[Bug gdb/27710] New: remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[Bug gdb/27710] New: remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

            Bug ID: 27710
           Summary: remote-notif.c:113: internal-error: void
                    remote_async_get_pending_events_handler(gdb_client_dat
                    a): Assertion `target_is_non_stop_p ()' failed
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: gdb
          Assignee: unassigned at sourceware dot org
          Reporter: vries at gcc dot gnu.org
  Target Milestone: ---

With master I ran into:
...
PASS: gdb.multi/multi-target-continue.exp: inferior 2
PASS: gdb.multi/multi-target-continue.exp: inferior 5
WARNING: Timed out waiting for EOF in server after monitor exit
...

In more detail:
...
(gdb) PASS: gdb.multi/multi-target-continue.exp: inferior 5
Remote debugging from host ::1, port 49990^M
Process
/home/vries/gdb_versions/devel/build/gdb/testsuite/outputs/gdb.multi/multi-target-continue/multi-target-continue
created; pid = 31241^M
/home/vries/gdb_versions/devel/src/gdb/remote-notif.c:113: internal-error: void
remote_async_get_pending_events_handler(gdb_client_data): Assertion
`target_is_non_stop_p ()' failed.^M
A problem internal to GDB has been detected,^M
further debugging may prove unreliable.^M
Quit this debugging session? (y or n) inferior 5^M
Please answer y or n.^M
/home/vries/gdb_versions/devel/src/gdb/remote-notif.c:113: internal-error: void
remote_async_get_pending_events_handler(gdb_client_data): Assertion
`target_is_non_stop_p ()' failed.^M
A problem internal to GDB has been detected,^M
further debugging may prove unreliable.^M
Quit this debugging session? (y or n) monitor exit^M
Please answer y or n.^M
...

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #1 from Tom de Vries <vries at gcc dot gnu.org> ---
There is a possibility that is related to a change in configuration I just did:
--with-included-regex (for PR27681 investigation), so there's a possibility
that this is triggered by this, that needs to be investigated.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #2 from Tom de Vries <vries at gcc dot gnu.org> ---
Not reproducible sofar...

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

Simon Marchi <simark at simark dot ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |simark at simark dot ca

--- Comment #3 from Simon Marchi <simark at simark dot ca> ---
(In reply to Tom de Vries from comment #1)
> There is a possibility that is related to a change in configuration I just
> did: --with-included-regex (for PR27681 investigation), so there's a
> possibility that this is triggered by this, that needs to be investigated.

It would be surprising for the regex library to end up screwing this.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #4 from Tom de Vries <vries at gcc dot gnu.org> ---
Created attachment 13366
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13366&action=edit
core.gz

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #5 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #4)
> Created attachment 13366 [details]
> core.gz

Matching gdb can be found here:
https://ftp.suse.com/pub/people/tdevries/gdb/pr27710/

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #6 from Tom de Vries <vries at gcc dot gnu.org> ---
Backtrace:
...
(gdb) bt
#0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
#1  0x00007fd1b7b43a01 in __GI_abort () at abort.c:79
#2  0x0000000000b1db10 in dump_core () at
/home/vries/gdb_versions/devel/src/gdb/utils.c:204
#3  0x0000000000b1e022 in internal_vproblem(internal_problem *, const char *,
int, const char *, typedef __va_list_tag __va_list_tag *) (problem=0x16d7bc0
<internal_error_problem>, 
    file=0xf4c5b0 "/home/vries/gdb_versions/devel/src/gdb/remote-notif.c",
line=113, 
    fmt=0xf4c584 "%s: Assertion `%s' failed.", ap=0x7ffe5cb2a788)
    at /home/vries/gdb_versions/devel/src/gdb/utils.c:414
#4  0x0000000000b1e0ea in internal_verror (
    file=0xf4c5b0 "/home/vries/gdb_versions/devel/src/gdb/remote-notif.c",
line=113, 
    fmt=0xf4c584 "%s: Assertion `%s' failed.", ap=0x7ffe5cb2a788)
    at /home/vries/gdb_versions/devel/src/gdb/utils.c:439
#5  0x0000000000df5b79 in internal_error (
    file=0xf4c5b0 "/home/vries/gdb_versions/devel/src/gdb/remote-notif.c",
line=113, 
    fmt=0xf4c584 "%s: Assertion `%s' failed.")
    at /home/vries/gdb_versions/devel/src/gdbsupport/errors.cc:55
#6  0x000000000094be95 in remote_async_get_pending_events_handler
(data=0x40e34b0)
    at /home/vries/gdb_versions/devel/src/gdb/remote-notif.c:113
#7  0x00000000004d29e5 in check_async_event_handlers ()
    at /home/vries/gdb_versions/devel/src/gdb/async-event.c:335
#8  0x0000000000df5cb7 in gdb_do_one_event ()
    at /home/vries/gdb_versions/devel/src/gdbsupport/event-loop.cc:216
#9  0x00000000007fa045 in start_event_loop ()
    at /home/vries/gdb_versions/devel/src/gdb/main.c:348
#10 0x00000000007fa165 in captured_command_loop ()
    at /home/vries/gdb_versions/devel/src/gdb/main.c:408
#11 0x00000000007fb846 in captured_main (data=0x7ffe5cb2a9d0)
    at /home/vries/gdb_versions/devel/src/gdb/main.c:1242
#12 0x00000000007fb8ac in gdb_main (args=0x7ffe5cb2a9d0)
    at /home/vries/gdb_versions/devel/src/gdb/main.c:1257
#13 0x00000000004187ee in main (argc=15, argv=0x7ffe5cb2aad8)
    at /home/vries/gdb_versions/devel/src/gdb/gdb.c:32
...

So, we hit the assert:
...
  gdb_assert (target_is_non_stop_p ());
...

The function is defined as:
...
bool
target_is_non_stop_p ()
{
  return ((non_stop
           || target_non_stop_enabled == AUTO_BOOLEAN_TRUE
           || (target_non_stop_enabled == AUTO_BOOLEAN_AUTO
               && target_always_non_stop_p ()))
          && target_can_async_p ());
}
...
and:
...
(gdb) p non_stop
$1 = true
...
so presumably the assert fails because target_can_async_p () == false.

We have:
...
bool
target_can_async_p ()
{
  return current_inferior ()->top_target ()->can_async_p ();
}
...

AFAIU, current_inferior ()->top_target () translates to:
...
(gdb) p
current_inferior_.m_obj.m_target_stack.m_stack[current_inferior_.m_obj.m_target_stack.m_top]
$16 = (target_ops *) 0x16c6cd0 <exec_ops>
...

Exec_ops has type exec_target, which subclasses target_ops without implementing
a can_async_p override, so we have the default:
...
    virtual bool can_async_p ()
      TARGET_DEFAULT_RETURN (false);
...

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #7 from Tom de Vries <vries at gcc dot gnu.org> ---
OTOH, we have:
...
(gdb) p *notif_state.remote
$33 = {<process_stratum_target> = {<target_ops> = {<refcounted_object> =
{m_refcount = 1}, 
      _vptr.target_ops = 0xf54570 <vtable for extended_remote_target+16>}, 
    threads_executing = false, connection_number = 4, commit_resumed_state =
false}, 
...

So, is the problem the assert?  That is, it checks something about the current
inferior, but instead it should check something about the inferior related to
notif_state.remote ?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #8 from Simon Marchi <simark at simark dot ca> ---
(In reply to Tom de Vries from comment #5)
> (In reply to Tom de Vries from comment #4)
> > Created attachment 13366 [details]
> > core.gz
> 
> Matching gdb can be found here:
> https://ftp.suse.com/pub/people/tdevries/gdb/pr27710/

What sha1 is that build of GDB?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #9 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Simon Marchi from comment #8)
> (In reply to Tom de Vries from comment #5)
> > (In reply to Tom de Vries from comment #4)
> > > Created attachment 13366 [details]
> > > core.gz
> > 
> > Matching gdb can be found here:
> > https://ftp.suse.com/pub/people/tdevries/gdb/pr27710/
> 
> What sha1 is that build of GDB?

$ file gdb
gdb: ELF 64-bit LSB executable, x86-64, version 1 (GNU/Linux), dynamically
linked, interpreter /lib64/ld-linux-x86-64.so.2,
BuildID[sha1]=dd1d8ec51a4a365d2935ede260768b506dc7a49b, for GNU/Linux 3.2.0,
with debug_info, not stripped

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #10 from Simon Marchi <simark at simark dot ca> ---
(In reply to Tom de Vries from comment #5)
> We have:
> ...
> bool
> target_can_async_p ()
> {
>   return current_inferior ()->top_target ()->can_async_p ();
> }
> ...
> 
> AFAIU, current_inferior ()->top_target () translates to:
> ...
> (gdb) p
> current_inferior_.m_obj.m_target_stack.m_stack[current_inferior_.m_obj.
> m_target_stack.m_top]
> $16 = (target_ops *) 0x16c6cd0 <exec_ops>
> ...
> 
> Exec_ops has type exec_target, which subclasses target_ops without
> implementing a can_async_p override, so we have the default:
> ...
>     virtual bool can_async_p ()
>       TARGET_DEFAULT_RETURN (false);
> ...

Since the test is related to multi-inferior, my first question would be, is the
current inferior the right one (one that belongs to the remote target receiving
the notification)?

We have the remote target in notif_state->remote_target, so maybe we need to
switch to it before calling target_is_non_stop_p.  From a quick look, it
doesn't seem like remote_notif_process cares about the current inferior though.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #11 from Simon Marchi <simark at simark dot ca> ---
(In reply to Tom de Vries from comment #9)
> (In reply to Simon Marchi from comment #8)
> > (In reply to Tom de Vries from comment #5)
> > > (In reply to Tom de Vries from comment #4)
> > > > Created attachment 13366 [details]
> > > > core.gz
> > > 
> > > Matching gdb can be found here:
> > > https://ftp.suse.com/pub/people/tdevries/gdb/pr27710/
> > 
> > What sha1 is that build of GDB?
> 
> $ file gdb
> gdb: ELF 64-bit LSB executable, x86-64, version 1 (GNU/Linux), dynamically
> linked, interpreter /lib64/ld-linux-x86-64.so.2,
> BuildID[sha1]=dd1d8ec51a4a365d2935ede260768b506dc7a49b, for GNU/Linux 3.2.0,
> with debug_info, not stripped

Haha sorry, I meant the git hash, to be able to checkout the corresponding
source.

Anyway, I am not able to load the core and get something meaningful, I would
probably need the libraries from your environment.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #12 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Simon Marchi from comment #11)
> (In reply to Tom de Vries from comment #9)
> > (In reply to Simon Marchi from comment #8)
> > > (In reply to Tom de Vries from comment #5)
> > > > (In reply to Tom de Vries from comment #4)
> > > > > Created attachment 13366 [details]
> > > > > core.gz
> > > > 
> > > > Matching gdb can be found here:
> > > > https://ftp.suse.com/pub/people/tdevries/gdb/pr27710/
> > > 
> > > What sha1 is that build of GDB?
> > 
> > $ file gdb
> > gdb: ELF 64-bit LSB executable, x86-64, version 1 (GNU/Linux), dynamically
> > linked, interpreter /lib64/ld-linux-x86-64.so.2,
> > BuildID[sha1]=dd1d8ec51a4a365d2935ede260768b506dc7a49b, for GNU/Linux 3.2.0,
> > with debug_info, not stripped
> 
> Haha sorry, I meant the git hash, to be able to checkout the corresponding
> source.
> 

Ah, that makes sense, I was wondering what you wanted with the buildid :)

I didn't keep track of the commit hash, but the sources have at least this date
stamp:
...
$ ./gdb -v
GNU gdb (GDB) 11.0.50.20210414-git
...

> Anyway, I am not able to load the core and get something meaningful, I would
> probably need the libraries from your environment.

Hmm, that's too bad.  FWIW, the environemnt is openSUSE Leap 15.2, so in that
context it should work.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #13 from Tom de Vries <vries at gcc dot gnu.org> ---
Created attachment 13368
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13368&action=edit
debug patch

I wrote this debug patch and ran the test-case, and got:
...
$ grep INFERIOR gdb.log  
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 2
(gdb) CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 5
(gdb) CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE: 5
(gdb) CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
(gdb) CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
(gdb) CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
(gdb) CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
(gdb) CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
(gdb) CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 2
(gdb) CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 5
Delete all breakpoints? (y or n) CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE:
2
(gdb) CURRENT INFERIOR: 1, INFERIOR FOR NOTIF_STATE: 5
Delete all breakpoints? (y or n) CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE:
5
(gdb) CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE: 5
Delete all breakpoints? (y or n) CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE:
2
(gdb) CURRENT INFERIOR: 4, INFERIOR FOR NOTIF_STATE: 5
CURRENT INFERIOR: 5, INFERIOR FOR NOTIF_STATE: 2
CURRENT INFERIOR: 2, INFERIOR FOR NOTIF_STATE: 5
...

It seems rather common that the current inferior and the inferior related to
notif_state are different.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #14 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Simon Marchi from comment #10)
> (In reply to Tom de Vries from comment #5)
> > We have:
> > ...
> > bool
> > target_can_async_p ()
> > {
> >   return current_inferior ()->top_target ()->can_async_p ();
> > }
> > ...
> > 
> > AFAIU, current_inferior ()->top_target () translates to:
> > ...
> > (gdb) p
> > current_inferior_.m_obj.m_target_stack.m_stack[current_inferior_.m_obj.
> > m_target_stack.m_top]
> > $16 = (target_ops *) 0x16c6cd0 <exec_ops>
> > ...
> > 
> > Exec_ops has type exec_target, which subclasses target_ops without
> > implementing a can_async_p override, so we have the default:
> > ...
> >     virtual bool can_async_p ()
> >       TARGET_DEFAULT_RETURN (false);
> > ...
> 
> Since the test is related to multi-inferior, my first question would be, is
> the current inferior the right one (one that belongs to the remote target
> receiving the notification)?
> 
> We have the remote target in notif_state->remote_target, so maybe we need to
> switch to it before calling target_is_non_stop_p.  From a quick look, it
> doesn't seem like remote_notif_process cares about the current inferior
> though.

I just stumbled upon remote_unpush_target, and it seems one remote_target can
be associated with multiple inferiors.  So at this point I'm not sure how one
would select the "right" inferior.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #15 from Simon Marchi <simark at simark dot ca> ---
(In reply to Tom de Vries from comment #14)
> I just stumbled upon remote_unpush_target, and it seems one remote_target
> can be associated with multiple inferiors.  So at this point I'm not sure
> how one would select the "right" inferior.

Indeed, one remote target can handle multiple inferiors (if using
extended-remote).
The right inferior in this case would be any inferior of that target.  Making
any inferior of this remote target the current inferior implicitly makes this
remote target the current one, such that target_is_non_stop_p will hit that
remote target.  If the current inferior is one that does not belong to that
remote target, then target_is_non_stop_p is really asking whether some other
unrelated target is non-stop.

So adding a call to switch_to_target_no_thread should make it work (it just
makes an arbitrary inferior of the target the current one).

Another option, instead of switching the current inferior, would be to make
target_is_non_stop accept a target:

  gdb_assert (target_is_non_stop_p (notif_state->remote));

And pass down the target to everything below.  It would be a bit more work, and
we would have to make sure everything called by target_is_non_stop_p doesn't
use the global context.

This looks like a problem that happens quite rarely though, it would be nice to
have a way to reproduce it all the time, by adding some delay somewhere or
otherwise.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[cbiesinger at google dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

Christian Biesinger <cbiesinger at google dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |cbiesinger at google dot com

--- Comment #16 from Christian Biesinger <cbiesinger at google dot com> ---
It seems like target_is_non_stop_p should really be a method on target?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #17 from Simon Marchi <simark at simark dot ca> ---
(In reply to Christian Biesinger from comment #16)
> It seems like target_is_non_stop_p should really be a method on target?

Indeed, it could be (a non-virtual method).

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #18 from Tom de Vries <vries at gcc dot gnu.org> ---
Created attachment 13369
  --> https://sourceware.org/bugzilla/attachment.cgi?id=13369&action=edit
Tentative patch

(In reply to Simon Marchi from comment #15)
> So adding a call to switch_to_target_no_thread should make it work (it just
> makes an arbitrary inferior of the target the current one).
> 

This patch implements that approach.  I've done a build and reg-test on
x86_64-linux.  Also I've done a 1000 iteration run of the test-case and the
assert doesn't trigger anymore.

It's rather unfortunate that we need to move remote_target to remote.h for the
dynamic cast.  This more simple solution at least compiles without that change:
...
+  scoped_restore_current_thread restore_thread;
+  process_stratum_target *remote
+    = (process_stratum_target *) notif_state->remote;
+  switch_to_target_no_thread (remote);
...
but I'm not sure how legal that is given that here remote_target is an
incomplete class.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug gdb/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[simark at simark dot ca]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #19 from Simon Marchi <simark at simark dot ca> ---
I don't really understand this supposed separation of concerns between remote.c
and remote-notif.c.  They both know about each other: remote_target (well,
remote_state) uses remote_notif_state, and remote_notif_state (and
notif_client) only works with remote_target as its user.  Anyway.

To avoid all this churn, I'd be fine with a new util function
"remote_target_is_non_stop", implemented in remote.c, like:

bool
remote_target_is_non_stop_p (remote_target *t)
{
  scoped_restore_current_thread restore_thread;
  switch_to_target_no_thread (t);

  return target_is_non_stop_p ();
}

You could then use

gdb_assert (remote_target_is_non_stop_p (remote_target))

in remote-notif.c.

If / when target_is_non_stop_p becomes a method on target_ops and hopefully
independent from the current context, it could be simplified to:

bool
remote_target_is_non_stop_p (remote_target *t)
{
  return t->is_non_stop_p ();
}

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug remote/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|gdb                         |remote

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug remote/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #20 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Simon Marchi from comment #19)
> I don't really understand this supposed separation of concerns between
> remote.c and remote-notif.c.  They both know about each other: remote_target
> (well, remote_state) uses remote_notif_state, and remote_notif_state (and
> notif_client) only works with remote_target as its user.  Anyway.
> 
> To avoid all this churn, I'd be fine with a new util function
> "remote_target_is_non_stop", implemented in remote.c, like:
> 
> bool
> remote_target_is_non_stop_p (remote_target *t)
> {
>   scoped_restore_current_thread restore_thread;
>   switch_to_target_no_thread (t);
> 
>   return target_is_non_stop_p ();
> }
> 
> You could then use
> 
> gdb_assert (remote_target_is_non_stop_p (remote_target))
> 
> in remote-notif.c.
> 
> If / when target_is_non_stop_p becomes a method on target_ops and hopefully
> independent from the current context, it could be simplified to:
> 
> bool
> remote_target_is_non_stop_p (remote_target *t)
> {
>   return t->is_non_stop_p ();
> }

Submitted as
https://sourceware.org/pipermail/gdb-patches/2021-April/178027.html

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug remote/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[cvs-commit at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

--- Comment #21 from cvs-commit at gcc dot gnu.org <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Tom de Vries <vries@sourceware.org>:

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=c39ebbf43faa932a6860fd9c3607503641659898

commit c39ebbf43faa932a6860fd9c3607503641659898
Author: Tom de Vries <tdevries@suse.de>
Date:   Thu Apr 22 17:01:00 2021 +0200

    [gdb] Fix assert in remote_async_get_pending_events_handler

    Occassionally I run into the following assert:
    ...
    (gdb) PASS: gdb.multi/multi-target-continue.exp: inferior 5
    Remote debugging from host ::1, port 49990^M
    Process multi-target-continue created; pid = 31241^M
    src/gdb/remote-notif.c:113: internal-error: \
      void remote_async_get_pending_events_handler(gdb_client_data): \
      Assertion `target_is_non_stop_p ()' failed.^M
    ...

    The assert checks target_is_non_stop_p, which is related to the current
    target.

    Fix this by changing the assert such that it checks non-stopness related to
    the event it's handling.

    Tested on x86_64-linux.

    gdb/ChangeLog:

    2021-04-22  Simon Marchi  <simon.marchi@polymtl.ca>
                Tom de Vries  <tdevries@suse.de>

            PR remote/27710
            * remote.c (remote_target_is_non_stop_p): New function.
            * remote.h (remote_target_is_non_stop_p): Declare.
            * remote-notif.c (remote_async_get_pending_events_handler): Fix
assert
            to check non-stopness using notif_state->remote rather current
target.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug remote/27710] remote-notif.c:113: internal-error: void remote_async_get_pending_events_handler(gdb_client_data): Assertion `target_is_non_stop_p ()' failed

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27710

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |11.1

--- Comment #22 from Tom de Vries <vries at gcc dot gnu.org> ---
Patch committed, marking resolved-fixed.

-- 
You are receiving this mail because:
You are on the CC list for the bug.