From: Markus Metzger <markus.t.metzger@intel.com>
To: gdb-patches@sourceware.org
Cc: pedro@palves.net
Subject: [PATCH v3 2/4] gdb, infrun, record: fix hang when step-over fails with no-history
Date: Mon, 4 Jul 2022 13:54:05 +0200 [thread overview]
Message-ID: <20220704115407.1239498-3-markus.t.metzger@intel.com> (raw)
In-Reply-To: <20220704115407.1239498-1-markus.t.metzger@intel.com>
When trying to step over a breakpoint at the end of the trace while
another thread is replaying, the step-over will fail with no-history.
This does not clear step_over_info so a subsequent resume will cause GDB
to not resume the thread and expect a SIGTRAP to complete the step-over.
This will never come causing GDB to hang in the wait-for-event poll.
This is a variant of the issue fixed in the parent commit. That commit
addressed the issue for a single-threaded process and fixed an issue with
reverse/replay stepping in general.
This commit addresses the issue for a multi-threaded process. In this
case, the single-step does not complete.
Finish an in-flight step-over when a thread stopped with NO_HISTORY.
Since we did not move, we will simply start the step-over again.
---
gdb/infrun.c | 14 +++
| 88 +++++++++++++++++++
2 files changed, 102 insertions(+)
create mode 100644 gdb/testsuite/gdb.btrace/multi-thread-break-hang.exp
diff --git a/gdb/infrun.c b/gdb/infrun.c
index a821ba06b22..3fa262efb01 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -74,6 +74,8 @@
#include "gdbsupport/common-debug.h"
#include "gdbsupport/buildargv.h"
+struct execution_control_state;
+
/* Prototypes for local functions */
static void sig_print_info (enum gdb_signal);
@@ -103,6 +105,8 @@ static bool start_step_over (void);
static bool step_over_info_valid_p (void);
+static int finish_step_over (execution_control_state *ecs);
+
/* Asynchronous signal handler registered as event loop source for
when we have pending events ready to be passed to the core. */
static struct async_event_handler *infrun_async_inferior_event_token;
@@ -5872,6 +5876,16 @@ handle_inferior_event (struct execution_control_state *ecs)
return;
gdb::observers::no_history.notify ();
+
+ /* Cancel an in-flight step-over. It will not succeed since we
+ won't be able to step at the end of the execution history. */
+ {
+ /* finish_step_over may call restart_threads, which may change the
+ current thread. make sure we leave the event thread as the
+ current thread. */
+ scoped_restore_current_thread restore_thread;
+ finish_step_over (ecs);
+ }
stop_waiting (ecs);
return;
}
--git a/gdb/testsuite/gdb.btrace/multi-thread-break-hang.exp b/gdb/testsuite/gdb.btrace/multi-thread-break-hang.exp
new file mode 100644
index 00000000000..df0c984cee0
--- /dev/null
+++ b/gdb/testsuite/gdb.btrace/multi-thread-break-hang.exp
@@ -0,0 +1,88 @@
+# This testcase is part of GDB, the GNU debugger.
+#
+# Copyright 2021-2022 Free Software Foundation, Inc.
+#
+# 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 we cancel an in-flight step-over at the end of the execution
+# history as long as some other thread is still replaying.
+#
+# This used to cause GDB to hang in poll ().
+
+if { [skip_btrace_tests] } {
+ unsupported "target does not support record-btrace"
+ return -1
+}
+
+standard_testfile multi-thread-step.c
+if [prepare_for_testing "failed to prepare" $testfile $srcfile {debug pthreads}] {
+ return -1
+}
+
+if ![runto_main] {
+ untested "failed to run to main"
+ return -1
+}
+
+# Set up breakpoints.
+set bp_1 [gdb_get_line_number "bp.1" $srcfile]
+set bp_2 [gdb_get_line_number "bp.2" $srcfile]
+
+# Trace the code between the two breakpoints.
+gdb_breakpoint $srcfile:$bp_1
+gdb_continue_to_breakpoint "continue to bp.1" ".*$srcfile:$bp_1\r\n.*"
+
+gdb_test_no_output "record btrace"
+
+# We have two threads at or close to bp.1 but handled only one stop event.
+# Remove the breakpoint so we do not need to deal with the 2nd event.
+delete_breakpoints
+gdb_breakpoint $srcfile:$bp_2
+gdb_continue_to_breakpoint "continue to bp.2" ".*$srcfile:$bp_2\r\n.*"
+
+# Determine the thread that reported the breakpoint.
+set thread [get_integer_valueof "\$_thread" bad]
+
+# Determine the other thread.
+set other "bad"
+if { $thread == 1 } {
+ set other 2
+} elseif { $thread == 2 } {
+ set other 1
+}
+
+# This test requires scheduler-locking 'on' or 'step'; 'replay' would
+# implicitly stop replaying, avoiding the problem; 'off' would step one
+# and resume the other.
+#
+# With the current record-btrace implementation that steps all resumed
+# threads in lock-step, 'off' might actually pass but we don't want to
+# bake that behavior into tests.
+gdb_test_no_output "set scheduler-locking step"
+
+# Start replaying the other thread. This will prevent stepping the thread
+# that reported the event.
+gdb_test "thread apply $other record goto begin" ".*"
+gdb_test "thread apply $other info record" "Replay in progress.*"
+
+# We're at a breakpoint so this triggers step-over. Since we're at the
+# end of the trace, the step will fail.
+gdb_test "stepi" "No more reverse-execution history.*" "stepi.1"
+
+# We used to hang at the second step since step-over insisted on polling
+# the next event.
+gdb_test "stepi" "No more reverse-execution history.*" "stepi.2"
+
+# Do one more just in case.
+gdb_test "stepi" "No more reverse-execution history.*" "stepi.3"
--
2.35.3
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
next prev parent reply other threads:[~2022-07-04 12:58 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-07-04 11:54 [PATCH v3 0/4] gdb, btrace: infrun fixes Markus Metzger
2022-07-04 11:54 ` [PATCH v3 1/4] gdb, infrun, btrace: fix reverse/replay stepping at end of execution history Markus Metzger
2022-07-04 11:54 ` Markus Metzger [this message]
2022-07-04 11:54 ` [PATCH v3 3/4] gdb, infrun, record: move no-history notification into normal_stop Markus Metzger
2022-07-04 11:54 ` [PATCH v3 4/4] gdb, infrun: fix multi-threaded reverse stepping Markus Metzger
2022-08-29 4:41 ` [PATCH v3 0/4] gdb, btrace: infrun fixes Metzger, Markus T
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20220704115407.1239498-3-markus.t.metzger@intel.com \
--to=markus.t.metzger@intel.com \
--cc=gdb-patches@sourceware.org \
--cc=pedro@palves.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).