| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Compiling GDB with current GCC (1b4a63593b) runs into this:
src/gdb/location.c: In function 'int event_location_empty_p(const event_location*)':
src/gdb/location.c:963:38: error: the address of 'event_location::<unnamed union>::explicit_loc' will never be NULL [-Werror=address]
963 | return (EL_EXPLICIT (location) == NULL
| ^
src/gdb/location.c:57:30: note: 'event_location::<unnamed union>::explicit_loc' declared here
57 | struct explicit_location explicit_loc;
| ^~~~~~~~~~~~
GCC is right, EL_EXPLICIT is defined as returning the address of an
union field:
/* An explicit location. */
struct explicit_location explicit_loc;
#define EL_EXPLICIT(P) (&((P)->u.explicit_loc))
and thus must always be non-NULL.
Change-Id: Ie74fee7834495a93affcefce03c06e4d83ad8191
|
|
The 'show user' command (which shows the definition of non-python/scheme
user defined commands) is currently missing a completer. This is
mentioned in PR 16238. Having one can improve the user experience.
In this commit I propose an implementation for such completer as well as
the associated tests.
Tested on x86_64 GNU/Linux.
All feedbacks are welcome.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=16238
|
|
The test stops at <signal_handler called> which is the call to the handler
rather than in the handler as intended. This patch replaces the
gdb_test "$enter_cmd to handler" with a gdb_test_multiple test. The multiple
test looks for the stop at <signal_handler called>. If found, the command
is issued again. The test passes if gdb stops in the handler as expected.
(gdb) PASS: gdb.base/sigstep.exp: stepi to handler, nothing in handler, step
from handler: continue to signal
stepi
<signal handler called>
1: x/i $pc
=> 0x7ffff7f80440 <__kernel_start_sigtramp_rt64>: bctrl
(gdb) stepi
handler (sig=551) at sigstep.c:32
32 {
1: x/i $pc
=> 0x10000097c <handler>: addis r2,r12,2
(gdb) PASS: gdb.base/sigstep.exp: stepi to handler, nothing in handler,
step from handler: stepi to handler
Patch has been tested on x86_64-linux and ppc64le-linux with no test failures.
|
|
On OBS I ran into:
...
(gdb) PASS: gdb.base/foll-vfork.exp: exit: \
vfork relations in info inferiors: continue to child exit
info inferiors^M
Num Description Connection Executable ^M
1 <null> foll-vfork-exit ^M
* 2 <null> foll-vfork-exit ^M
(gdb) I'm the proud parent of child #5044!^M
FAIL: gdb.base/foll-vfork.exp: exit: vfork relations in info inferiors: \
vfork relation no longer appears in info inferiors (timeout)
...
Fix this by removing the '$' anchor in the corresponding '$gdb_prompt $'
regexps.
Tested on x86_64-linux.
|
|
GDB hangs when doing this:
- launch inferior with multiple threads
- multiple threads hit some breakpoint(s)
- one breakpoint hit is presented as a stop, the rest are saved as
pending wait statuses
- "set scheduler-locking on"
- resume the currently selected thread (because of scheduler-locking,
it's the only one resumed), let it execute until exit
- GDB hangs, not showing the prompt, impossible to interrupt with ^C
When the resumed thread exits, we expect the target to return a
TARGET_WAITKIND_NO_RESUMED event, and that's what we see:
[infrun] fetch_inferior_event: enter
[infrun] scoped_disable_commit_resumed: reason=handling event
[infrun] random_pending_event_thread: None found.
[Thread 0x7ffff7d9c700 (LWP 309357) exited]
[infrun] print_target_wait_results: target_wait (-1.0.0 [process -1], status) =
[infrun] print_target_wait_results: -1.0.0 [process -1],
[infrun] print_target_wait_results: status->kind = no-resumed
[infrun] handle_inferior_event: status->kind = no-resumed
[infrun] handle_no_resumed: TARGET_WAITKIND_NO_RESUMED (ignoring: found resumed)
[infrun] prepare_to_wait: prepare_to_wait
[infrun] reset: reason=handling event
[infrun] maybe_set_commit_resumed_all_targets: not requesting commit-resumed for target native, no resumed threads
[infrun] fetch_inferior_event: exit
The problem is in handle_no_resumed: we check if some other thread is
actually resumed, to see if we should ignore that event (see comments in
that function for more info). If this condition is true:
(thread->executing () || thread->has_pending_waitstatus ())
... then we ignore the event. The problem is that there are some non-resumed
threads with a pending event, which makes us ignore the event. But these
threads are not resumed, so we end up waiting while nothing executes, hence
waiting for ever.
My first fix was to change the condition to:
(thread->executing ()
|| (thread->resumed () && thread->has_pending_waitstatus ()))
... but then it occured to me that we could simply check for:
(thread->resumed ())
Since "executing" implies "resumed", checking simply for "resumed"
covers threads that are resumed and executing, as well as threads that
are resumed with a pending status, which is what we want.
Change-Id: Ie796290f8ae7f34c026ca3a8fcef7397414f4780
|
|
Add a regression test-case for commit a50bdb99afe "[gdb/tdep, rs6000] Don't
skip system call in skip_prologue":
- set a breakpoint on a local copy of glibc's _exit, and
- verify that it triggers.
The test-case uses an assembly file by default, but also has the possibility
to use a C source file instead.
Tested on ppc64le-linux. Verified that the test-case fails without
aforementioned commit, and passes with the commit. Both with assembly
and C source.
|
|
Simon pointed out that my recent patches to .debug_loclists caused
some regressions. After a brief discussion we realized it was because
his system compiler defaults to PIE.
This patch changes this code to unconditionally apply the text offset
here. It also changes loclist_describe_location to work more like
dwarf2_find_location_expression.
I tested this by running the gdb.dwarf2 tests both with and without
-pie.
|
|
Commit e86fc4a5bc37 ("PR 28447: implement multiple parameters for .file
on XCOFF") changes the structure associated to the internal
representation of files in COFF formats. However, gdb directory update
has been forgotten, leading to compilation errors of this kind:
CXX coffread.o
/home/simark/src/binutils-gdb/gdb/coffread.c: In function 'const char* coff_getfilename(internal_auxent*)':
/home/simark/src/binutils-gdb/gdb/coffread.c:1343:29: error: 'union internal_auxent::<unnamed struct>::<unnamed>' has no member named 'x_zeroes'
1343 | if (aux_entry->x_file.x_n.x_zeroes == 0)
| ^~~~~~~~
Fix it by adjusting the COFF code in GDB.
Change-Id: I703fa134bc722d47515efbd72b88fa5650af6c3c
|
|
Add a test-case to excercise the problem scenario reported in PR28527 and
fixed in commit a50bdb99afe "[gdb/tdep, rs6000] Don't skip system call in
skip_prologue":
- set a breakpoint on _exit, and
- verify that it triggers.
Note that this is not a regression test for that commit. Since the actual
code in _exit may vary across os instances, we cannot guarantee that the
problem will always trigger with this test-case.
Rather, this test-case is a version of the original test-case
(gdb.threads/process-dies-while-detaching.exp) that is minimal while still
reproducing the problem reported in PR28527, in that same setting.
The benefit of this test-case is that it exercise real-life code and may
expose similar problems in other settings. Also, it provides a much easier
test-case to investigate in case a similar problem occurs.
Tested on x86_64-linux and ppc64le-linux.
|
|
The purpose of this test is described in the comments in
dprintf-execution-x-script.exp.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=28308
The name of this new test was based on that of an existing test,
bp-cmds-execution-x-script.exp. I started off by copying that test,
adding to it, and then rewriting almost all of it. It's different
enough that I decided that listing the copyright year as 2021
was sufficient.
|
|
This commit fixes Bug 28308, titled "Strange interactions with
dprintf and break/commands":
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=28308
Since creating that bug report, I've found a somewhat simpler way of
reproducing the problem. I've encapsulated it into the GDB test case
which I've created along with this bug fix. The name of the new test
is gdb.base/dprintf-execution-x-script.exp, I'll demonstrate the
problem using this test case, though for brevity, I've placed all
relevant files in the same directory and have renamed the files to all
start with 'dp-bug' instead of 'dprintf-execution-x-script'.
The script file, named dp-bug.gdb, consists of the following commands:
dprintf increment, "dprintf in increment(), vi=%d\n", vi
break inc_vi
commands
continue
end
run
Note that the final command in this script is 'run'. When 'run' is
instead issued interactively, the bug does not occur. So, let's look
at the interactive case first in order to see the correct/expected
output:
$ gdb -q -x dp-bug.gdb dp-bug
... eliding buggy output which I'll discuss later ...
(gdb) run
Starting program: /mesquite2/sourceware-git/f34-master/bld/gdb/tmp/dp-bug
vi=0
dprintf in increment(), vi=0
Breakpoint 2, inc_vi () at dprintf-execution-x-script.c:26
26 in dprintf-execution-x-script.c
vi=1
dprintf in increment(), vi=1
Breakpoint 2, inc_vi () at dprintf-execution-x-script.c:26
26 in dprintf-execution-x-script.c
vi=2
dprintf in increment(), vi=2
Breakpoint 2, inc_vi () at dprintf-execution-x-script.c:26
26 in dprintf-execution-x-script.c
vi=3
[Inferior 1 (process 1539210) exited normally]
In this run, in which 'run' was issued from the gdb prompt (instead
of at the end of the script), there are three dprintf messages along
with three 'Breakpoint 2' messages. This is the correct output.
Now let's look at the output that I snipped above; this is the output
when 'run' is issued from the script loaded via GDB's -x switch:
$ gdb -q -x dp-bug.gdb dp-bug
Reading symbols from dp-bug...
Dprintf 1 at 0x40116e: file dprintf-execution-x-script.c, line 38.
Breakpoint 2 at 0x40113a: file dprintf-execution-x-script.c, line 26.
vi=0
dprintf in increment(), vi=0
Breakpoint 2, inc_vi () at dprintf-execution-x-script.c:26
26 dprintf-execution-x-script.c: No such file or directory.
vi=1
Breakpoint 2, inc_vi () at dprintf-execution-x-script.c:26
26 in dprintf-execution-x-script.c
vi=2
Breakpoint 2, inc_vi () at dprintf-execution-x-script.c:26
26 in dprintf-execution-x-script.c
vi=3
[Inferior 1 (process 1539175) exited normally]
In the output shown above, only the first dprintf message is printed.
The 2nd and 3rd dprintf messages are missing! However, all three
'Breakpoint 2...' messages are still printed.
Why does this happen?
bpstat_do_actions_1() in gdb/breakpoint.c contains the following
comment and code near the start of the function:
/* Avoid endless recursion if a `source' command is contained
in bs->commands. */
if (executing_breakpoint_commands)
return 0;
scoped_restore save_executing
= make_scoped_restore (&executing_breakpoint_commands, 1);
Also, as described by this comment prior to the 'async' field
in 'struct ui' in top.h, the main UI starts off in sync mode
when processing command line arguments:
/* True if the UI is in async mode, false if in sync mode. If in
sync mode, a synchronous execution command (e.g, "next") does not
return until the command is finished. If in async mode, then
running a synchronous command returns right after resuming the
target. Waiting for the command's completion is later done on
the top event loop. For the main UI, this starts out disabled,
until all the explicit command line arguments (e.g., `gdb -ex
"start" -ex "next"') are processed. */
This combination of things, the state of the static global
'executing_breakpoint_commands' plus the state of the async
field in the main UI causes this behavior.
This is a backtrace after hitting the dprintf breakpoint for
the second time when doing 'run' from the script file, i.e.
non-interactively:
Thread 1 "gdb" hit Breakpoint 3, bpstat_do_actions_1 (bsp=0x7fffffffc2b8)
at /ironwood1/sourceware-git/f34-master/bld/../../worktree-master/gdb/breakpoint.c:4431
4431 if (executing_breakpoint_commands)
#0 bpstat_do_actions_1 (bsp=0x7fffffffc2b8)
at gdb/breakpoint.c:4431
#1 0x00000000004d8bc6 in dprintf_after_condition_true (bs=0x1538090)
at gdb/breakpoint.c:13048
#2 0x00000000004c5caa in bpstat_stop_status (aspace=0x116dbc0, bp_addr=0x40116e, thread=0x137f450, ws=0x7fffffffc718,
stop_chain=0x1538090) at gdb/breakpoint.c:5498
#3 0x0000000000768d98 in handle_signal_stop (ecs=0x7fffffffc6f0)
at gdb/infrun.c:6172
#4 0x00000000007678d3 in handle_inferior_event (ecs=0x7fffffffc6f0)
at gdb/infrun.c:5662
#5 0x0000000000763cd5 in fetch_inferior_event ()
at gdb/infrun.c:4060
#6 0x0000000000746d7d in inferior_event_handler (event_type=INF_REG_EVENT)
at gdb/inf-loop.c:41
#7 0x00000000007a702f in handle_target_event (error=0, client_data=0x0)
at gdb/linux-nat.c:4207
#8 0x0000000000b8cd6e in gdb_wait_for_event (block=block@entry=0)
at gdbsupport/event-loop.cc:701
#9 0x0000000000b8d032 in gdb_wait_for_event (block=0)
at gdbsupport/event-loop.cc:597
#10 gdb_do_one_event () at gdbsupport/event-loop.cc:212
#11 0x00000000009d19b6 in wait_sync_command_done ()
at gdb/top.c:528
#12 0x00000000009d1a3f in maybe_wait_sync_command_done (was_sync=0)
at gdb/top.c:545
#13 0x00000000009d2033 in execute_command (p=0x7fffffffcb18 "", from_tty=0)
at gdb/top.c:676
#14 0x0000000000560d5b in execute_control_command_1 (cmd=0x13b9bb0, from_tty=0)
at gdb/cli/cli-script.c:547
#15 0x000000000056134a in execute_control_command (cmd=0x13b9bb0, from_tty=0)
at gdb/cli/cli-script.c:717
#16 0x00000000004c3bbe in bpstat_do_actions_1 (bsp=0x137f530)
at gdb/breakpoint.c:4469
#17 0x00000000004c3d40 in bpstat_do_actions ()
at gdb/breakpoint.c:4533
#18 0x00000000006a473a in command_handler (command=0x1399ad0 "run")
at gdb/event-top.c:624
#19 0x00000000009d182e in read_command_file (stream=0x113e540)
at gdb/top.c:443
#20 0x0000000000563697 in script_from_file (stream=0x113e540, file=0x13bb0b0 "dp-bug.gdb")
at gdb/cli/cli-script.c:1642
#21 0x00000000006abd63 in source_gdb_script (extlang=0xc44e80 <extension_language_gdb>, stream=0x113e540,
file=0x13bb0b0 "dp-bug.gdb") at gdb/extension.c:188
#22 0x0000000000544400 in source_script_from_stream (stream=0x113e540, file=0x7fffffffd91a "dp-bug.gdb",
file_to_open=0x13bb0b0 "dp-bug.gdb")
at gdb/cli/cli-cmds.c:692
#23 0x0000000000544557 in source_script_with_search (file=0x7fffffffd91a "dp-bug.gdb", from_tty=1, search_path=0)
at gdb/cli/cli-cmds.c:750
#24 0x00000000005445cf in source_script (file=0x7fffffffd91a "dp-bug.gdb", from_tty=1)
at gdb/cli/cli-cmds.c:759
#25 0x00000000007cf6d9 in catch_command_errors (command=0x5445aa <source_script(char const*, int)>,
arg=0x7fffffffd91a "dp-bug.gdb", from_tty=1, do_bp_actions=false)
at gdb/main.c:523
#26 0x00000000007cf85d in execute_cmdargs (cmdarg_vec=0x7fffffffd1b0, file_type=CMDARG_FILE, cmd_type=CMDARG_COMMAND,
ret=0x7fffffffd18c) at gdb/main.c:615
#27 0x00000000007d0c8e in captured_main_1 (context=0x7fffffffd3f0)
at gdb/main.c:1322
#28 0x00000000007d0eba in captured_main (data=0x7fffffffd3f0)
at gdb/main.c:1343
#29 0x00000000007d0f25 in gdb_main (args=0x7fffffffd3f0)
at gdb/main.c:1368
#30 0x00000000004186dd in main (argc=5, argv=0x7fffffffd508)
at gdb/gdb.c:32
There are two frames for bpstat_do_actions_1(), one at frame #16 and
the other at frame #0. The one at frame #16 is processing the actions
for Breakpoint 2, which is a 'continue'. The one at frame #0 is attempting
to process the dprintf breakpoint action. However, at this point,
the value of 'executing_breakpoint_commands' is 1, forcing an early
return, i.e. prior to executing the command(s) associated with the dprintf
breakpoint.
For the sake of comparison, this is what the stack looks like when hitting
the dprintf breakpoint for the second time when issuing the 'run'
command from the GDB prompt.
Thread 1 "gdb" hit Breakpoint 3, bpstat_do_actions_1 (bsp=0x7fffffffccd8)
at /ironwood1/sourceware-git/f34-master/bld/../../worktree-master/gdb/breakpoint.c:4431
4431 if (executing_breakpoint_commands)
#0 bpstat_do_actions_1 (bsp=0x7fffffffccd8)
at gdb/breakpoint.c:4431
#1 0x00000000004d8bc6 in dprintf_after_condition_true (bs=0x16b0290)
at gdb/breakpoint.c:13048
#2 0x00000000004c5caa in bpstat_stop_status (aspace=0x116dbc0, bp_addr=0x40116e, thread=0x13f0e60, ws=0x7fffffffd138,
stop_chain=0x16b0290) at gdb/breakpoint.c:5498
#3 0x0000000000768d98 in handle_signal_stop (ecs=0x7fffffffd110)
at gdb/infrun.c:6172
#4 0x00000000007678d3 in handle_inferior_event (ecs=0x7fffffffd110)
at gdb/infrun.c:5662
#5 0x0000000000763cd5 in fetch_inferior_event ()
at gdb/infrun.c:4060
#6 0x0000000000746d7d in inferior_event_handler (event_type=INF_REG_EVENT)
at gdb/inf-loop.c:41
#7 0x00000000007a702f in handle_target_event (error=0, client_data=0x0)
at gdb/linux-nat.c:4207
#8 0x0000000000b8cd6e in gdb_wait_for_event (block=block@entry=0)
at gdbsupport/event-loop.cc:701
#9 0x0000000000b8d032 in gdb_wait_for_event (block=0)
at gdbsupport/event-loop.cc:597
#10 gdb_do_one_event () at gdbsupport/event-loop.cc:212
#11 0x00000000007cf512 in start_event_loop ()
at gdb/main.c:421
#12 0x00000000007cf631 in captured_command_loop ()
at gdb/main.c:481
#13 0x00000000007d0ebf in captured_main (data=0x7fffffffd3f0)
at gdb/main.c:1353
#14 0x00000000007d0f25 in gdb_main (args=0x7fffffffd3f0)
at gdb/main.c:1368
#15 0x00000000004186dd in main (argc=5, argv=0x7fffffffd508)
at gdb/gdb.c:32
This relatively short backtrace is due to the current UI's async field
being set to 1.
Yet another thing to be aware of regarding this problem is the
difference in the way that commands associated to dprintf breakpoints
versus regular breakpoints are handled. While they both use a command
list associated with the breakpoint, regular breakpoints will place
the commands to be run on the bpstat chain constructed in
bp_stop_status(). These commands are run later on. For dprintf
breakpoints, commands are run via the 'after_condition_true' function
pointer directly from bpstat_stop_status(). (The 'commands' field in
the bpstat is cleared in dprintf_after_condition_true(). This
prevents the dprintf commands from being run again later on when other
commands on the bpstat chain are processed.)
Another thing that I noticed is that dprintf breakpoints are the only
type of breakpoint which use 'after_condition_true'. This suggests
that one possible way of fixing this problem, that of making dprintf
breakpoints work more like regular breakpoints, probably won't work.
(I must admit, however, that my understanding of this code isn't
complete enough to say why. I'll trust that whoever implemented it
had a good reason for doing it this way.)
The comment referenced earlier regarding 'executing_breakpoint_commands'
states that the reason for checking this variable is to avoid
potential endless recursion when a 'source' command appears in
bs->commands. We know that a dprintf command is constrained to either
1) execution of a GDB printf command, 2) an inferior function call of
a printf-like function, or 3) execution of an agent-printf command.
Therefore, infinite recursion due to a 'source' command cannot happen
when executing commands upon hitting a dprintf breakpoint.
I chose to fix this problem by having dprintf_after_condition_true()
directly call execute_control_commands(). This means that it no
longer attempts to go through bpstat_do_actions_1() avoiding the
infinite recursion check for potential 'source' commands on the
command chain. I think it simplifies this code a little bit too, a
definite bonus.
Summary:
* breakpoint.c (dprintf_after_condition_true): Don't call
bpstat_do_actions_1(). Call execute_control_commands()
instead.
|
|
Change-Id: I2141b0b8a09f6521a59908599eb5ba1a19b18dc6
|
|
AddressSanitizer
This test fails for me, showing:
ERROR: tcl error sourcing /home/smarchi/src/binutils-gdb/gdb/testsuite/gdb.debuginfod/fetch_src_and_symbols.exp.
ERROR: This GDB was configured as follows:
configure --host=x86_64-pc-linux-gnu --target=x86_64-pc-linux-gnu
--with-auto-load-dir=$debugdir:$datadir/auto-load
--with-auto-load-safe-path=$debugdir:$datadir/auto-load
... and much more ...
The problem is that TCL's exec throws an error as soon as the exec'ed
process outputs on stderr. When GDB is built with ASan, it prints some
warnings about pre-existing signal handlers:
warning: Found custom handler for signal 7 (Bus error) preinstalled.
warning: Found custom handler for signal 8 (Floating point exception) preinstalled.
warning: Found custom handler for signal 11 (Segmentation fault) preinstalled.
Pass --quiet to GDB to avoid these warnings.
Change-Id: I3751d89b9b1df646da19149d7cb86775e2d3e80f
|
|
When the code to handle DW_LLE_start_end was added (as part of some
DWARF 5 work), it was written to add the base address. However, this
seems incorrect -- the DWARF standard describes this as an address,
not an offset from the base address.
This patch changes a couple of spots in dwarf2/loc.c to fix this
problem. It then changes decode_debug_loc_addresses to return
DEBUG_LOC_OFFSET_PAIR instead, which preserves the previous semantics.
This only showed up on the RISC-V target internally, due to the
combination of DWARF 5 and a newer version of GCC. I've updated a
couple of existing loclists test cases to demonstrate the bug.
|
|
The rhES5 build failed due to an upstream import a while back. The
bug here is that, while the 'personality' function exists,
ADDR_NO_RANDOMIZE is only defined in <linux/personality.h>, not
<sys/personality.h>.
However, <linux/personality.h> does not declare the 'personality'
function, and <sys/personality.h> and <linux/personality.h> cannot
both be included.
This patch restores one of the removed configure checks and updates
the code to check it.
We had this as a local patch at AdaCore, because it seemed like there
was no interest upstream. However, now it turns out that this fixes
PR build/28555, so I'm sending it now.
|
|
While reading the interface of gdb::array_view, I realized that the
constructor that builds an array_view on top of a contiguous container
(such as std::vector, std::array or even gdb::array_view) can be
missused.
Lets consider the following code sample:
struct Parent
{
Parent (int a): a { a } {}
int a;
};
std::ostream &operator<< (std::ostream& os, const Parent & p)
{ os << "Parent {a=" << p.a << "}"; return os; }
struct Child : public Parent
{
Child (int a, int b): Parent { a }, b { b } {}
int b;
};
std::ostream &operator<< (std::ostream& os, const Child & p)
{ os << "Child {a=" << p.a << ", b=" << p.b << "}"; return os; }
template <typename T>
void print (const gdb::array_view<const T> &p)
{
std::for_each (p.begin (), p.end (), [](const T &p) { std::cout << p << '\n'; });
}
Then with the current interface nothinng prevents this usage of
array_view to be done:
const std::array<Child, 3> elts = {
Child {1, 2},
Child {3, 4},
Child {5, 6}
};
print_all<Parent> (elts);
This compiles fine and produces the following output:
Parent {a=1}
Parent {a=2}
Parent {a=3}
which is obviously wrong. There is nowhere in memory a Parent-like
object for which the A member is 2 and this call to print_all<Parent>
shold not compile at all (calling print_all<Child> is however fine).
This comes down to the fact that a Child* is convertible into a Parent*,
and that an array view is constructed to a pointer to the first element
and a size. The valid type pointed to that can be used with this
constructor are restricted using SFINAE, which requires that a
pointer to a member into the underlying container can be converted into a
pointer the array_view's data type.
This patch proposes to change the constraints on the gdb::array_view
ctor which accepts a container now requires that the (decayed) type of
the elements in the container match the (decayed) type of the array_view
being constructed.
Applying this change required minimum adjustment in GDB codebase, which
are also included in this patch.
Tested by rebuilding.
|
|
This commits adds const versions for the GET and AS_ARRAX_VIEW methods
of gdb_argv. Those methods will be required in the following patch of
the series.
|
|
Change-Id: I04403bd85ec3fa75ea14130d68daba675a2a8aeb
|
|
non-stop
When doing "continue -a" in non-stop mode, each thread is individually
resumed while the commit resumed state is enabled. This forces the
target to commit each resumption immediately, instead of being able to
batch things.
The reason is that there is no scoped_disable_commit_resumed around the
loop over threads in continue_1, when "non_stop && all_threads" is true.
Since the proceed function is called once for each thread, the
scoped_disable_commit_resumed in proceed therefore forces commit-resumed
between each thread resumption. Add the necessary
scoped_disable_commit_resumed in continue_1 to avoid that.
I looked at the MI side of things, the function exec_continue, and found
that it was correct. There is a similar iteration over threads, and
there is a scoped_disable_commit_resumed at the function scope. This is
not wrong, but a bit more than we need. The branches that just call
continue_1 do not need it, as continue_1 takes care of disabling commit
resumed. So, move the scoped_disable_commit_resumed to the inner scope
where we iterate on threads and proceed them individually.
Here's an example debugging a multi-threaded program attached by
gdbserver (debug output trimmed for brevity):
$ ./gdb -nx -q --data-directory=data-directory -ex "set non-stop" -ex "tar rem :1234"
(gdb) set debug remote
(gdb) set debug infrun
(gdb) c -a
Continuing.
[infrun] proceed: enter
[infrun] scoped_disable_commit_resumed: reason=proceeding
[remote] Sending packet: $vCont;c:p14388.14388#90
[infrun] reset: reason=proceeding
[infrun] maybe_set_commit_resumed_all_targets: enabling commit-resumed for target remote
[infrun] maybe_call_commit_resumed_all_targets: calling commit_resumed for target remote
[infrun] proceed: exit
[infrun] proceed: enter
[infrun] scoped_disable_commit_resumed: reason=proceeding
[remote] Sending packet: $vCont;c:p14388.1438a#b9
[infrun] reset: reason=proceeding
[infrun] maybe_set_commit_resumed_all_targets: enabling commit-resumed for target remote
[infrun] maybe_call_commit_resumed_all_targets: calling commit_resumed for target remote
[infrun] proceed: exit
... and so on for each thread ...
Notice how we send one vCont;c for each thread. With the patch applied, we
send a single vCont;c at the end:
[infrun] scoped_disable_commit_resumed: reason=continue all threads in non-stop
[infrun] proceed: enter
[infrun] scoped_disable_commit_resumed: reason=proceeding
[infrun] reset: reason=proceeding
[infrun] proceed: exit
[infrun] clear_proceed_status_thread: Thread 85790.85792
[infrun] proceed: enter
[infrun] scoped_disable_commit_resumed: reason=proceeding
[infrun] reset: reason=proceeding
[infrun] proceed: exit
... proceeding threads individually ...
[infrun] reset: reason=continue all threads in non-stop
[infrun] maybe_set_commit_resumed_all_targets: enabling commit-resumed for target remote
[infrun] maybe_call_commit_resumed_all_targets: calling commit_resumed for target remote
[remote] Sending packet: $vCont;c#a8
Change-Id: I331dd2473c5aa5114f89854196fed2a8fdd122bb
|
|
While reading another patch, I saw that this function didn't need to
take a dwarf2_per_objfile, but could take a dwarf2_per_bfd instead.
It doesn't change the behavior, but doing this shows that this function
is objfile-independent (can work with only the shared per-bfd data).
Change-Id: I58f9c9cef6688902e95226480285da2d0005d77f
|
|
I don't find that the bpstat typedef, which hides a pointer, is
particularly useful. In fact, it confused me many times, and I just see
it as something to remember that adds cognitive load. Also, with C++,
we might want to be able to pass bpstats objects by const-reference, not
necessarily by pointer.
So, remove the bpstat typedef and rename struct bpstats to bpstat (since
it represents one bpstat, it makes sense that it is singular).
Change-Id: I52e763b6e54ee666a9e045785f686d37b4f5f849
|
|
This adds a new make_unique_xstrndup function, which is the "n"
analogue of make_unique_xstrdup. It also updates a couple existing
places to use this function.
|
|
PR 28065 (gdb.threads/access-mem-running-thread-exit.exp intermittent
failure) shows that GDB can hit an unexpected scenario -- it can
happen that the kernel manages to open a /proc/PID/task/LWP/mem file,
but then reading from the file returns 0/EOF, even though the process
hasn't exited or execed.
"0" out of read/write is normally what you get when the address space
of the process the file was open for is gone, because the process
execed or exited. So when GDB gets the 0, it returns memory access
failure. In the bad case in question, the process hasn't execed or
exited, so GDB fails a memory access when the access should have
worked.
GDB has code in place to gracefully handle the case of opening the
/proc/PID/task/LWP/mem just while the LWP is exiting -- most often the
open fails with EACCES or ENOENT. When it happens, GDB just tries
opening the file for a different thread of the process. The testcase
is written such that it stresses GDB's logic of closing/reopening the
/proc/PID/task/LWP/mem file, by constantly spawning short lived
threads.
However, there's a window where the kernel manages to find the thread,
but the thread exits just after and clears its address space pointer.
In this case, the kernel creates a file successfully, but the file
ends up with no address space associated, so a subsequent read/write
returns 0/EOF too, just like if the whole process had execed or
exited. This is the case in question that GDB does not handle.
Oleg Nesterov gave this suggestion as workaround for that race:
gdb can open(/proc/pid/mem) and then read (say) /proc/pid/statm.
If statm reports something non-zero, then open() was "successfull".
I think that might work. However, I didn't try it, because I realized
we have another nasty race that that wouldn't fix.
The other race I realized is that because we close/reopen the
/proc/PID/task/LWP/mem file when GDB switches to a different inferior,
then it can happen that GDB reopens /proc/PID/task/LWP/mem just after
a thread execs, and before GDB has seen the corresponding exec event.
I.e., we can open a /proc/PID/task/LWP/mem file accessing the
post-exec address space thinking we're accessing the pre-exec address
space.
A few months back, Simon, Oleg and I discussed a similar race:
[Bug gdb/26754] Race condition when resuming threads and one does an exec
https://sourceware.org/bugzilla/show_bug.cgi?id=26754
The solution back then was to make the kernel fail any ptrace
operation until the exec event is consumed, with this kernel commit:
commit dbb5afad100a828c97e012c6106566d99f041db6
Author: Oleg Nesterov <oleg@redhat.com>
AuthorDate: Wed May 12 15:33:08 2021 +0200
Commit: Linus Torvalds <torvalds@linux-foundation.org>
CommitDate: Wed May 12 10:45:22 2021 -0700
ptrace: make ptrace() fail if the tracee changed its pid unexpectedly
This however, only applies to ptrace, not to the /proc/pid/mem file
opening case. Also, even if it did apply to the file open case, we
would want to support current kernels until such a fix is more wide
spread anyhow.
So all in all, this commit gives up on the idea of only ever keeping
one /proc/pid/mem file descriptor open. Instead, make GDB open a
/proc/pid/mem per inferior, and keep it open until the inferior exits,
is detached or execs. Make GDB open the file right after the inferior
is created or is attached to or forks, at which point we know the
inferior is stable and stopped and isn't thus going to exec, or have a
thread exit, and so the file open won't fail (unless the whole process
is SIGKILLed from outside GDB, at which point it doesn't matter
whether we open the file).
This way, we avoid both races described above, at the expense of using
more file descriptors (one per inferior).
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=28065
Change-Id: Iff943b95126d0f98a7973a07e989e4f020c29419
|
|
Replaces a hard coded line number with a use of gdb_get_line_number.
I suspect that the line number has, over time, come adrift from where
it was supposed to be stopping. When the test was first added, line
770 pointed at the final 'return 0' in function main. Over time, as
things have been added, line 770 now points at some random location in
the middle of main.
So, I've marked the 'return 0' with a comment, and now the test will
always stop there.
I also removed an old comment from 1997 talking about how these tests
will only pass with the HP compiler, followed by an additional comment
from 2000 saying that the tests now pass with GCC.
I get the same results before and after this change.
|
|
The "set index-cache" command is used at the same time as a prefix
command (prefix for "set index-cache directory", for example), and a
boolean setting for turning the index-cache on and off. Even though I
did introduce that, I now don't think it's a good idea to do something
non-standard like this.
First, there's no dedicated CLI command to show whether the index-cache
is enabled, so it has to be custom output in the "show index-cache
handler". Also, it means there's no good way a MI frontend can find out
if the index-cache is enabled. "-gdb-show index-cache" doesn't show it
in the MI output record:
(gdb) interpreter-exec mi "-gdb-show index-cache"
~"\n"
~"The index cache is currently disabled.\n"
^done,showlist={option={name="directory",value="/home/simark/.cache/gdb"}}
Fix this by introducing "set/show index-cache enabled on/off", regular
boolean setting commands. Keep commands "set index-cache on" and "set
index-cache off" as deprecated aliases of "set index-cache enabled",
with respectively the default arguments "on" and "off".
Update tests using "set index-cache on/off" to use the new command.
Update the regexps in gdb.base/maint.exp to figure out whether the
index-cache is enabled or not. Update the doc to mention the new
commands.
Change-Id: I7d5aaaf7fd22bf47bd03e0023ef4fbb4023b37b3
|
|
The getter and setter in struct setting always receive and return values
by const reference. This is not necessary for scalar values (like bool
and int), but more importantly it makes it a bit annoying to write a
getter, you have to use a scratch static variable or something similar
that you can refer to:
const bool &
my_getter ()
{
static bool value;
value = function_returning_bool ();
return value;
}
Change the getter and setter function signatures to receive and return
value by value instead of by reference, when the underlying data type is
scalar. This means that string-based settings will still use
references, but all others will be by value. The getter above would
then be re-written as:
bool
my_getter ()
{
return function_returning_bool ();
}
This is useful for a patch later in this series that defines a boolean
setting with a getter and a setter.
Change-Id: Ieca3a2419fcdb75a6f75948b2c920b548a0af0fd
|
|
The class_deprecated enumerator isn't assigned anywhere, so remove it.
Commands that are deprecated have cmd_list_element::cmd_deprecated set
instead.
Change-Id: Ib35e540915c52aa65f13bfe9b8e4e22e6007903c
|
|
Remove two unnecessary nullptr checks. If aliases is nullptr, then the
for loops will simply be skipped.
Change-Id: I9132063bb17798391f8d019af305383fa8e0229f
|
|
When building gnu-nat.c, we get:
CXX gnu-nat.o
gnu-nat.c: In member function 'virtual void gnu_nat_target::create_inferior(const char*, const string&, char**, int)':
gnu-nat.c:2117:13: error: 'struct inf' has no member named 'target_is_pushed'
2117 | if (!inf->target_is_pushed (this))
| ^~~~~~~~~~~~~~~~
gnu-nat.c:2118:10: error: 'struct inf' has no member named 'push_target'
2118 | inf->push_target (this);
| ^~~~~~~~~~~
This is because of a confusion between the generic `struct inferior`
variable and the gnu-nat-specific `struct inf` variable. Fix by
referring to `inferior`, not `inf`.
Adjust the comment on top of `struct inf` to clarify the purpose of that
type.
Co-Authored-By: Andrea Monaco <andrea.monaco@autistici.org>
Change-Id: I2fe2f7f6ef61a38d79860fd262b08835c963fc77
|
|
We see some additional failures when running the testsuite against a GDB
compiled with ASan, compared to a GDB compiled without ASan. Some of
them are caused by the memory leak report shown by the GDB process when
it exits, and the fact that it makes it exit with a non-zero exit code.
I generally try to remember to set ASAN_OPTIONS=detect_leaks=0 in my
environment when running the tests, but I don't always do it. I think
it would be nice if the testsuite did it. I don't see any use to have
leak detection when running the tests. That is, unless we ever have a
test that ensures GDB doesn't leak memory, which isn't going to happen
any time soon.
Here are some tests I found that were affected by this:
gdb.base/batch-exit-status.exp
gdb.base/many-headers.exp
gdb.base/quit.exp
gdb.base/with-mf.exp
gdb.dwarf2/gdb-add-index.exp
gdb.dwarf2/gdb-add-index-symlink.exp
gdb.dwarf2/imported-unit-runto-main.exp
Change-Id: I784c7df8a13979eb96587f735c1d33ba2cc6e0ca
|
|
The current register set selection mechanism for AArch64 is static, based
on a pre-populated array of register sets.
This means that we might potentially probe register sets that are not
available. This is OK if the kernel errors out during ptrace, but probing the
tag_ctl register, for example, does not result in a ptrace error if the kernel
supports the tagged address ABI but not MTE (PR 28355).
Making the register set selection dynamic, based on feature checks, solves
this and simplifies the code a bit. It allows us to list all of the register
sets only once, and pick and choose based on HWCAP/HWCAP2 or other properties.
I plan to backport this fix to GDB 11 as well.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=28355
|
|
I ran into a case where a breakpoint on _exit never triggered, because it was
set past the end of the _exit prologue, past the end of the exit_group system
call (which does not return).
More concretely, the breakpoint was set at the last insn show here:
...
Dump of assembler code for function _exit:
0x00007ffff7e42ea0 <+0>: 12 00 4c 3c addis r2,r12,18
0x00007ffff7e42ea4 <+4>: 60 43 42 38 addi r2,r2,17248
0x00007ffff7e42ea8 <+8>: 00 00 00 60 nop
0x00007ffff7e42eac <+12>: f8 ff e1 fb std r31,-8(r1)
0x00007ffff7e42eb0 <+16>: 78 1b 7f 7c mr r31,r3
0x00007ffff7e42eb4 <+20>: f0 ff c1 fb std r30,-16(r1)
0x00007ffff7e42eb8 <+24>: ea 00 00 38 li r0,234
0x00007ffff7e42ebc <+28>: a0 8b 22 e9 ld r9,-29792(r2)
0x00007ffff7e42ec0 <+32>: 78 fb e3 7f mr r3,r31
0x00007ffff7e42ec4 <+36>: 14 6a c9 7f add r30,r9,r13
0x00007ffff7e42ec8 <+40>: 02 00 00 44 sc
0x00007ffff7e42ecc <+44>: 26 00 00 7c mfcr r0
0x00007ffff7e42ed0 <+48>: 00 10 09 74 andis. r9,r0,4096
...
Fix this by treating system calls the same as branches in skip_prologue:
by default, don't skip, such that the breakpoint is set at 0x00007ffff7e42eb8
instead.
Tested on ppc64le-linux, on a power 8 machine.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=28527
|
|
On powerpc64le-linux, with test-case gdb.arch/powerpc-addpcis.exp I run into
SIGILL:
...
(gdb) PASS: gdb.arch/powerpc-addpcis.exp: get hexadecimal valueof "$r3"
stepi^M
^M
Program terminated with signal SIGILL, Illegal instruction.^M
The program no longer exists.^M
(gdb) PASS: gdb.arch/powerpc-addpcis.exp: set r4
...
because it's a power9 insn, and I'm running on a power8 machine.
Fix this by handling the SIGILL. Likewise in gdb.arch/powerpc-lnia.exp.
Tested on powerpc64le-linux.
|
|
gdb:
* MAINTAINERS (Global Maintainers): Update my address.
(Responsible Maintainers): Likewise.
(Write After Approval): Likewise.
sim:
* MAINTAINERS (Global Maintainers): Update my address.
|
|
When running test-case gdb.base/step-indirect-call-thunk.exp with target board
unix/-m32/-fPIE/-pie, I run into:
...
(gdb) stepi^M
0x5655552e 22 { /* inc.1 */^M
(gdb) stepi^M
0x56555530 22 { /* inc.1 */^M
(gdb) stepi^M
0x565555f7 in __x86.get_pc_thunk.ax ()^M
(gdb) FAIL: gdb.base/step-indirect-call-thunk.exp: stepi into return thunk
...
In contrast, with unix/-m32 we have instead:
...
(gdb) stepi^M
0x08048407 22 { /* inc.1 */^M
(gdb) stepi^M
23 return x + 1; /* inc.2 */^M
(gdb) stepi^M
0x0804840c 23 return x + 1; /* inc.2 */^M
(gdb) stepi^M
24 } /* inc.3 */^M
(gdb) stepi^M
0x08048410 24 } /* inc.3 */^M
(gdb) stepi^M
0x0804848f in __x86_return_thunk ()^M
(gdb) PASS: gdb.base/step-indirect-call-thunk.exp: stepi into return thunk
...
The test-case doesn't expect to run into __x86.get_pc_thunk.ax, which is a
PIC helper function for x86_64-linux.
Fix this by insn-stepping through it.
Likewise in a few other test-cases.
Tested on x86_64-linux.
|
|
Updated manpages to be consistent with help information provided by the
binary. The main changes are:
* Making all long-form options have '--', instead of a single '-';
* added most of the missing options to the manpage;
* removed the information about using '+' instead of '-', since it
doesn't seem to be supported anymore.
This also fixes 2 upstream bugs:
* https://sourceware.org/bugzilla/show_bug.cgi?id=23965; by adding
--args to the manpage
* https://sourceware.org/bugzilla/show_bug.cgi?id=10619; by adding the
double dashes
|
|
On my branch that replaces the DWARF psymtab reader,
dw2-stack-boundary.exp started failing. However, when I look at the
output in gdb.log, it is correct:
file /home/tromey/gdb/build/gdb/testsuite/outputs/gdb.dwarf2/dw2-stack-boundary/dw2-stack-boundary
Reading symbols from /home/tromey/gdb/build/gdb/testsuite/outputs/gdb.dwarf2/dw2-stack-boundary/dw2-stack-boundary...
During symbol reading: location description stack overflow
During symbol reading: location description stack underflow
What happens to cause the failure is that the two branches in
gdb_test_multiple appear in this order:
-re "\r\nDuring symbol reading: location description stack underflow" {
[...]
-re "\r\nDuring symbol reading: location description stack overflow" {
The first one will match the above, without causing the second one to
ever match -- leading to a spurious failure.
Anchoring the regexps seems to fix the problem, and works for the
current gdb as well.
|
|
The 'set debuginfod' change caused a regression in unittest.exp:
Running selftest help_doc_invariants.
help doc broken invariant: command 'info set debuginfod' help doc first line is not terminated with a '.' character
help doc broken invariant: command 'set debuginfod' help doc first line is not terminated with a '.' character
help doc broken invariant: command 'show debuginfod' help doc first line is not terminated with a '.' character
Self test failed: self-test failed at ../../binutils-gdb/gdb/unittests/command-def-selftests.c:100
This patch fixes the problem. I'm checking it in.
|
|
On a machine with gcc 11, I get:
FAIL: gdb.gdb/unittest.exp: test_completion: tab complete "maintenance selftest string_v" (second tab) (timeout)
FAIL: gdb.gdb/unittest.exp: test_completion: tab complete "maintenance selftest string_vie" (timeout)
That's because when compiling with C++ >= 17, we use the standard
version of string_view, and don't have a selftest for it. So the list
of selftests shown by the tab completion when completing "string_v"
differs.
Change the test to use the copy_* tests instead.
Change-Id: I85f6aa44ee5fc9652b9bd4451e0506b89773526b
|
|
Add section describing GDB's usage of debuginfod.
Refer to this new section in the description of the '--with-debuginfod'
configure option.
Mention debuginfod in the 'Separate Debug Files' section.
|
|
Add 'set debuginfod' command. Accepts 'on', 'off' or 'ask' as an
argument. 'on' enables debuginfod for the current session. 'off'
disables debuginfod for the current session. 'ask' will prompt
the user to either enable or disable debuginfod when the next query
is about to be performed:
This GDB supports auto-downloading debuginfo from the following URLs:
<URL1> <URL2> ...
Enable debuginfod for this session? (y or [n]) y
Debuginfod has been enabled.
To make this setting permanent, add 'set debuginfod on' to .gdbinit.
For interactive sessions, 'ask' is the default. For non-interactive
sessions, 'off' is the default.
Add 'show debuginfod status' command. Displays whether debuginfod
is set to 'on', 'off' or 'ask'.
Add 'set/show debuginfod urls' commands. Accepts a string of
space-separated debuginfod server URLs to be queried. The default
value is copied from the DEBUGINFOD_URLS environment variable.
Finally add 'set/show debuginfod verbose' commands to control whether
debuginfod-related output is displayed. Verbose output is enabled
by default.
(gdb) run
Starting program: /bin/sleep 5
Download failed: No route to host. Continuing without debug info for /lib64/libc.so.6.
If GDB is not built with debuginfod then these commands will just display
Support for debuginfod is not compiled into GDB.
|
|
Remove TYPE_FIELD_DWARF_BLOCK, replace with type::field +
field::loc_dwarf_block.
Change-Id: I10af9410bb5f46d342b8358a7956998c7e804b64
|
|
Remove TYPE_FIELD_STATIC_PHYSADDR replace with type::field +
field::loc_physaddr.
Change-Id: Ica9bc4a48f34750ec82ec86c298d3ecece81bcbd
|
|
Remove TYPE_FIELD_STATIC_PHYSNAME, replace with type::field +
field::loc_physname.
Change-Id: Ie35d446b67dd1d02f39998b406001bdb7e6d5abb
|
|
Remove TYPE_FIELD_ENUMVAL, replace with type::field +
field::loc_enumval.
Change-Id: I2ada73e4635aad3363ce2eb22c1dc52698ee2072
|
|
Remove TYPE_FIELD_BITPOS, replace its uses with type::field +
field::loc_bitpos.
Change-Id: Iccd8d5a77e5352843a837babaa6bd284162e0320
|
|
Remove TYPE_FIELD_LOC_KIND, replace its uses with type::field +
field::loc_kind.
Change-Id: Ib124a26365df82ac1d23df7962d954192913bd90
|
|
Remove FIELD_DWARF_BLOCK, replace its uses with field::loc_dwarf_block.
Change-Id: I66b7d6a960cb5e341e61e21bd3cc9a6ac26de6a8
|
|
Remove FIELD_LOC_KIND_PHYSADDR, replace its uses with
field::loc_physaddr.
Change-Id: Ifd8b2bdaad75f42bfb1404ef8c396ffe7e10ac55
|
|
Remove FIELD_STATIC_PHYSNAME, replace its uses with field::loc_physname.
Change-Id: Iaa8952410403b4eb5bbd68411feea27e2405d657
|
