From patchwork Thu Oct 3 13:51:08 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tom de Vries X-Patchwork-Id: 34817 Received: (qmail 75574 invoked by alias); 3 Oct 2019 13:51:14 -0000 Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org Delivered-To: mailing list gdb-patches@sourceware.org Received: (qmail 75565 invoked by uid 89); 3 Oct 2019 13:51:14 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-25.5 required=5.0 tests=AWL, BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, SPF_PASS autolearn=ham version=3.3.1 spammy= X-HELO: mx1.suse.de Received: from mx2.suse.de (HELO mx1.suse.de) (195.135.220.15) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Thu, 03 Oct 2019 13:51:12 +0000 Received: from relay2.suse.de (unknown [195.135.220.254]) by mx1.suse.de (Postfix) with ESMTP id 94C5CB18C; Thu, 3 Oct 2019 13:51:10 +0000 (UTC) Subject: Re: [PATCH][gdb] Only force INTERP_CONSOLE ui_out for breakpoint commands in MI mode To: Pedro Alves , Andrew Burgess Cc: gdb-patches@sourceware.org, Tom Tromey , Simon Marchi References: <20190905095815.GA27717@delia> <20190924173122.GN4962@embecosm.com> <20190925083642.GO4962@embecosm.com> <6a13b015-6b46-a2a3-0d6b-89a01ef3ab16@suse.de> <6dc5e6b5-4f63-23a8-581b-c78a184d8f08@suse.de> <817a4663-ff93-57dc-d067-e8ee7b8a2804@redhat.com> From: Tom de Vries Openpgp: preference=signencrypt Message-ID: <03ba9add-28c6-ab08-1d96-920d8a57bf67@suse.de> Date: Thu, 3 Oct 2019 15:51:08 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.8.0 MIME-Version: 1.0 In-Reply-To: <817a4663-ff93-57dc-d067-e8ee7b8a2804@redhat.com> X-IsSubscribed: yes On 03-10-19 14:10, Pedro Alves wrote: > On 10/3/19 11:01 AM, Tom de Vries wrote: >> The problem is that the '#0 main () at test.c:19' ends up in the gdb output >> output rather than in gdb.txt. This is due to the fact that the redirect is >> setup for the current ui_out (which is tui->interp_ui_out ()), while the >> backtrace output is printed to the INTERP_CONSOLE ui_out. >> >> Fix this by limiting switching to INTERP_CONSOLE ui_out to when INTERP_MI is >> active. > >> --- a/gdb/cli/cli-script.c >> +++ b/gdb/cli/cli-script.c >> @@ -697,6 +697,9 @@ execute_control_command_1 (struct command_line *cmd, int from_tty) >> enum command_control_type >> execute_control_command (struct command_line *cmd, int from_tty) >> { >> + if (!current_interp_named_p (INTERP_MI)) >> + return execute_control_command_1 (cmd, from_tty); > > Wouldn't we need to handle INTERP_MI1/INTERP_MI2/INTERP_MI3 as well? > > Would > > if (current_uiout->is_mi_like_p ()) > > instead work? Yep, that works as well. Patch updated accordingly and re-tested. OK for trunk? Thanks, - Tom [gdb] Only force INTERP_CONSOLE ui_out for breakpoint commands in MI mode The problem reported in PR mi/25055 is that the output of the backtrace command, when executed as breakpoint command does not show when executing using the MI interpreter: ... $ gdb a.out Reading symbols from a.out... (gdb) break main Breakpoint 1 at 0x4003c0: file test.c, line 19. (gdb) commands Type commands for breakpoint(s) 1, one per line. End with a line saying just "end". >bt >end (gdb) interpreter-exec mi "-exec-run" ^done Breakpoint 1, main () at test.c:19 19 return foo (4); (gdb) ... Interestingly, the function print_frame is called twice during -exec-run: - once during tui_on_normal_stop where the ui_out is temporarily set to tui->interp_ui_out (), resulting in the part after the comma in "Breakpoint 1, main () at test.c:19" - once during execute_control_command, where the ui_out is the default for the current interpreter: mi_ui_out, which ignores calls to output text. The commit 3a87ae656c2 "Use console uiout when executing breakpoint commands" fixes the problem by temporarily switching to the ui_out of INTERP_CONSOLE in execute_control_command. This however caused a regression in redirection (escaping '#' using '\' for git commit message convenience): ... $ rm -f gdb.txt; gdb a.out Reading symbols from a.out... (gdb) break main Breakpoint 1 at 0x4003c0: file test.c, line 19. (gdb) commands Type commands for breakpoint(s) 1, one per line. End with a line saying just "end". >bt >end (gdb) set logging redirect on (gdb) set logging on Redirecting output to gdb.txt. Copying debug output to gdb.txt. (gdb) run \#0 main () at test.c:19 (gdb) q A debugging session is active. Inferior 1 [process 22428] will be killed. Quit anyway? (y or n) y $ cat gdb.txt Starting program: /data/gdb_versions/devel/a.out Breakpoint 1, main () at test.c:19 19 return foo (4); ... The problem is that the '#0 main () at test.c:19' ends up in the gdb output output rather than in gdb.txt. This is due to the fact that the redirect is setup for the current ui_out (which is tui->interp_ui_out ()), while the backtrace output is printed to the INTERP_CONSOLE ui_out. Fix this by limiting switching to INTERP_CONSOLE ui_out to when INTERP_MI is active. Tested on x86_64-linux. gdb/ChangeLog: 2019-10-02 Tom de Vries PR gdb/24956 * cli/cli-script.c (execute_control_command): Only switch to INTERP_CONSOLE's ui_out when INTERP_MI is active. gdb/testsuite/ChangeLog: 2019-10-02 Tom de Vries PR gdb/24956 * gdb.base/ui-redirect.exp: Test output of user-defined command. --- gdb/cli/cli-script.c | 3 +++ gdb/testsuite/gdb.base/ui-redirect.exp | 21 +++++++++++++++++++++ 2 files changed, 24 insertions(+) diff --git a/gdb/cli/cli-script.c b/gdb/cli/cli-script.c index 4fc9c70259c..c460792daff 100644 --- a/gdb/cli/cli-script.c +++ b/gdb/cli/cli-script.c @@ -697,6 +697,9 @@ execute_control_command_1 (struct command_line *cmd, int from_tty) enum command_control_type execute_control_command (struct command_line *cmd, int from_tty) { + if (!current_uiout->is_mi_like_p ()) + return execute_control_command_1 (cmd, from_tty); + /* Make sure we use the console uiout. It's possible that we are executing breakpoint commands while running the MI interpreter. */ interp *console = interp_lookup (current_ui, INTERP_CONSOLE); diff --git a/gdb/testsuite/gdb.base/ui-redirect.exp b/gdb/testsuite/gdb.base/ui-redirect.exp index efcac14b7cc..932c74af4b5 100644 --- a/gdb/testsuite/gdb.base/ui-redirect.exp +++ b/gdb/testsuite/gdb.base/ui-redirect.exp @@ -57,6 +57,24 @@ set cmds "$cmds\n" set outdir [standard_output_file {}] set cmds_file "$outdir/cmds.txt" +with_test_prefix "userdefined" { + set test "define userdefined" + gdb_test_multiple $test $test { + -re "End with a line saying just \"end\"\\.\r\n>$" { + pass $test + } + } + + set test "bt" + gdb_test_multiple $test $test { + -re "\r\n>$" { + pass $test + } + } + + gdb_test_no_output "end" +} + with_test_prefix "logging" { gdb_test_no_output "set logging file /dev/null" gdb_test "set logging on" \ @@ -64,6 +82,7 @@ with_test_prefix "logging" { gdb_test "save breakpoints $cmds_file" "Saved to file '$cmds_file'\\." \ "save breakpoints cmds.txt" cmp_file_string "$cmds_file" "$cmds" "cmds.txt" + gdb_test "userdefined" "#0 main ().*" gdb_test "set logging off" "Done logging to /dev/null\\." gdb_test "help" "List of classes of commands:.*" } @@ -74,6 +93,7 @@ with_test_prefix "redirect" { "Redirecting output to /dev/null.*Copying debug output to /dev/null\\." gdb_test_no_output "save breakpoints $cmds_file" "save breakpoints cmds.txt" cmp_file_string "$cmds_file" "$cmds" "cmds.txt" + gdb_test_no_output "userdefined" gdb_test "set logging off" "Done logging to /dev/null\\." gdb_test "help" "List of classes of commands:.*" } @@ -87,6 +107,7 @@ with_test_prefix "redirect while already logging" { gdb_test "save breakpoints $cmds_file" "Saved to file '$cmds_file'\\." \ "save breakpoints cmds.txt" cmp_file_string "$cmds_file" "$cmds" "cmds.txt" + gdb_test "userdefined" "#0 main ().*" gdb_test "set logging off" "Done logging to /dev/null\\." gdb_test "help" "List of classes of commands:.*" gdb_test_no_output "set logging redirect off"