main \n"
=thread-group-started,id="i1",pid="17751"
=thread-created,id="1",group-id="i1"
^running
*running,thread-id="all"
(gdb)
=library-loaded,id="/lib64/ld-linux-x86-64.so.2",target-name="/lib64/ld-linux-x86-64.so.2",host-name="/lib64/ld-linux-x86-64.so.2",symbols-loaded="0",thread-group="i1"
~"Stopped due to shared library event (no libraries added or removed)\n"
*stopped,reason="solib-event",frame={addr="0x000000379180f990",func="_dl_debug_state",args=[],from="/lib64/ld-linux-x86-64.so.2"},thread-id="1",stopped-threads="all",core="3"
(gdb)
-exec-run
=thread-exited,id="1",group-id="i1"
=thread-group-exited,id="i1"
=library-unloaded,id="/lib64/ld-linux-x86-64.so.2",target-name="/lib64/ld-linux-x86-64.so.2",host-name="/lib64/ld-linux-x86-64.so.2",thread-group="i1"
=thread-group-started,id="i1",pid="17754"
=thread-created,id="1",group-id="i1"
^running
*running,thread-id="all"
(gdb)
=library-loaded,id="/lib64/ld-linux-x86-64.so.2",target-name="/lib64/ld-linux-x86-64.so.2",host-name="/lib64/ld-linux-x86-64.so.2",symbols-loaded="0",thread-group="i1"
*stopped,reason="solib-event",thread-id="1",stopped-threads="all",core="1"
=thread-selected,id="1"
(gdb)
As seen above, with MI commands, the *stopped event _doesn't_ have
frame info. This is because normal_stop, as commanded by the result
of bpstat_print, skips printing frame info in this case (it's an
"event", not a "breakpoint"), and when the interpreter is MI,
mi_on_normal_stop skips calling print_stack_frame, as the normal_stop
call was already done with the MI uiout. This explains why the async
output is different even with a CLI command. Its because in async
mode, the mi_on_normal_stop path is always taken; it is always reached
with the MI uiout, because the stop is handled from the event loop,
instead of from within `proceed -> wait_for_inferior -> normal_stop'
with the interpreter overridden, as in sync mode.
This patch fixes the issue by making all cases output the same
*stopped event, by factoring out the print code from normal_stop, and
using it from mi_on_normal_stop as well. I chose the *stopped output
without a frame, mainly because that is what you already get if you
use MI execution commands, the commands frontends are supposed to use
(except when implementing a console). This patch makes it simpler to
tweak the MI output differently if desired, as we only have to change
the centralized print_stop_event (taking into account whether the
uiout is MI-like), and all different modes will change accordingly.
Tested on x86_64 Fedora 17, no regressions. The mi-solib.exp test no
longer fails in async mode with this patch, so the patch removes the
kfail.
2014-03-18 Pedro Alves <palves@redhat.com>
PR gdb/13860
* inferior.h (print_stop_event): Declare.
* infrun.c (print_stop_event): New, factored out from ...
(normal_stop): ... this.
* mi/mi-interp.c (mi_on_normal_stop): Use print_stop_event instead
of bpstat_print/print_stack_frame.
2014-03-18 Pedro Alves <palves@redhat.com>
PR gdb/13860
* gdb.mi/mi-solib.exp: Remove gdb/13860 kfail.
* lib/mi-support.exp (mi_expect_stop): Add special handling for
solib-event.
The destructor code in ui-out.c has a latent bug, which is hidden by
the fact that nothing uses this right now. This patch fixes the
problem. The bug is that we don't always clear a pointer in the
ui-out object, leading to a bad free.
2014-03-17 Tom Tromey <tromey@redhat.com>
* ui-out.c (clear_table, ui_out_new): Clear uiout->table.id.
Consider the following declarations:
type Packed_Array is array (Natural range <>) of Boolean;
pragma Pack (Packed_Array);
function Make (H, L : Natural) return Packed_Array is
begin
return (H .. L => False);
end Make;
A1 : Packed_Array := Make (1, 2);
A2 : Packed_Array renames A1;
One possible DWARF translation for A2 is:
<3><1e4>: Abbrev Number: 21 (DW_TAG_variable)
<1e5> DW_AT_name : a2
<1ea> DW_AT_type : <0x1d9>
<3><1d9>: Abbrev Number: 22 (DW_TAG_const_type)
<1da> DW_AT_type : <0x1de>
<3><1de>: Abbrev Number: 23 (DW_TAG_reference_type)
<1e0> DW_AT_type : <0x1a3>
<3><1a3>: Abbrev Number: 17 (DW_TAG_array_type)
<1a4> DW_AT_name : foo__Ta1S___XP1
<1a8> DW_AT_GNAT_descriptive_type: <0x16b>
<3><16b>: Abbrev Number: 6 (DW_TAG_typedef)
<16c> DW_AT_name : foo__Ta1S
<172> DW_AT_type : <0x176>
<3><176>: Abbrev Number: 17 (DW_TAG_array_type)
<177> DW_AT_name : foo__Ta1S
<17b> DW_AT_GNAT_descriptive_type: <0x223>
Here, foo__Ta1S___XP1 is the type used for the code generation while
foo__Ta1S is the source-level type. Both form a valid GNAT encoding for
a packed array type.
Trying to print A2 (1) can make GDB crash. This is because A2 is defined
as a reference to a GNAT encoding for a packed array. When decoding
constrained packed arrays, the ada_coerce_ref subprogram follows
references and returns a fixed type from the target type, peeling
the GNAT encoding for packed arrays. The remaining code assumes that
the resulting type is still such an encoding while we only have
a standard GDB array type, hence the crash:
arr = ada_coerce_ref (arr);
[...]
type = decode_constrained_packed_array_type (value_type (arr));
decode_constrained_packed_array_type assumes that its argument is
such an encoding. From its front comment:
/* The array type encoded by TYPE, where
ada_is_constrained_packed_array_type (TYPE). */
This patch simply replaces the call to ada_coerce_ref with a call
to coerce_ref in order to avoid prematurely transforming
the packed array type as a side-effect. This way, the remaining code
will always work with a GNAT encoding.
gdb/ChangeLog:
* ada-lang.c (decode_constrained_packed_array): Perform a
minimal coercion for reference with coerce_ref instead of
ada_coerce_ref.
