From patchwork Tue Mar 12 11:34:20 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Metzger, Markus T" X-Patchwork-Id: 87074 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 6206F3858008 for ; Tue, 12 Mar 2024 11:35:06 +0000 (GMT) X-Original-To: gdb-patches@sourceware.org Delivered-To: gdb-patches@sourceware.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.12]) by sourceware.org (Postfix) with ESMTPS id 98F3D3858281 for ; Tue, 12 Mar 2024 11:34:30 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 98F3D3858281 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=intel.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org 98F3D3858281 Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=192.198.163.12 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243274; cv=none; b=LY+xy4NKA6C+LA/mhKEcJ8YTTGPJY931R2kiAFGJ8ENoF1E7QWqVTLHHYyPvMfzDMDe2gyT1Qxirk0fvw6AY9mW33N2ABOdmS9yrO11dlk6MOSXI5S5FAG6BcUSiQcU3UHzULZbNECSnggHy/XO9qGyN8P/XeOeuzaKEUd8QRWU= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243274; c=relaxed/simple; bh=uNc0+MQ1a66pAMtT2Vj5duxeaGG4rdmjhCwmt5O8R9g=; h=DKIM-Signature:From:To:Subject:Date:Message-Id:MIME-Version; b=RU9q8h93XJPojU5YXL/HwVGbBzUFBGV1sQkPSaZI6nON3AXqSFLkfo37k9jsFqtB9OTmvJ4Xi7XbcUd4p935lgvZlkrBdEx6ahUDhGpnntgQx6bsSwCBM8Hd/VO4wYIOa2TQ6dcIWv4cwfJfBtUMLBKW09YaQK2ult1a4EFuzus= ARC-Authentication-Results: i=1; server2.sourceware.org DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1710243270; x=1741779270; h=from:to:subject:date:message-id:in-reply-to:references: mime-version:content-transfer-encoding; bh=uNc0+MQ1a66pAMtT2Vj5duxeaGG4rdmjhCwmt5O8R9g=; b=OBsUhNojk6FBSlWJQx56hvunPyloar+t+97g0IwJjMBW5YLUO3Nt7dhH hP7UxvHmsTp+uVHmmV+i68P+C/mXXCspAqksnueG372/E/3jAp8Cz/Tpo yEmkGa2nmYl+TF8ZU7wdH8RaGKASvLAPp6uksOaKnHtL6sbgYdnzfrQCu XG/Wsgzo9w9nTT2Pn/4w0sWFNVpF6P+BvT+ZKeaznmBRRuMkD68RUilmh i5/in/w7BNzT99kSTqhpNDNJF3C/vEI5Ov+SOx/gbSu/vL+OfU+XMLl/M c2hJJvHjSi15zKOHwvYDGAjwPgUbVJWXC08kokwlZ9gd7N58t9NS/BfhK Q==; X-IronPort-AV: E=McAfee;i="6600,9927,11010"; a="8709955" X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="8709955" Received: from fmviesa005.fm.intel.com ([10.60.135.145]) by fmvoesa106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:30 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="15999861" Received: from gkldtt-dev-004.igk.intel.com (HELO localhost) ([10.123.221.202]) by fmviesa005-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:28 -0700 From: Markus Metzger To: gdb-patches@sourceware.org Subject: [PATCH v4 1/4] gdb, infrun, btrace: fix reverse/replay stepping at end of execution history Date: Tue, 12 Mar 2024 11:34:20 +0000 Message-Id: <20240312113423.3543956-2-markus.t.metzger@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240312113423.3543956-1-markus.t.metzger@intel.com> References: <20240312113423.3543956-1-markus.t.metzger@intel.com> MIME-Version: 1.0 X-Spam-Status: No, score=-10.0 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_NONE, TXREP, T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.30 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: gdb-patches-bounces+patchwork=sourceware.org@sourceware.org When trying to step over a breakpoint at the end of the trace, 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. That step-over failed after actually completing the step. This is wrong. The step-over itself should have failed and the step should not have completed. Fix it by moving the end of execution history check to before we are stepping. This exposes another issue, however. When completing a step-over at the end of the execution history, we implicitly stop replaying that thread. A continue command would resume after the step-over and, since we're no longer replaying, would continue recording. Fix that by recording the replay state in the thread's control state and failing with no-history in keep_going if we're switching from replay to recording. Reviewed-By: Guinevere Larsen --- gdb/gdbthread.h | 3 ++ gdb/infrun.c | 25 +++++++++++++ gdb/record-btrace.c | 19 +++++----- gdb/testsuite/gdb.btrace/cont-hang.exp | 43 ++++++++++++++++++++++ gdb/testsuite/gdb.btrace/step-hang.exp | 42 ++++++++++++++++++++++ gdb/testsuite/gdb.btrace/stepn.exp | 50 ++++++++++++++++++++++++++ 6 files changed, 173 insertions(+), 9 deletions(-) create mode 100644 gdb/testsuite/gdb.btrace/cont-hang.exp create mode 100644 gdb/testsuite/gdb.btrace/step-hang.exp create mode 100644 gdb/testsuite/gdb.btrace/stepn.exp diff --git a/gdb/gdbthread.h b/gdb/gdbthread.h index 11c553a99ca..4931a0cc2f1 100644 --- a/gdb/gdbthread.h +++ b/gdb/gdbthread.h @@ -173,6 +173,9 @@ struct thread_control_state command. This is used to decide whether "set scheduler-locking step" behaves like "on" or "off". */ int stepping_command = 0; + + /* Whether the thread was replaying when the command was issued. */ + bool is_replaying = false; }; /* Inferior thread specific part of `struct infcall_suspend_state'. */ diff --git a/gdb/infrun.c b/gdb/infrun.c index bbb98f6dcdb..f38d96b64df 100644 --- a/gdb/infrun.c +++ b/gdb/infrun.c @@ -3089,6 +3089,8 @@ clear_proceed_status_thread (struct thread_info *tp) /* Discard any remaining commands or status from previous stop. */ bpstat_clear (&tp->control.stop_bpstat); + + tp->control.is_replaying = target_record_is_replaying (tp->ptid); } /* Notify the current interpreter and observers that the target is about to @@ -8972,6 +8974,29 @@ keep_going_pass_signal (struct execution_control_state *ecs) gdb_assert (ecs->event_thread->ptid == inferior_ptid); gdb_assert (!ecs->event_thread->resumed ()); + /* When a thread reaches the end of its execution history, it automatically + stops replaying. This is so the user doesn't need to explicitly stop it + with a separate command. + + We do not want a single command (e.g. continue) to transition from + replaying to recording, though, e.g. when starting from a breakpoint we + needed to step over at the end of the trace. When we reach the end of the + execution history during stepping, stop with no-history. + + The other direction is fine. When we're at the end of the execution + history, we may reverse-continue to start replaying. */ + if (ecs->event_thread->control.is_replaying + && !target_record_is_replaying (ecs->event_thread->ptid)) + { + interps_notify_no_history (); + ecs->ws.set_no_history (); + set_last_target_status (ecs->target, ecs->ptid, ecs->ws); + stop_print_frame = true; + stop_waiting (ecs); + normal_stop (); + return; + } + /* Save the pc before execution, to compare with pc after stop. */ ecs->event_thread->prev_pc = regcache_read_pc_protected (get_thread_regcache (ecs->event_thread)); diff --git a/gdb/record-btrace.c b/gdb/record-btrace.c index 6350400c318..ae273fda507 100644 --- a/gdb/record-btrace.c +++ b/gdb/record-btrace.c @@ -2340,6 +2340,16 @@ record_btrace_single_step_forward (struct thread_info *tp) if (replay == NULL) return btrace_step_no_history (); + /* The execution trace contains (and ends with) the current instruction. + This instruction has not been executed, yet, so the trace really ends + one instruction earlier. + + We'd fail later on in btrace_insn_next () but we must not trigger + breakpoints as we're not really able to step. */ + btrace_insn_end (&end, btinfo); + if (btrace_insn_cmp (replay, &end) == 0) + return btrace_step_no_history (); + /* Check if we're stepping a breakpoint. */ if (record_btrace_replay_at_breakpoint (tp)) return btrace_step_stopped (); @@ -2362,15 +2372,6 @@ record_btrace_single_step_forward (struct thread_info *tp) } while (btrace_insn_get (replay) == NULL); - /* Determine the end of the instruction trace. */ - btrace_insn_end (&end, btinfo); - - /* The execution trace contains (and ends with) the current instruction. - This instruction has not been executed, yet, so the trace really ends - one instruction earlier. */ - if (btrace_insn_cmp (replay, &end) == 0) - return btrace_step_no_history (); - return btrace_step_spurious (); } diff --git a/gdb/testsuite/gdb.btrace/cont-hang.exp b/gdb/testsuite/gdb.btrace/cont-hang.exp new file mode 100644 index 00000000000..cddcf68b3ab --- /dev/null +++ b/gdb/testsuite/gdb.btrace/cont-hang.exp @@ -0,0 +1,43 @@ +# This testcase is part of GDB, the GNU debugger. +# +# Copyright 2024 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 . + +# Test that we do not hang when trying to continue over a breakpoint at +# the end of the trace. + +require allow_btrace_tests + +standard_testfile record_goto.c +if [prepare_for_testing "failed to prepare" $testfile $srcfile] { + return -1 +} + +if ![runto_main] { + return -1 +} + +# Trace the call to the test function. +gdb_test_no_output "record btrace" +gdb_test "next" "main\.3.*" + +# We need to be replaying, otherwise, we'd just continue recording. +gdb_test "reverse-stepi" +gdb_test "break" + +# Continuing will step over the breakpoint and then run into the end of +# the execution history. This ends replay, so we can continue recording. +gdb_test "continue" "No more reverse-execution history.*" +gdb_continue_to_end diff --git a/gdb/testsuite/gdb.btrace/step-hang.exp b/gdb/testsuite/gdb.btrace/step-hang.exp new file mode 100644 index 00000000000..91ea813955d --- /dev/null +++ b/gdb/testsuite/gdb.btrace/step-hang.exp @@ -0,0 +1,42 @@ +# This testcase is part of GDB, the GNU debugger. +# +# Copyright 2024 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 . + +# Test that we do not hang when trying to step over a breakpoint at the +# end of the trace. + +require allow_btrace_tests + +standard_testfile record_goto.c +if [prepare_for_testing "failed to prepare" $testfile $srcfile] { + return -1 +} + +if ![runto_main] { + return -1 +} + +# Trace the call to the test function. +gdb_test_no_output "record btrace" +gdb_test "next" "main\.3.*" + +# We need to be replaying, otherwise, we'd just continue recording. +gdb_test "reverse-stepi" +gdb_test "break" + +# Stepping over the breakpoint ends replaying and we can continue recording. +gdb_test "step" "main\.3.*" +gdb_continue_to_end diff --git a/gdb/testsuite/gdb.btrace/stepn.exp b/gdb/testsuite/gdb.btrace/stepn.exp new file mode 100644 index 00000000000..4aec90adc65 --- /dev/null +++ b/gdb/testsuite/gdb.btrace/stepn.exp @@ -0,0 +1,50 @@ +# This testcase is part of GDB, the GNU debugger. +# +# Copyright 2024 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 . + +# Test that step n does not start recording when issued while replaying. + +require allow_btrace_tests + +standard_testfile record_goto.c +if [prepare_for_testing "failed to prepare" $testfile $srcfile] { + return -1 +} + +if ![runto_main] { + return -1 +} + +# Trace the call to the test function. +gdb_test_no_output "record btrace" +gdb_test "next" "main\.3.*" + +# Stepping should bring us to the end of the execution history, but should +# not resume recording. +with_test_prefix "stepi" { + gdb_test "reverse-stepi" + gdb_test "stepi 5" "No more reverse-execution history.*main\.3.*" +} + +with_test_prefix "step" { + gdb_test "reverse-step" + gdb_test "step 5" "No more reverse-execution history.*main\.3.*" +} + +with_test_prefix "next" { + gdb_test "reverse-next" + gdb_test "next 5" "No more reverse-execution history.*main\.3.*" +} From patchwork Tue Mar 12 11:34:21 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Metzger, Markus T" X-Patchwork-Id: 87076 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id C9AF83858020 for ; Tue, 12 Mar 2024 11:35:46 +0000 (GMT) X-Original-To: gdb-patches@sourceware.org Delivered-To: gdb-patches@sourceware.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.12]) by sourceware.org (Postfix) with ESMTPS id 35BF03858D39 for ; Tue, 12 Mar 2024 11:34:33 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 35BF03858D39 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=intel.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org 35BF03858D39 Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=192.198.163.12 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243276; cv=none; b=lVKzwoMZcvtC+I13oj5VBIdp8Nms8LCRCTbFuUCfgKBNbEJA8CMucb51iIioP9cXUibW4fR04j4HxXCZ6emjGtB7iqr7nLXezo9MLIHZD5TZLviv4zo7CSGuUFdSZgdO5r/AYwl6jaLluGY15tUjENQNtc+pZqOEi8Zn8iFc6Pw= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243276; c=relaxed/simple; bh=gLZanBYSGaBp+zZ3LIRFvwdw+M+J9N1R5zE6Z018nnE=; h=DKIM-Signature:From:To:Subject:Date:Message-Id:MIME-Version; b=sE0mKIaNN9OXkNUXc+BGaH1rmPLrv99iyXgpY3OBh7pjcdF5TlFeQeqEnMu9Z212WUDsmgSEohAxqQlxNuIELaIjOyo0IBQ5jq5buwwJCLbR/QDvt/8JSeOSwzu0Y3k7V5WQeyl3GEIpAsv6BZmuAFUNTqXcn1QcvNl9vV3XI7U= ARC-Authentication-Results: i=1; server2.sourceware.org DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1710243273; x=1741779273; h=from:to:subject:date:message-id:in-reply-to:references: mime-version:content-transfer-encoding; bh=gLZanBYSGaBp+zZ3LIRFvwdw+M+J9N1R5zE6Z018nnE=; b=T/3uwtzJX3Ap7lVE3lao4P1/vAu0BOp2lPkLgTOudj1L792dsiTX8Mdc xTZFrqnqcOS++ItQe8nOa4jbPRYYhzcQ2E3VLYvDLWwc66OSPrivXC7zY qizY316wdF883yNrAR3k0AR9RwSSK81WRZKT5tYr7u08efIFmVdRxLhkM XkB2drI/S5rj6+nUiWpf+xZrPyeBE5MOzFjUBQCG539i7PzcCUhcFlNE+ GGhI8w3VDlofO9CKe/XmGmQx9KIPvydhO5cvpBEfn+LPEEs9tvTm9t/1x gQPZVFn4Go2HutwxPjbXgtvVw61KWztKdFivwRRYoOj/QbzlmOptNj1Ib Q==; X-IronPort-AV: E=McAfee;i="6600,9927,11010"; a="8709962" X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="8709962" Received: from fmviesa005.fm.intel.com ([10.60.135.145]) by fmvoesa106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:33 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="15999871" Received: from gkldtt-dev-004.igk.intel.com (HELO localhost) ([10.123.221.202]) by fmviesa005-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:31 -0700 From: Markus Metzger To: gdb-patches@sourceware.org Subject: [PATCH v4 2/4] gdb, infrun, record: fix hang when step-over fails with no-history Date: Tue, 12 Mar 2024 11:34:21 +0000 Message-Id: <20240312113423.3543956-3-markus.t.metzger@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240312113423.3543956-1-markus.t.metzger@intel.com> References: <20240312113423.3543956-1-markus.t.metzger@intel.com> MIME-Version: 1.0 X-Spam-Status: No, score=-10.0 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_NONE, TXREP, T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.30 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: gdb-patches-bounces+patchwork=sourceware.org@sourceware.org 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. Reviewed-By: Guinevere Larsen --- gdb/infrun.c | 14 ++++ .../gdb.btrace/multi-thread-break-hang.exp | 84 +++++++++++++++++++ 2 files changed, 98 insertions(+) create mode 100644 gdb/testsuite/gdb.btrace/multi-thread-break-hang.exp diff --git a/gdb/infrun.c b/gdb/infrun.c index f38d96b64df..eb5164b4066 100644 --- a/gdb/infrun.c +++ b/gdb/infrun.c @@ -78,6 +78,8 @@ #include "disasm.h" #include "interps.h" +struct execution_control_state; + /* Prototypes for local functions */ static void sig_print_info (enum gdb_signal); @@ -109,6 +111,8 @@ static bool step_over_info_valid_p (void); static bool schedlock_applies (struct thread_info *tp); +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; @@ -6535,6 +6539,16 @@ handle_inferior_event (struct execution_control_state *ecs) return; interps_notify_no_history (); + + /* 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; } diff --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..33edcf05612 --- /dev/null +++ b/gdb/testsuite/gdb.btrace/multi-thread-break-hang.exp @@ -0,0 +1,84 @@ +# This testcase is part of GDB, the GNU debugger. +# +# Copyright 2024 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 . + +# 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 (). + +require allow_btrace_tests + +standard_testfile multi-thread-step.c +if [prepare_for_testing "failed to prepare" $testfile $srcfile {debug pthreads}] { + return -1 +} + +if ![runto_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" From patchwork Tue Mar 12 11:34:22 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Metzger, Markus T" X-Patchwork-Id: 87075 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 2E6943858014 for ; Tue, 12 Mar 2024 11:35:43 +0000 (GMT) X-Original-To: gdb-patches@sourceware.org Delivered-To: gdb-patches@sourceware.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.12]) by sourceware.org (Postfix) with ESMTPS id A031938582AF for ; Tue, 12 Mar 2024 11:34:35 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org A031938582AF Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=intel.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org A031938582AF Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=192.198.163.12 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243280; cv=none; b=L/7PeIXesLZd0hlc0HOqCcXUjcT/LeDhKpTxQZMTAWYB/KAzjlN3E0JRJOIneGs0VmGXgJt6p1Ho9iKNq3CA1VO7Fb+SX3dxenAKX/EnCUowaRHSoMQXyBU54fBealubufJ8liNEx4tgDktNctIiR5IvWeGVGpKKs1fg5Q9laL8= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243280; c=relaxed/simple; bh=WBAYW8mKVgHH8ranLhi58QZf/aDs311wRzZeZLSLrsY=; h=DKIM-Signature:From:To:Subject:Date:Message-Id:MIME-Version; b=xViQHaASIvH2Z1ei1jsAzhJk+Ix2TZmpgQ/tOhb7tiR980g1EQd3pqFyuWSso7oO/9NJBDKy9yfhi20WISQ05BqJKlLhW5yFFGWZwCoCRTFT0ByunEwdZqTnGc7ClKnwjw3FoyivR8isXz55gxJ0bUDBWzscbZU+NPgTB1/50UM= ARC-Authentication-Results: i=1; server2.sourceware.org DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1710243276; x=1741779276; h=from:to:subject:date:message-id:in-reply-to:references: mime-version:content-transfer-encoding; bh=WBAYW8mKVgHH8ranLhi58QZf/aDs311wRzZeZLSLrsY=; b=R99SpIE32djFLkHQxG0wmeK2egHuzD4ll9fe0gdeZem9Il6MLxEvAzPa +EtXqznsSDiNHmwvJZnScjOE6kREZkp2NmjtDoGTlu2PymhKB9boPIFva c21xTEHg+iUAdZqo/HR8wwsftcy0aHRh2JBKFD34/yaZwTHaujhYzrbuG 6XUmXYFkpcJMkXxjhRJFrjwE2eCmCJCXTDub8EK50D25FKCAaRS7YhaGd 8qBScIiaxyNQ5XLPefftNE3YtyI6DeZ9a4uJKbUqzP5RaCqki6gbcPDhj OR8f7RABh1Rg+h4hBmU72UnKRvSBv+DbQF13GHR6XZmO2545EWlZ6vTMz A==; X-IronPort-AV: E=McAfee;i="6600,9927,11010"; a="8709969" X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="8709969" Received: from fmviesa005.fm.intel.com ([10.60.135.145]) by fmvoesa106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:35 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="15999877" Received: from gkldtt-dev-004.igk.intel.com (HELO localhost) ([10.123.221.202]) by fmviesa005-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:34 -0700 From: Markus Metzger To: gdb-patches@sourceware.org Subject: [PATCH v4 3/4] gdb, infrun, record: move no-history notification into normal_stop Date: Tue, 12 Mar 2024 11:34:22 +0000 Message-Id: <20240312113423.3543956-4-markus.t.metzger@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240312113423.3543956-1-markus.t.metzger@intel.com> References: <20240312113423.3543956-1-markus.t.metzger@intel.com> MIME-Version: 1.0 X-Spam-Status: No, score=-10.0 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_NONE, TXREP, T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.30 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: gdb-patches-bounces+patchwork=sourceware.org@sourceware.org Leave calling gdb::observers::no_history.notify to normal_stop based on the last waitstatus. --- gdb/infrun.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/gdb/infrun.c b/gdb/infrun.c index eb5164b4066..f92f529412f 100644 --- a/gdb/infrun.c +++ b/gdb/infrun.c @@ -6538,8 +6538,6 @@ handle_inferior_event (struct execution_control_state *ecs) if (handle_stop_requested (ecs)) return; - interps_notify_no_history (); - /* 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. */ { @@ -9002,7 +9000,6 @@ keep_going_pass_signal (struct execution_control_state *ecs) if (ecs->event_thread->control.is_replaying && !target_record_is_replaying (ecs->event_thread->ptid)) { - interps_notify_no_history (); ecs->ws.set_no_history (); set_last_target_status (ecs->target, ecs->ptid, ecs->ws); stop_print_frame = true; @@ -9642,6 +9639,9 @@ normal_stop () if (saved_context.changed ()) return true; + if (last.kind () == TARGET_WAITKIND_NO_HISTORY) + interps_notify_no_history (); + /* Notify observers about the stop. This is where the interpreters print the stop event. */ notify_normal_stop ((inferior_ptid != null_ptid From patchwork Tue Mar 12 11:34:23 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Metzger, Markus T" X-Patchwork-Id: 87077 Return-Path: X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id D25B638582BB for ; Tue, 12 Mar 2024 11:36:08 +0000 (GMT) X-Original-To: gdb-patches@sourceware.org Delivered-To: gdb-patches@sourceware.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.12]) by sourceware.org (Postfix) with ESMTPS id 773A9385802A for ; Tue, 12 Mar 2024 11:34:38 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 773A9385802A Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=intel.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org 773A9385802A Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=192.198.163.12 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243283; cv=none; b=BKiaoKYGbeXeBEn9ajr3tRQSTCoPzNJHoB0ysidDCWPzVBiMyDjWfJa8u7B7zoCUVu7VMZQ4ASDg2RQVR0y8USCHibbea2T06LCz32SRPQSAYXK+GN2PZCXh/AkxJBmreQ0EZX0ID9dJyqHaFkr4SeoZo7iktNEtMtOYKBQ+B3Q= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1710243283; c=relaxed/simple; bh=Cr1IA2b2mwB5Ypuqs043rdcC3OCWaZZF/neiZDEqn7w=; h=DKIM-Signature:From:To:Subject:Date:Message-Id:MIME-Version; b=SjMMUTpYlhLfTL88Y/YPY7REq/S2hCmGr5UiLmJur9ZPZRm97gKad7ENAapxGzqaDOj8iqiGu++dtiaoiBX69NNY4/GpOSnP5WzbSjdCXvrq7t8/XgWphQOX1p/oh+nIYUqSw6SYhg/OdyHhaZE8ZLodo2HQpHHelvxzBczNfR4= ARC-Authentication-Results: i=1; server2.sourceware.org DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1710243279; x=1741779279; h=from:to:subject:date:message-id:in-reply-to:references: mime-version:content-transfer-encoding; bh=Cr1IA2b2mwB5Ypuqs043rdcC3OCWaZZF/neiZDEqn7w=; b=SVixUU6fJ4t77WL/K16nG7bBvquJAXQNdkTQK9xnd0aSx0RoomauiNjv R20TEG6rFnWU1n2MXpuhEqyfF9GOtKl5uS+JJ8IA5RM4NVzCMjUAFpYrO Y2jzjWXrYCfI/qfEYFWR6cnGPkk+bb2yClzTMQhIyfiWw9NhpNSWQoLAU c8gZoMnwdA4Vaf/ZBXDBrVTk7BbnD72PicLgAgV2oi1qroj7VSqOJaRno gxckWM/LAhUSiI6Fy/HuE3b1oqQ+5fpAnDi++FtYyf4PfE8J2lBe8hFbe 2DFBA0n5yq31svnEHlmF6pjqL0jI7Gxqw7R3RpA2JjIS8sNSu05i0ALJy w==; X-IronPort-AV: E=McAfee;i="6600,9927,11010"; a="8709978" X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="8709978" Received: from fmviesa005.fm.intel.com ([10.60.135.145]) by fmvoesa106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:38 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.07,119,1708416000"; d="scan'208";a="15999883" Received: from gkldtt-dev-004.igk.intel.com (HELO localhost) ([10.123.221.202]) by fmviesa005-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 12 Mar 2024 04:34:37 -0700 From: Markus Metzger To: gdb-patches@sourceware.org Subject: [PATCH v4 4/4] gdb, infrun: fix multi-threaded reverse stepping Date: Tue, 12 Mar 2024 11:34:23 +0000 Message-Id: <20240312113423.3543956-5-markus.t.metzger@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240312113423.3543956-1-markus.t.metzger@intel.com> References: <20240312113423.3543956-1-markus.t.metzger@intel.com> MIME-Version: 1.0 X-Spam-Status: No, score=-10.1 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_NONE, TXREP, T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.30 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: gdb-patches-bounces+patchwork=sourceware.org@sourceware.org When reverse-stepping a thread that has a pending breakpoint event, the thread is not resumed as part of the infcmd function. A first resume notices the event and returns without resuming the target. If the corresponding breakpoint has been deleted, event processing results in a second resume that performs the intended stepping action. That resume happens after the infcmd function returned and the temporarily modified execution direction was restored. We end up resuming in the wrong direction. Store the direction in a thread's control state and change most of infrun to take it from there rather than using the global variable. --- gdb/gdbthread.h | 10 +++ gdb/infrun.c | 47 ++++++---- gdb/infrun.h | 7 -- .../gdb.btrace/implicit-stop-replaying.exp | 90 +++++++++++++++++++ 4 files changed, 128 insertions(+), 26 deletions(-) create mode 100644 gdb/testsuite/gdb.btrace/implicit-stop-replaying.exp diff --git a/gdb/gdbthread.h b/gdb/gdbthread.h index 4931a0cc2f1..64f9e231f38 100644 --- a/gdb/gdbthread.h +++ b/gdb/gdbthread.h @@ -92,6 +92,13 @@ enum step_over_calls_kind STEP_OVER_UNDEBUGGABLE }; +/* Reverse execution. */ +enum exec_direction_kind + { + EXEC_FORWARD, + EXEC_REVERSE + }; + /* Inferior thread specific part of `struct infcall_control_state'. Inferior process counterpart is `struct inferior_control_state'. */ @@ -176,6 +183,9 @@ struct thread_control_state /* Whether the thread was replaying when the command was issued. */ bool is_replaying = false; + + /* The execution direction when the command was issued. */ + enum exec_direction_kind execution_direction = EXEC_FORWARD; }; /* Inferior thread specific part of `struct infcall_suspend_state'. */ diff --git a/gdb/infrun.c b/gdb/infrun.c index f92f529412f..9db396821d2 100644 --- a/gdb/infrun.c +++ b/gdb/infrun.c @@ -96,7 +96,8 @@ static void insert_step_resume_breakpoint_at_caller (const frame_info_ptr &); static void insert_longjmp_resume_breakpoint (struct gdbarch *, CORE_ADDR); -static bool maybe_software_singlestep (struct gdbarch *gdbarch); +static bool maybe_software_singlestep (const thread_info *tp, + gdbarch *gdbarch, CORE_ADDR pc); static void resume (gdb_signal sig); @@ -2372,11 +2373,12 @@ bool sched_multi = false; GDBARCH the current gdbarch. */ static bool -maybe_software_singlestep (struct gdbarch *gdbarch) +maybe_software_singlestep (const thread_info *tp, gdbarch *gdbarch, + CORE_ADDR pc) { bool hw_step = true; - if (execution_direction == EXEC_FORWARD + if (tp->control.execution_direction == EXEC_FORWARD && gdbarch_software_single_step_p (gdbarch)) hw_step = !insert_single_step_breakpoints (gdbarch); @@ -2527,6 +2529,10 @@ do_target_resume (ptid_t resume_ptid, bool step, enum gdb_signal sig) /* Install inferior's terminal modes. */ target_terminal::inferior (); + scoped_restore save_exec_dir + = make_scoped_restore (&execution_direction, + tp->control.execution_direction); + /* Avoid confusing the next resume, if the next stop/resume happens to apply to another thread. */ tp->set_stop_signal (GDB_SIGNAL_0); @@ -2787,6 +2793,7 @@ resume_1 (enum gdb_signal sig) insert_breakpoints (); resume_ptid = internal_resume_ptid (user_step); + do_target_resume (resume_ptid, false, GDB_SIGNAL_0); tp->set_resumed (true); return; @@ -2836,7 +2843,7 @@ resume_1 (enum gdb_signal sig) set_step_over_info (aspace, regcache_read_pc (regcache), 0, tp->global_num); - step = maybe_software_singlestep (gdbarch); + step = maybe_software_singlestep (tp, gdbarch, pc); insert_breakpoints (); } @@ -2855,7 +2862,7 @@ resume_1 (enum gdb_signal sig) /* Do we need to do it the hard way, w/temp breakpoints? */ else if (step) - step = maybe_software_singlestep (gdbarch); + step = maybe_software_singlestep (tp, gdbarch, pc); /* Currently, our software single-step implementation leads to different results than hardware single-stepping in one situation: when stepping @@ -2926,7 +2933,7 @@ resume_1 (enum gdb_signal sig) else resume_ptid = internal_resume_ptid (user_step); - if (execution_direction != EXEC_REVERSE + if (tp->control.execution_direction != EXEC_REVERSE && step && breakpoint_inserted_here_p (aspace, pc)) { /* There are two cases where we currently need to step a @@ -3095,6 +3102,7 @@ clear_proceed_status_thread (struct thread_info *tp) bpstat_clear (&tp->control.stop_bpstat); tp->control.is_replaying = target_record_is_replaying (tp->ptid); + tp->control.execution_direction = ::execution_direction; } /* Notify the current interpreter and observers that the target is about to @@ -3204,7 +3212,7 @@ schedlock_applies (struct thread_info *tp) && tp->control.stepping_command) || (scheduler_mode == schedlock_replay && target_record_will_replay (minus_one_ptid, - execution_direction))); + tp->control.execution_direction))); } /* Set process_stratum_target::COMMIT_RESUMED_STATE in all target @@ -3629,7 +3637,7 @@ proceed (CORE_ADDR addr, enum gdb_signal siggnal) if (cur_thr->stop_pc_p () && pc == cur_thr->stop_pc () && breakpoint_here_p (aspace, pc) == ordinary_breakpoint_here - && execution_direction != EXEC_REVERSE) + && cur_thr->control.execution_direction != EXEC_REVERSE) /* There is a breakpoint at the address we will resume at, step one instruction before inserting breakpoints so that we do not stop right away (and report a second hit at this @@ -4927,7 +4935,7 @@ adjust_pc_after_break (struct thread_info *thread, breakpoint at PC - 1. We'd then report a hit on B1, although INSN1 hadn't been de-executed yet. Doing nothing is the correct behaviour. */ - if (execution_direction == EXEC_REVERSE) + if (thread->control.execution_direction == EXEC_REVERSE) return; /* If the target can tell whether the thread hit a SW breakpoint, @@ -7526,7 +7534,7 @@ process_event_stop_test (struct execution_control_state *ecs) delete_step_resume_breakpoint (ecs->event_thread); if (ecs->event_thread->control.proceed_to_finish - && execution_direction == EXEC_REVERSE) + && ecs->event_thread->control.execution_direction == EXEC_REVERSE) { struct thread_info *tp = ecs->event_thread; @@ -7541,7 +7549,7 @@ process_event_stop_test (struct execution_control_state *ecs) } fill_in_stop_func (gdbarch, ecs); if (ecs->event_thread->stop_pc () == ecs->stop_func_start - && execution_direction == EXEC_REVERSE) + && ecs->event_thread->control.execution_direction == EXEC_REVERSE) { /* We are stepping over a function call in reverse, and just hit the step-resume breakpoint at the start address of @@ -7666,7 +7674,7 @@ process_event_stop_test (struct execution_control_state *ecs) if (pc_in_thread_step_range (ecs->event_thread->stop_pc (), ecs->event_thread) - && (execution_direction != EXEC_REVERSE + && (ecs->event_thread->control.execution_direction != EXEC_REVERSE || *curr_frame_id == original_frame_id)) { infrun_debug_printf @@ -7685,7 +7693,7 @@ process_event_stop_test (struct execution_control_state *ecs) CORE_ADDR stop_pc = ecs->event_thread->stop_pc (); if (stop_pc == ecs->event_thread->control.step_range_start && stop_pc != ecs->stop_func_start - && execution_direction == EXEC_REVERSE) + && ecs->event_thread->control.execution_direction == EXEC_REVERSE) end_stepping_range (ecs); else keep_going (ecs); @@ -7707,7 +7715,7 @@ process_event_stop_test (struct execution_control_state *ecs) backward through the trampoline code, and that's handled further down, so there is nothing for us to do here. */ - if (execution_direction != EXEC_REVERSE + if (ecs->event_thread->control.execution_direction != EXEC_REVERSE && ecs->event_thread->control.step_over_calls == STEP_OVER_UNDEBUGGABLE && in_solib_dynsym_resolve_code (ecs->event_thread->stop_pc ()) && (ecs->event_thread->control.step_start_function == nullptr @@ -7856,7 +7864,7 @@ process_event_stop_test (struct execution_control_state *ecs) /* Reverse stepping through solib trampolines. */ - if (execution_direction == EXEC_REVERSE + if (ecs->event_thread->control.execution_direction == EXEC_REVERSE && ecs->event_thread->control.step_over_calls != STEP_OVER_NONE && (gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc) || (ecs->stop_func_start == 0 @@ -7884,7 +7892,7 @@ process_event_stop_test (struct execution_control_state *ecs) stepped into (backwards), and continue to there. When we get there, we'll need to single-step back to the caller. */ - if (execution_direction == EXEC_REVERSE) + if (ecs->event_thread->control.execution_direction == EXEC_REVERSE) { /* If we're already at the start of the function, we've either just stepped backward into a single instruction function, @@ -7947,7 +7955,7 @@ process_event_stop_test (struct execution_control_state *ecs) tmp_sal) && !inline_frame_is_marked_for_skip (true, ecs->event_thread)) { - if (execution_direction == EXEC_REVERSE) + if (ecs->event_thread->control.execution_direction == EXEC_REVERSE) handle_step_into_function_backward (gdbarch, ecs); else handle_step_into_function (gdbarch, ecs); @@ -7965,7 +7973,7 @@ process_event_stop_test (struct execution_control_state *ecs) return; } - if (execution_direction == EXEC_REVERSE) + if (ecs->event_thread->control.execution_direction == EXEC_REVERSE) { /* If we're already at the start of the function, we've either just stepped backward into a single instruction function without line @@ -7994,7 +8002,7 @@ process_event_stop_test (struct execution_control_state *ecs) /* Reverse stepping through solib trampolines. */ - if (execution_direction == EXEC_REVERSE + if (ecs->event_thread->control.execution_direction == EXEC_REVERSE && ecs->event_thread->control.step_over_calls != STEP_OVER_NONE) { CORE_ADDR stop_pc = ecs->event_thread->stop_pc (); @@ -8577,6 +8585,7 @@ keep_going_stepped_thread (struct thread_info *tp) tp->set_resumed (true); resume_ptid = internal_resume_ptid (tp->control.stepping_command); + do_target_resume (resume_ptid, false, GDB_SIGNAL_0); } else diff --git a/gdb/infrun.h b/gdb/infrun.h index 6339fd997e1..1a52d6250e5 100644 --- a/gdb/infrun.h +++ b/gdb/infrun.h @@ -107,13 +107,6 @@ extern bool disable_randomization; current location. */ extern ULONGEST get_stop_id (void); -/* Reverse execution. */ -enum exec_direction_kind - { - EXEC_FORWARD, - EXEC_REVERSE - }; - /* The current execution direction. */ extern enum exec_direction_kind execution_direction; diff --git a/gdb/testsuite/gdb.btrace/implicit-stop-replaying.exp b/gdb/testsuite/gdb.btrace/implicit-stop-replaying.exp new file mode 100644 index 00000000000..20240da1dc1 --- /dev/null +++ b/gdb/testsuite/gdb.btrace/implicit-stop-replaying.exp @@ -0,0 +1,90 @@ +# This testcase is part of GDB, the GNU debugger. +# +# Copyright 2024 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 . + +# Test that we stop replaying other threads when stepping a thread at the +# end of its execution history. +# +# This is similar to the last test in multi-thread-step.exp, except that +# we reverse-step instead of record goto begin to start replaying and we +# step instead of continuing. +# +# This triggered a bug where GDB confused the execution direction and kept +# stepping both threads backwards instead of forwards. + +require allow_btrace_tests + +standard_testfile multi-thread-step.c +if [prepare_for_testing "failed to prepare" $testfile $srcfile \ + {debug pthreads}] { + return -1 +} + +if ![runto_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.*" + +# Make sure GDB knows about the new thread. +gdb_test "info threads" +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 "bad" +gdb_test_multiple "thread" "thread" { + -re -wrap "Current thread is \($decimal\).*" { + set thread $expect_out(1,string) + } +} + +# Determine the other thread. +set other "bad" +if { $thread == 1 } { + set other 2 +} elseif { $thread == 2 } { + set other 1 +} + +# This test only works for scheduler-locking 'replay'. +gdb_test_no_output "set scheduler-locking replay" + +# Remove breakpoints or we might run into it right away. +delete_breakpoints + +# Start replaying the other thread. +gdb_test "thread apply $other reverse-stepi" +gdb_test "thread apply $other info record" "Replay in progress.*" \ + "other thread is replaying" + +# Step the thread that reported the breakpoint, which is not replaying. +gdb_test "next" "return arg;.*" + +# The other thread stopped replaying. +gdb_test "thread apply $other info record" "Recorded \[^\\\r\\\n\]*" \ + "other thread stopped replaying"