2021-01-01 16:03:39 +08:00
|
|
|
# Copyright 2019-2021 Free Software Foundation, Inc.
|
Switch the inferior too in switch_to_program_space_and_thread
With multi-target, each inferior now has its own target connection.
The problem in switch_to_program_space_and_thread is that in the
current state GDB switches to "no thread" and also sets the program
space but because the inferior is not switched, potentially an
incorrect target remains selected.
Here is a sample scenario that exploits this flow:
On terminal 1, start a gdbserver on a program named foo:
$ gdbserver :1234 ./foo
On terminal 2, start gdb on a program named bar. Suppose foo and bar
are compiled from foo.c and bar.c. They are completely separate. So,
bar.c:2 has no meaning for foo.
$ gdb -q ./bar
Reading symbols from ./bar...
(gdb) add-inferior
[New inferior 2]
Added inferior 2
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) target remote :1234
...
(gdb) set debug remote 2
(gdb) break bar.c:2
Sending packet: $Hgp0.0#ad...Packet received: OK
Sending packet: $m5fa,12#f8...Packet received: E01
Sending packet: $m5fa,1#c6...Packet received: E01
Sending packet: $m5fb,3#c9...Packet received: E01
Sending packet: $m5fe,1#ca...Packet received: E01
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Here we have an unnecessary sending of the packets to the gdbserver.
With this fix in progspace-and-thread.c, we'll get this:
(gdb) break bar.c:2
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Now there is no sending of the packets to gdbserver.
The changes around clear_symtab_users calls are necessary because
otherwise we regress gdb.base/step-over-exit.exp, hitting the new
assertion in switch_to_program_space_and_thread. The problem is, a
forked child terminates, and when GDB decides to auto-purge that
inferior, GDB tries to switch to the pspace of that no-longer-existing
inferior.
The root of the problem is within the program_space destructor:
program_space::~program_space ()
{
...
set_current_program_space (this); # (1)
...
breakpoint_program_space_exit (this); # (2)
...
free_all_objfiles (); # (3)
...
}
We get here from delete_inferior -> delete_program_space.
So we're deleting an inferior, and the inferior to be
deleted is no longer in the inferior list.
At (2), we've deleted all the breakpoints and locations for the
program space being deleted.
The crash happens while doing a breakpoint re-set, called by
clear_symtab_users at the tail end of (3). That is, while recreating
breakpoints for the current program space, which is the program space
we're tearing down. During breakpoint re-set, we try to switch to the
new location's pspace (the current pspace set in (1), so the pspace
we're tearing down) with switch_to_program_space_and_thread, and that
hits the failed assertion. It's the fact that we recreate breakpoints
in the program_space destructor that is the latent bug here. Just
don't do that, and we don't end up in the crash situation.
My first approach to fix this added a symfile_add_flags parameter to
program_space::free_all_objfiles, and then passed that down to
clear_symtab_users. The program_space dtor would then pass down
SYMFILE_DEFER_BP_RESET to free_all_objfiles. I couldn't help feeling
that adding that parameter to free_all_objfiles looked a little
awkward, so I settled on something a little different -- hoist the
clear_symtab_users call to the callers. There are only two callers.
I felt that that didn't look as odd, particularly since
remove_symbol_file_command also does:
objf->unlink ();
clear_symtab_users (0);
I.e., objfile deletion is already separate from calling
clear_symtab_users in some places.
gdb/ChangeLog:
2020-01-10 Aleksandar Paunovic <aleksandar.paunovic@intel.com>
Pedro Alves <palves@redhat.com>
* progspace-and-thread.c (switch_to_program_space_and_thread):
Assert there's an inferior for PSPACE. Use
switch_to_inferior_no_thread to switch the inferior too.
* progspace.c (program_space::~program_space): Call
clear_symtab_users here, with SYMFILE_DEFER_BP_RESET.
(program_space::free_all_objfiles): Don't call clear_symtab_users
here.
* symfile.c (symbol_file_clear): Call clear_symtab_users here.
gdb/testsuite/ChangeLog:
2020-01-10 Pedro Alves <palves@redhat.com>
* gdb.server/bkpt-other-inferior.exp: New file.
2020-01-11 04:06:16 +08:00
|
|
|
|
|
|
|
# This program is free software; you can redistribute it and/or modify
|
|
|
|
# it under the terms of the GNU General Public License as published by
|
|
|
|
# the Free Software Foundation; either version 3 of the License, or
|
|
|
|
# (at your option) any later version.
|
|
|
|
#
|
|
|
|
# This program is distributed in the hope that it will be useful,
|
|
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
# GNU General Public License for more details.
|
|
|
|
#
|
|
|
|
# You should have received a copy of the GNU General Public License
|
|
|
|
# along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
|
|
|
|
# Test that GDB does not access the remote target's memory when
|
|
|
|
# setting a breakpoint on a function that only exists in an inferior
|
|
|
|
# that is not bound to the remote target.
|
|
|
|
|
|
|
|
load_lib gdbserver-support.exp
|
|
|
|
|
|
|
|
standard_testfile server.c
|
|
|
|
|
|
|
|
if { [skip_gdbserver_tests] } {
|
|
|
|
return 0
|
|
|
|
}
|
|
|
|
|
|
|
|
if { [prepare_for_testing "failed to prepare" ${binfile} "${srcfile}" \
|
|
|
|
{debug pthreads}] } {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
|
|
|
|
# Make sure we're disconnected, in case we're testing with an
|
|
|
|
# extended-remote board, therefore already connected.
|
|
|
|
gdb_test "disconnect" ".*"
|
|
|
|
|
|
|
|
# Leave inferior 1 with the exec target, not connected. Add another
|
|
|
|
# inferior, and connect it to gdbserver.
|
|
|
|
|
|
|
|
gdb_test "add-inferior" "Added inferior 2" \
|
|
|
|
"add inferior 2"
|
|
|
|
gdb_test "inferior 2" "Switching to inferior 2.*" \
|
|
|
|
"switch to inferior 2"
|
|
|
|
gdb_test "file ${binfile}" ".*" "load file in inferior 2"
|
|
|
|
|
|
|
|
set target_exec [gdbserver_download_current_prog]
|
|
|
|
|
|
|
|
# Start GDBserver.
|
|
|
|
set res [gdbserver_start "" $target_exec]
|
|
|
|
|
|
|
|
# Connect to GDBserver.
|
|
|
|
set gdbserver_protocol [lindex $res 0]
|
|
|
|
set gdbserver_gdbport [lindex $res 1]
|
|
|
|
gdb_target_cmd $gdbserver_protocol $gdbserver_gdbport
|
|
|
|
|
|
|
|
# Discard any symbol files that we have opened.
|
|
|
|
set test "discard symbol table"
|
|
|
|
gdb_test_multiple "file" $test {
|
|
|
|
-re "A program is being debugged already..*Are you sure you want to change the file.*y or n. $" {
|
|
|
|
gdb_test "y" ".*" $test \
|
|
|
|
{Discard symbol table from `.*'\? \(y or n\) } "y"
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
# At this point:
|
|
|
|
#
|
|
|
|
# - inferior 1 has symbols, and is not connected to any target.
|
|
|
|
# - inferior 2 has no symbols, and is connected to gdbserver.
|
|
|
|
|
|
|
|
# Setting a breakpoint at some function by name should set a
|
|
|
|
# breakpoint on inferior 1, since it has symbols, and should not
|
|
|
|
# result in any access to inferior 2's remote target.
|
|
|
|
|
|
|
|
foreach inf_sel {1 2} {
|
|
|
|
with_test_prefix "inf $inf_sel" {
|
|
|
|
gdb_test "inferior $inf_sel" "Switching to inferior $inf_sel.*" \
|
|
|
|
"switch to inferior"
|
|
|
|
|
Fix bkpt-other-inferior.exp race
When testing with "maint set target-non-stop on",
gdb.server/bkpt-other-inferior.exp sometimes fails like so:
(gdb) inferior 2
[Switching to inferior 2 [process 368191] (<noexec>)]
[Switching to thread 2.1 (Thread 368191.368191)]
[remote] Sending packet: $m7ffff7fd0100,1#5b
[remote] Packet received: 48
[remote] Sending packet: $m7ffff7fd0100,1#5b
[remote] Packet received: 48
[remote] Sending packet: $m7ffff7fd0100,9#63
[remote] Packet received: 4889e7e8e80c000049
#0 0x00007ffff7fd0100 in ?? ()
(gdb) PASS: gdb.server/bkpt-other-inferior.exp: inf 2: switch to inferior
break -q main
Breakpoint 2 at 0x1138: file /home/pedro/gdb/binutils-gdb/src/gdb/testsuite/gdb.server/server.c, line 21.
(gdb) PASS: gdb.server/bkpt-other-inferior.exp: inf 2: set breakpoint
delete breakpoints
Delete all breakpoints? (y or n) y
(gdb) [remote] wait: enter
[remote] wait: exit
FAIL: gdb.server/bkpt-other-inferior.exp: inf 2: delete all breakpoints in delete_breakpoints (timeout)
ERROR: breakpoints not deleted
Remote debugging from host ::1, port 55876
monitor exit
The problem is here:
(gdb) [remote] wait: enter
The testcase isn't expecting any output after the prompt.
Why is that "[remote] wait" output? What happens is that "delete
breakpoints" queries the user, and `query` disables/reenables target
async, which results in the remote target's async event handler ending
up marked:
(top-gdb) bt
#0 mark_async_event_handler (async_handler_ptr=0x556bffffffff) at ../../src/gdb/async-event.c:295
#1 0x0000556bf71b711f in infrun_async (enable=1) at ../../src/gdb/infrun.c:119
#2 0x0000556bf7471387 in target_async (enable=1) at ../../src/gdb/target.c:3684
#3 0x0000556bf748a0bd in gdb_readline_wrapper_cleanup::~gdb_readline_wrapper_cleanup (this=0x7ffe3cf30eb0, __in_chrg=<optimized out>) at ../../src/gdb/top.c:1074
#4 0x0000556bf74874e2 in gdb_readline_wrapper (prompt=0x556bfa17da60 "Delete all breakpoints? (y or n) ") at ../../src/gdb/top.c:1096
#5 0x0000556bf75111c5 in defaulted_query(const char *, char, typedef __va_list_tag __va_list_tag *) (ctlstr=0x556bf7717f34 "Delete all breakpoints? ", defchar=0 '\000', args=0x7ffe3cf31020) at ../../src/gdb/utils.c:893
#6 0x0000556bf751166f in query (ctlstr=0x556bf7717f34 "Delete all breakpoints? ") at ../../src/gdb/utils.c:985
#7 0x0000556bf6f11404 in delete_command (arg=0x0, from_tty=1) at ../../src/gdb/breakpoint.c:13500
...
... which then later results in a target_wait call:
(top-gdb) bt
#0 remote_target::wait_ns (this=0x7ffe3cf30f80, ptid=..., status=0xde530314f0802800, options=...) at ../../src/gdb/remote.c:7937
#1 0x0000556bf7369dcb in remote_target::wait (this=0x556bfa0b2180, ptid=..., status=0x7ffe3cf31568, options=...) at ../../src/gdb/remote.c:8173
#2 0x0000556bf745e527 in target_wait (ptid=..., status=0x7ffe3cf31568, options=...) at ../../src/gdb/target.c:2000
#3 0x0000556bf71be686 in do_target_wait_1 (inf=0x556bfa1573d0, ptid=..., status=0x7ffe3cf31568, options=...) at ../../src/gdb/infrun.c:3463
#4 0x0000556bf71be88b in <lambda(inferior*)>::operator()(inferior *) const (__closure=0x7ffe3cf31320, inf=0x556bfa1573d0) at ../../src/gdb/infrun.c:3526
#5 0x0000556bf71bebcd in do_target_wait (wait_ptid=..., ecs=0x7ffe3cf31540, options=...) at ../../src/gdb/infrun.c:3539
#6 0x0000556bf71bf97b in fetch_inferior_event () at ../../src/gdb/infrun.c:3879
#7 0x0000556bf71a27f8 in inferior_event_handler (event_type=INF_REG_EVENT) at ../../src/gdb/inf-loop.c:42
#8 0x0000556bf71cc8b7 in infrun_async_inferior_event_handler (data=0x0) at ../../src/gdb/infrun.c:9220
#9 0x0000556bf6ecb80f in check_async_event_handlers () at ../../src/gdb/async-event.c:327
#10 0x0000556bf76b011a in gdb_do_one_event () at ../../src/gdbsupport/event-loop.cc:216
...
... which returns TARGET_WAITKIND_IGNORE.
Fix this by only enabling remote output around setting the breakpoint.
gdb/testsuite/ChangeLog:
* gdb.server/bkpt-other-inferior.exp: Only enable remote output
around setting the breakpoint.
Change-Id: I2fd152fd9c46b1c5e7fa678cc4d4054dac0b2bd4
2021-02-14 00:25:26 +08:00
|
|
|
gdb_test_no_output "set debug remote 1"
|
|
|
|
|
Switch the inferior too in switch_to_program_space_and_thread
With multi-target, each inferior now has its own target connection.
The problem in switch_to_program_space_and_thread is that in the
current state GDB switches to "no thread" and also sets the program
space but because the inferior is not switched, potentially an
incorrect target remains selected.
Here is a sample scenario that exploits this flow:
On terminal 1, start a gdbserver on a program named foo:
$ gdbserver :1234 ./foo
On terminal 2, start gdb on a program named bar. Suppose foo and bar
are compiled from foo.c and bar.c. They are completely separate. So,
bar.c:2 has no meaning for foo.
$ gdb -q ./bar
Reading symbols from ./bar...
(gdb) add-inferior
[New inferior 2]
Added inferior 2
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) target remote :1234
...
(gdb) set debug remote 2
(gdb) break bar.c:2
Sending packet: $Hgp0.0#ad...Packet received: OK
Sending packet: $m5fa,12#f8...Packet received: E01
Sending packet: $m5fa,1#c6...Packet received: E01
Sending packet: $m5fb,3#c9...Packet received: E01
Sending packet: $m5fe,1#ca...Packet received: E01
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Here we have an unnecessary sending of the packets to the gdbserver.
With this fix in progspace-and-thread.c, we'll get this:
(gdb) break bar.c:2
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Now there is no sending of the packets to gdbserver.
The changes around clear_symtab_users calls are necessary because
otherwise we regress gdb.base/step-over-exit.exp, hitting the new
assertion in switch_to_program_space_and_thread. The problem is, a
forked child terminates, and when GDB decides to auto-purge that
inferior, GDB tries to switch to the pspace of that no-longer-existing
inferior.
The root of the problem is within the program_space destructor:
program_space::~program_space ()
{
...
set_current_program_space (this); # (1)
...
breakpoint_program_space_exit (this); # (2)
...
free_all_objfiles (); # (3)
...
}
We get here from delete_inferior -> delete_program_space.
So we're deleting an inferior, and the inferior to be
deleted is no longer in the inferior list.
At (2), we've deleted all the breakpoints and locations for the
program space being deleted.
The crash happens while doing a breakpoint re-set, called by
clear_symtab_users at the tail end of (3). That is, while recreating
breakpoints for the current program space, which is the program space
we're tearing down. During breakpoint re-set, we try to switch to the
new location's pspace (the current pspace set in (1), so the pspace
we're tearing down) with switch_to_program_space_and_thread, and that
hits the failed assertion. It's the fact that we recreate breakpoints
in the program_space destructor that is the latent bug here. Just
don't do that, and we don't end up in the crash situation.
My first approach to fix this added a symfile_add_flags parameter to
program_space::free_all_objfiles, and then passed that down to
clear_symtab_users. The program_space dtor would then pass down
SYMFILE_DEFER_BP_RESET to free_all_objfiles. I couldn't help feeling
that adding that parameter to free_all_objfiles looked a little
awkward, so I settled on something a little different -- hoist the
clear_symtab_users call to the callers. There are only two callers.
I felt that that didn't look as odd, particularly since
remove_symbol_file_command also does:
objf->unlink ();
clear_symtab_users (0);
I.e., objfile deletion is already separate from calling
clear_symtab_users in some places.
gdb/ChangeLog:
2020-01-10 Aleksandar Paunovic <aleksandar.paunovic@intel.com>
Pedro Alves <palves@redhat.com>
* progspace-and-thread.c (switch_to_program_space_and_thread):
Assert there's an inferior for PSPACE. Use
switch_to_inferior_no_thread to switch the inferior too.
* progspace.c (program_space::~program_space): Call
clear_symtab_users here, with SYMFILE_DEFER_BP_RESET.
(program_space::free_all_objfiles): Don't call clear_symtab_users
here.
* symfile.c (symbol_file_clear): Call clear_symtab_users here.
gdb/testsuite/ChangeLog:
2020-01-10 Pedro Alves <palves@redhat.com>
* gdb.server/bkpt-other-inferior.exp: New file.
2020-01-11 04:06:16 +08:00
|
|
|
set test "set breakpoint"
|
2020-09-03 22:38:12 +08:00
|
|
|
gdb_test_multiple "break -q main" $test {
|
Switch the inferior too in switch_to_program_space_and_thread
With multi-target, each inferior now has its own target connection.
The problem in switch_to_program_space_and_thread is that in the
current state GDB switches to "no thread" and also sets the program
space but because the inferior is not switched, potentially an
incorrect target remains selected.
Here is a sample scenario that exploits this flow:
On terminal 1, start a gdbserver on a program named foo:
$ gdbserver :1234 ./foo
On terminal 2, start gdb on a program named bar. Suppose foo and bar
are compiled from foo.c and bar.c. They are completely separate. So,
bar.c:2 has no meaning for foo.
$ gdb -q ./bar
Reading symbols from ./bar...
(gdb) add-inferior
[New inferior 2]
Added inferior 2
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) target remote :1234
...
(gdb) set debug remote 2
(gdb) break bar.c:2
Sending packet: $Hgp0.0#ad...Packet received: OK
Sending packet: $m5fa,12#f8...Packet received: E01
Sending packet: $m5fa,1#c6...Packet received: E01
Sending packet: $m5fb,3#c9...Packet received: E01
Sending packet: $m5fe,1#ca...Packet received: E01
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Here we have an unnecessary sending of the packets to the gdbserver.
With this fix in progspace-and-thread.c, we'll get this:
(gdb) break bar.c:2
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Now there is no sending of the packets to gdbserver.
The changes around clear_symtab_users calls are necessary because
otherwise we regress gdb.base/step-over-exit.exp, hitting the new
assertion in switch_to_program_space_and_thread. The problem is, a
forked child terminates, and when GDB decides to auto-purge that
inferior, GDB tries to switch to the pspace of that no-longer-existing
inferior.
The root of the problem is within the program_space destructor:
program_space::~program_space ()
{
...
set_current_program_space (this); # (1)
...
breakpoint_program_space_exit (this); # (2)
...
free_all_objfiles (); # (3)
...
}
We get here from delete_inferior -> delete_program_space.
So we're deleting an inferior, and the inferior to be
deleted is no longer in the inferior list.
At (2), we've deleted all the breakpoints and locations for the
program space being deleted.
The crash happens while doing a breakpoint re-set, called by
clear_symtab_users at the tail end of (3). That is, while recreating
breakpoints for the current program space, which is the program space
we're tearing down. During breakpoint re-set, we try to switch to the
new location's pspace (the current pspace set in (1), so the pspace
we're tearing down) with switch_to_program_space_and_thread, and that
hits the failed assertion. It's the fact that we recreate breakpoints
in the program_space destructor that is the latent bug here. Just
don't do that, and we don't end up in the crash situation.
My first approach to fix this added a symfile_add_flags parameter to
program_space::free_all_objfiles, and then passed that down to
clear_symtab_users. The program_space dtor would then pass down
SYMFILE_DEFER_BP_RESET to free_all_objfiles. I couldn't help feeling
that adding that parameter to free_all_objfiles looked a little
awkward, so I settled on something a little different -- hoist the
clear_symtab_users call to the callers. There are only two callers.
I felt that that didn't look as odd, particularly since
remove_symbol_file_command also does:
objf->unlink ();
clear_symtab_users (0);
I.e., objfile deletion is already separate from calling
clear_symtab_users in some places.
gdb/ChangeLog:
2020-01-10 Aleksandar Paunovic <aleksandar.paunovic@intel.com>
Pedro Alves <palves@redhat.com>
* progspace-and-thread.c (switch_to_program_space_and_thread):
Assert there's an inferior for PSPACE. Use
switch_to_inferior_no_thread to switch the inferior too.
* progspace.c (program_space::~program_space): Call
clear_symtab_users here, with SYMFILE_DEFER_BP_RESET.
(program_space::free_all_objfiles): Don't call clear_symtab_users
here.
* symfile.c (symbol_file_clear): Call clear_symtab_users here.
gdb/testsuite/ChangeLog:
2020-01-10 Pedro Alves <palves@redhat.com>
* gdb.server/bkpt-other-inferior.exp: New file.
2020-01-11 04:06:16 +08:00
|
|
|
-re "Sending packet.*$gdb_prompt $" {
|
|
|
|
fail $test
|
|
|
|
}
|
2020-09-03 22:38:12 +08:00
|
|
|
-re "^break -q main\r\nBreakpoint .* at .*$gdb_prompt $" {
|
Switch the inferior too in switch_to_program_space_and_thread
With multi-target, each inferior now has its own target connection.
The problem in switch_to_program_space_and_thread is that in the
current state GDB switches to "no thread" and also sets the program
space but because the inferior is not switched, potentially an
incorrect target remains selected.
Here is a sample scenario that exploits this flow:
On terminal 1, start a gdbserver on a program named foo:
$ gdbserver :1234 ./foo
On terminal 2, start gdb on a program named bar. Suppose foo and bar
are compiled from foo.c and bar.c. They are completely separate. So,
bar.c:2 has no meaning for foo.
$ gdb -q ./bar
Reading symbols from ./bar...
(gdb) add-inferior
[New inferior 2]
Added inferior 2
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) target remote :1234
...
(gdb) set debug remote 2
(gdb) break bar.c:2
Sending packet: $Hgp0.0#ad...Packet received: OK
Sending packet: $m5fa,12#f8...Packet received: E01
Sending packet: $m5fa,1#c6...Packet received: E01
Sending packet: $m5fb,3#c9...Packet received: E01
Sending packet: $m5fe,1#ca...Packet received: E01
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Here we have an unnecessary sending of the packets to the gdbserver.
With this fix in progspace-and-thread.c, we'll get this:
(gdb) break bar.c:2
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Now there is no sending of the packets to gdbserver.
The changes around clear_symtab_users calls are necessary because
otherwise we regress gdb.base/step-over-exit.exp, hitting the new
assertion in switch_to_program_space_and_thread. The problem is, a
forked child terminates, and when GDB decides to auto-purge that
inferior, GDB tries to switch to the pspace of that no-longer-existing
inferior.
The root of the problem is within the program_space destructor:
program_space::~program_space ()
{
...
set_current_program_space (this); # (1)
...
breakpoint_program_space_exit (this); # (2)
...
free_all_objfiles (); # (3)
...
}
We get here from delete_inferior -> delete_program_space.
So we're deleting an inferior, and the inferior to be
deleted is no longer in the inferior list.
At (2), we've deleted all the breakpoints and locations for the
program space being deleted.
The crash happens while doing a breakpoint re-set, called by
clear_symtab_users at the tail end of (3). That is, while recreating
breakpoints for the current program space, which is the program space
we're tearing down. During breakpoint re-set, we try to switch to the
new location's pspace (the current pspace set in (1), so the pspace
we're tearing down) with switch_to_program_space_and_thread, and that
hits the failed assertion. It's the fact that we recreate breakpoints
in the program_space destructor that is the latent bug here. Just
don't do that, and we don't end up in the crash situation.
My first approach to fix this added a symfile_add_flags parameter to
program_space::free_all_objfiles, and then passed that down to
clear_symtab_users. The program_space dtor would then pass down
SYMFILE_DEFER_BP_RESET to free_all_objfiles. I couldn't help feeling
that adding that parameter to free_all_objfiles looked a little
awkward, so I settled on something a little different -- hoist the
clear_symtab_users call to the callers. There are only two callers.
I felt that that didn't look as odd, particularly since
remove_symbol_file_command also does:
objf->unlink ();
clear_symtab_users (0);
I.e., objfile deletion is already separate from calling
clear_symtab_users in some places.
gdb/ChangeLog:
2020-01-10 Aleksandar Paunovic <aleksandar.paunovic@intel.com>
Pedro Alves <palves@redhat.com>
* progspace-and-thread.c (switch_to_program_space_and_thread):
Assert there's an inferior for PSPACE. Use
switch_to_inferior_no_thread to switch the inferior too.
* progspace.c (program_space::~program_space): Call
clear_symtab_users here, with SYMFILE_DEFER_BP_RESET.
(program_space::free_all_objfiles): Don't call clear_symtab_users
here.
* symfile.c (symbol_file_clear): Call clear_symtab_users here.
gdb/testsuite/ChangeLog:
2020-01-10 Pedro Alves <palves@redhat.com>
* gdb.server/bkpt-other-inferior.exp: New file.
2020-01-11 04:06:16 +08:00
|
|
|
pass $test
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Fix bkpt-other-inferior.exp race
When testing with "maint set target-non-stop on",
gdb.server/bkpt-other-inferior.exp sometimes fails like so:
(gdb) inferior 2
[Switching to inferior 2 [process 368191] (<noexec>)]
[Switching to thread 2.1 (Thread 368191.368191)]
[remote] Sending packet: $m7ffff7fd0100,1#5b
[remote] Packet received: 48
[remote] Sending packet: $m7ffff7fd0100,1#5b
[remote] Packet received: 48
[remote] Sending packet: $m7ffff7fd0100,9#63
[remote] Packet received: 4889e7e8e80c000049
#0 0x00007ffff7fd0100 in ?? ()
(gdb) PASS: gdb.server/bkpt-other-inferior.exp: inf 2: switch to inferior
break -q main
Breakpoint 2 at 0x1138: file /home/pedro/gdb/binutils-gdb/src/gdb/testsuite/gdb.server/server.c, line 21.
(gdb) PASS: gdb.server/bkpt-other-inferior.exp: inf 2: set breakpoint
delete breakpoints
Delete all breakpoints? (y or n) y
(gdb) [remote] wait: enter
[remote] wait: exit
FAIL: gdb.server/bkpt-other-inferior.exp: inf 2: delete all breakpoints in delete_breakpoints (timeout)
ERROR: breakpoints not deleted
Remote debugging from host ::1, port 55876
monitor exit
The problem is here:
(gdb) [remote] wait: enter
The testcase isn't expecting any output after the prompt.
Why is that "[remote] wait" output? What happens is that "delete
breakpoints" queries the user, and `query` disables/reenables target
async, which results in the remote target's async event handler ending
up marked:
(top-gdb) bt
#0 mark_async_event_handler (async_handler_ptr=0x556bffffffff) at ../../src/gdb/async-event.c:295
#1 0x0000556bf71b711f in infrun_async (enable=1) at ../../src/gdb/infrun.c:119
#2 0x0000556bf7471387 in target_async (enable=1) at ../../src/gdb/target.c:3684
#3 0x0000556bf748a0bd in gdb_readline_wrapper_cleanup::~gdb_readline_wrapper_cleanup (this=0x7ffe3cf30eb0, __in_chrg=<optimized out>) at ../../src/gdb/top.c:1074
#4 0x0000556bf74874e2 in gdb_readline_wrapper (prompt=0x556bfa17da60 "Delete all breakpoints? (y or n) ") at ../../src/gdb/top.c:1096
#5 0x0000556bf75111c5 in defaulted_query(const char *, char, typedef __va_list_tag __va_list_tag *) (ctlstr=0x556bf7717f34 "Delete all breakpoints? ", defchar=0 '\000', args=0x7ffe3cf31020) at ../../src/gdb/utils.c:893
#6 0x0000556bf751166f in query (ctlstr=0x556bf7717f34 "Delete all breakpoints? ") at ../../src/gdb/utils.c:985
#7 0x0000556bf6f11404 in delete_command (arg=0x0, from_tty=1) at ../../src/gdb/breakpoint.c:13500
...
... which then later results in a target_wait call:
(top-gdb) bt
#0 remote_target::wait_ns (this=0x7ffe3cf30f80, ptid=..., status=0xde530314f0802800, options=...) at ../../src/gdb/remote.c:7937
#1 0x0000556bf7369dcb in remote_target::wait (this=0x556bfa0b2180, ptid=..., status=0x7ffe3cf31568, options=...) at ../../src/gdb/remote.c:8173
#2 0x0000556bf745e527 in target_wait (ptid=..., status=0x7ffe3cf31568, options=...) at ../../src/gdb/target.c:2000
#3 0x0000556bf71be686 in do_target_wait_1 (inf=0x556bfa1573d0, ptid=..., status=0x7ffe3cf31568, options=...) at ../../src/gdb/infrun.c:3463
#4 0x0000556bf71be88b in <lambda(inferior*)>::operator()(inferior *) const (__closure=0x7ffe3cf31320, inf=0x556bfa1573d0) at ../../src/gdb/infrun.c:3526
#5 0x0000556bf71bebcd in do_target_wait (wait_ptid=..., ecs=0x7ffe3cf31540, options=...) at ../../src/gdb/infrun.c:3539
#6 0x0000556bf71bf97b in fetch_inferior_event () at ../../src/gdb/infrun.c:3879
#7 0x0000556bf71a27f8 in inferior_event_handler (event_type=INF_REG_EVENT) at ../../src/gdb/inf-loop.c:42
#8 0x0000556bf71cc8b7 in infrun_async_inferior_event_handler (data=0x0) at ../../src/gdb/infrun.c:9220
#9 0x0000556bf6ecb80f in check_async_event_handlers () at ../../src/gdb/async-event.c:327
#10 0x0000556bf76b011a in gdb_do_one_event () at ../../src/gdbsupport/event-loop.cc:216
...
... which returns TARGET_WAITKIND_IGNORE.
Fix this by only enabling remote output around setting the breakpoint.
gdb/testsuite/ChangeLog:
* gdb.server/bkpt-other-inferior.exp: Only enable remote output
around setting the breakpoint.
Change-Id: I2fd152fd9c46b1c5e7fa678cc4d4054dac0b2bd4
2021-02-14 00:25:26 +08:00
|
|
|
gdb_test_no_output "set debug remote 0"
|
|
|
|
|
Switch the inferior too in switch_to_program_space_and_thread
With multi-target, each inferior now has its own target connection.
The problem in switch_to_program_space_and_thread is that in the
current state GDB switches to "no thread" and also sets the program
space but because the inferior is not switched, potentially an
incorrect target remains selected.
Here is a sample scenario that exploits this flow:
On terminal 1, start a gdbserver on a program named foo:
$ gdbserver :1234 ./foo
On terminal 2, start gdb on a program named bar. Suppose foo and bar
are compiled from foo.c and bar.c. They are completely separate. So,
bar.c:2 has no meaning for foo.
$ gdb -q ./bar
Reading symbols from ./bar...
(gdb) add-inferior
[New inferior 2]
Added inferior 2
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) target remote :1234
...
(gdb) set debug remote 2
(gdb) break bar.c:2
Sending packet: $Hgp0.0#ad...Packet received: OK
Sending packet: $m5fa,12#f8...Packet received: E01
Sending packet: $m5fa,1#c6...Packet received: E01
Sending packet: $m5fb,3#c9...Packet received: E01
Sending packet: $m5fe,1#ca...Packet received: E01
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Here we have an unnecessary sending of the packets to the gdbserver.
With this fix in progspace-and-thread.c, we'll get this:
(gdb) break bar.c:2
Breakpoint 1 at 0x5fe: file bar.c, line 2.
(gdb)
Now there is no sending of the packets to gdbserver.
The changes around clear_symtab_users calls are necessary because
otherwise we regress gdb.base/step-over-exit.exp, hitting the new
assertion in switch_to_program_space_and_thread. The problem is, a
forked child terminates, and when GDB decides to auto-purge that
inferior, GDB tries to switch to the pspace of that no-longer-existing
inferior.
The root of the problem is within the program_space destructor:
program_space::~program_space ()
{
...
set_current_program_space (this); # (1)
...
breakpoint_program_space_exit (this); # (2)
...
free_all_objfiles (); # (3)
...
}
We get here from delete_inferior -> delete_program_space.
So we're deleting an inferior, and the inferior to be
deleted is no longer in the inferior list.
At (2), we've deleted all the breakpoints and locations for the
program space being deleted.
The crash happens while doing a breakpoint re-set, called by
clear_symtab_users at the tail end of (3). That is, while recreating
breakpoints for the current program space, which is the program space
we're tearing down. During breakpoint re-set, we try to switch to the
new location's pspace (the current pspace set in (1), so the pspace
we're tearing down) with switch_to_program_space_and_thread, and that
hits the failed assertion. It's the fact that we recreate breakpoints
in the program_space destructor that is the latent bug here. Just
don't do that, and we don't end up in the crash situation.
My first approach to fix this added a symfile_add_flags parameter to
program_space::free_all_objfiles, and then passed that down to
clear_symtab_users. The program_space dtor would then pass down
SYMFILE_DEFER_BP_RESET to free_all_objfiles. I couldn't help feeling
that adding that parameter to free_all_objfiles looked a little
awkward, so I settled on something a little different -- hoist the
clear_symtab_users call to the callers. There are only two callers.
I felt that that didn't look as odd, particularly since
remove_symbol_file_command also does:
objf->unlink ();
clear_symtab_users (0);
I.e., objfile deletion is already separate from calling
clear_symtab_users in some places.
gdb/ChangeLog:
2020-01-10 Aleksandar Paunovic <aleksandar.paunovic@intel.com>
Pedro Alves <palves@redhat.com>
* progspace-and-thread.c (switch_to_program_space_and_thread):
Assert there's an inferior for PSPACE. Use
switch_to_inferior_no_thread to switch the inferior too.
* progspace.c (program_space::~program_space): Call
clear_symtab_users here, with SYMFILE_DEFER_BP_RESET.
(program_space::free_all_objfiles): Don't call clear_symtab_users
here.
* symfile.c (symbol_file_clear): Call clear_symtab_users here.
gdb/testsuite/ChangeLog:
2020-01-10 Pedro Alves <palves@redhat.com>
* gdb.server/bkpt-other-inferior.exp: New file.
2020-01-11 04:06:16 +08:00
|
|
|
delete_breakpoints
|
|
|
|
}
|
|
|
|
}
|