From: Andrew Burgess <aburgess@redhat.com>
To: Tom de Vries <tdevries@suse.de>, gdb-patches@sourceware.org
Subject: Re: [PATCH] [gdb/tui] Make tui_setup_io more robust
Date: Thu, 02 Apr 2026 10:18:14 +0100 [thread overview]
Message-ID: <87fr5d67qh.fsf@redhat.com> (raw)
In-Reply-To: <20260221153831.2743758-1-tdevries@suse.de>
Tom de Vries <tdevries@suse.de> writes:
> The following sequence of events may happen when enabling TUI:
> - tui_enable () is called,
> - a SIGFPE happens, before tui_setup_io (1) is called,
> - the SIGFPE triggers handle_fatal_signal, which calls tui_disable (),
> - during tui_disable (), tui_setup_io (0) is called, and
> - tui_setup_io (0) tries to restore things that were saved during a previous
> tui_setup_io (1) call, but tui_setup_io (1) was never called so the saving
> never happened.
>
> This can cause current_ui.m_ui_stderr to be nullptr, which then can cause a
> crash in sig_write (PR33918).
If this bug number is worth mentioning, then we should add a 'Bug:...'
tag to the commit message. I like them if for nothing else they are
clickable in my terminal :)
>
> Fix this by:
> - adding a variable tui_io_mode, and
> - using the variable to bail out of tui_setup_io in case the current mode
> doesn't change.
What's one more global, the tui already has so many!
This seems a reasonable change to me.
Approved-By: Andrew Burgess <aburgess@redhat.com>
Thanks,
Andrew
>
> Tested on x86_64-linux.
> ---
> gdb/tui/tui-io.c | 12 ++++++++++++
> 1 file changed, 12 insertions(+)
>
> diff --git a/gdb/tui/tui-io.c b/gdb/tui/tui-io.c
> index f673fbf36f6..5a3bc604592 100644
> --- a/gdb/tui/tui-io.c
> +++ b/gdb/tui/tui-io.c
> @@ -824,6 +824,10 @@ tui_rl_display_match_list (char **matches, int len, int max)
> gdb_display_match_list (matches, len, max, &displayer);
> }
>
> +/* Whether the IO is setup for curses (1) or non-curses (0). */
> +
> +static int tui_io_mode;
> +
> /* Setup the IO for curses or non-curses mode.
> - In non-curses mode, readline and gdb use the standard input and
> standard output/error directly.
> @@ -837,6 +841,14 @@ tui_setup_io (int mode)
> {
> extern int _rl_echoing_p;
>
> + if (tui_io_mode == mode)
> + {
> + /* Nothing to do. Also, this makes sure that we don't try to restore
> + things before having saved them. */
> + return;
> + }
> + tui_io_mode = mode;
> +
> if (mode)
> {
> /* Ensure that readline has been initialized before saving any
>
> base-commit: 646982f4295bf0a1e64867606d496c34c9a15a0c
> --
> 2.51.0
next prev parent reply other threads:[~2026-04-02 9:18 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-21 15:38 Tom de Vries
2026-04-02 9:18 ` Andrew Burgess [this message]
2026-04-02 21:22 ` Tom de Vries
2026-04-03 10:31 ` Tom de Vries
2026-04-03 11:35 ` Tom de Vries
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=87fr5d67qh.fsf@redhat.com \
--to=aburgess@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=tdevries@suse.de \
/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