unofficial mirror of emacs-devel@gnu.org 
 help / color / mirror / code / Atom feed
From: Ken Brown <kbrown@cornell.edu>
To: emacs-devel@gnu.org, Po Lu <luangruo@yahoo.com>
Subject: Re: master 24d01af3dd: Cancel DND correctly during display disconnect on a different display
Date: Thu, 26 May 2022 12:34:16 -0400	[thread overview]
Message-ID: <ba9d463d-8b15-3894-4504-ec1a5b2f9aa1@cornell.edu> (raw)
In-Reply-To: <2bd53fb7-8699-6a4c-cc65-4d7173341318@cornell.edu>

On 5/26/2022 10:34 AM, Ken Brown wrote:
> 
> 
> On 4/28/2022 9:29 PM, Po Lu via Mailing list for Emacs changes wrote:
>> branch: master
>> commit 24d01af3dde22ccd03b614d981129c6eacb0d56e
>> Author: Po Lu <luangruo@yahoo.com>
>> Commit: Po Lu <luangruo@yahoo.com>
>>
>>      Cancel DND correctly during display disconnect on a different display
>>      * src/xterm.c (x_dnd_io_error_handler): New function.
>>      (x_connection_closed): Send the messages to cancel DND if a
>>      different display was disconnected.
>> ---
>>   src/xterm.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++++++-------
>>   1 file changed, 70 insertions(+), 8 deletions(-)
>>
>> diff --git a/src/xterm.c b/src/xterm.c
>> index c4e0305a6e..35f6bc7fe8 100644
>> --- a/src/xterm.c
>> +++ b/src/xterm.c
>> @@ -1005,6 +1005,10 @@ static int x_dnd_movement_x, x_dnd_movement_y;
>>   static unsigned int x_dnd_keyboard_state;
>>   #endif
>> +/* jmp_buf that gets us out of the IO error handler if an error occurs
>> +   terminating DND as part of the display disconnect handler.  */
>> +static jmp_buf x_dnd_disconnect_handler;
>> +
>>   struct x_client_list_window
>>   {
>>     Window window;
>> @@ -2571,8 +2575,17 @@ x_dnd_compute_toplevels (struct x_display_info *dpyinfo)
>>     return 0;
>>   }
>> -#define X_DND_SUPPORTED_VERSION 5
>> +static _Noreturn int
>> +x_dnd_io_error_handler (Display *display)
>> +{
>> +#ifdef USE_GTK
>> +  emacs_abort ();
>> +#else
>> +  longjmp (x_dnd_disconnect_handler, 1);
>> +#endif
>> +}
>> +#define X_DND_SUPPORTED_VERSION 5
>>   static int x_dnd_get_window_proto (struct x_display_info *, Window);
>>   static Window x_dnd_get_window_proxy (struct x_display_info *, Window);
>> @@ -20027,6 +20040,9 @@ x_connection_closed (Display *dpy, const char 
>> *error_message, bool ioerror)
>>     struct x_display_info *dpyinfo = x_display_info_for_display (dpy);
>>     Lisp_Object frame, tail;
>>     specpdl_ref idx = SPECPDL_INDEX ();
>> +  void *io_error_handler;
>> +  xm_drop_start_message dmsg;
>> +  struct frame *f;
>>     error_msg = alloca (strlen (error_message) + 1);
>>     strcpy (error_msg, error_message);
>> @@ -20034,16 +20050,61 @@ x_connection_closed (Display *dpy, const char 
>> *error_message, bool ioerror)
>>     /* Inhibit redisplay while frames are being deleted. */
>>     specbind (Qinhibit_redisplay, Qt);
>> -  /* If drag-and-drop is in progress and the DND frame's display is
>> -     DPY, cancel drag-and-drop.  Don't reset event masks or try to
>> -     send responses to other programs because the display is going
>> +  /* If drag-and-drop is in progress, cancel drag-and-drop.  If DND
>> +     frame's display is DPY, don't reset event masks or try to send
>> +     responses to other programs because the display is going
>>        away.  */
>> -  if ((x_dnd_in_progress || x_dnd_waiting_for_finish)
>> -      && dpy == (x_dnd_waiting_for_finish
>> -         ? x_dnd_finish_display
>> -         : FRAME_X_DISPLAY (x_dnd_frame)))
>> +  if (x_dnd_in_progress || x_dnd_waiting_for_finish)
>>       {
>> +      /* Handle display disconnect errors here because this function
>> +     is not reentrant at this particular spot.  */
>> +      io_error_handler = XSetIOErrorHandler (x_dnd_io_error_handler);
>> +
>> +      if (!sigsetjmp (x_dnd_disconnect_handler, 1)
>> +      && x_dnd_in_progress
>> +      && dpy != (x_dnd_waiting_for_finish
>> +             ? x_dnd_finish_display
>> +             : FRAME_X_DISPLAY (x_dnd_frame)))
>> +    {
>> +      /* Clean up drag and drop if the drag frame's display isn't
>> +         the one being disconnected.  */
>> +      f = x_dnd_frame;
>> +
>> +      if (x_dnd_last_seen_window != None
>> +          && x_dnd_last_protocol_version != -1)
>> +        x_dnd_send_leave (x_dnd_frame,
>> +                  x_dnd_last_seen_window);
>> +      else if (x_dnd_last_seen_window != None
>> +           && !XM_DRAG_STYLE_IS_DROP_ONLY (x_dnd_last_motif_style)
>> +           && x_dnd_last_motif_style != XM_DRAG_STYLE_NONE
>> +           && x_dnd_motif_setup_p)
>> +        {
>> +          dmsg.reason = XM_DRAG_REASON (XM_DRAG_ORIGINATOR_INITIATOR,
>> +                        XM_DRAG_REASON_DROP_START);
>> +          dmsg.byte_order = XM_BYTE_ORDER_CUR_FIRST;
>> +          dmsg.timestamp = FRAME_DISPLAY_INFO (f)->last_user_time;
>> +          dmsg.side_effects
>> +        = XM_DRAG_SIDE_EFFECT (xm_side_effect_from_action (FRAME_DISPLAY_INFO 
>> (f),
>> +                                   x_dnd_wanted_action),
>> +                       XM_DROP_SITE_VALID,
>> +                       xm_side_effect_from_action (FRAME_DISPLAY_INFO (f),
>> +                                   x_dnd_wanted_action),
>> +                       XM_DROP_ACTION_DROP_CANCEL);
>> +          dmsg.x = 0;
>> +          dmsg.y = 0;
>> +          dmsg.index_atom = FRAME_DISPLAY_INFO (f)->Xatom_XdndSelection;
>> +          dmsg.source_window = FRAME_X_WINDOW (f);
>> +
>> +          x_dnd_send_xm_leave_for_drop (FRAME_DISPLAY_INFO (f), f,
>> +                        x_dnd_last_seen_window, 0);
>> +          xm_send_drop_message (FRAME_DISPLAY_INFO (f), FRAME_X_WINDOW (f),
>> +                    x_dnd_last_seen_window, &dmsg);
>> +        }
>> +    }
>> +
>> +      XSetIOErrorHandler (io_error_handler);
>> +
>>         x_dnd_last_seen_window = None;
>>         x_dnd_last_seen_toplevel = None;
>>         x_dnd_in_progress = false;
>> @@ -20234,6 +20295,7 @@ x_io_error_quitter (Display *display)
>>           DisplayString (display));
>>     x_connection_closed (display, buf, true);
>>   }
>> +
>>   \f
>>   /* Changing the font of the frame.  */
>>
> 
> After this commit I'm getting the following warning on 64-bit Cygwin:
> 
> ../../master/src/xterm.c:20064:12: warning: initialization of ‘long int (*)[34]’ 
> from incompatible pointer type ‘long int (*)[32]’ [-Wincompatible-pointer-types]
> 20064 |       if (!sigsetjmp (x_dnd_disconnect_handler, 1)
>        |            ^~~~~~~~~
> 
> A second such warning appeared after commit cc297af:
> 
> ../../master/src/xterm.c: In function ‘xm_get_drag_window’:
> ../../master/src/xterm.c:1447:11: warning: initialization of ‘long int (*)[34]’ 
> from incompatible pointer type ‘long int (*)[32]’ [-Wincompatible-pointer-types]
>   1447 |       if (sigsetjmp (x_dnd_disconnect_handler, 1))
>        |           ^~~~~~~~~
> 
> 
> With current git HEAD, the line numbers of the sigsetjmp calls are 1664 and 21315.

The problem seems to be that x_dnd_disconnect_handler is of type jmp_buf instead 
of sigjmp_buf.  But we can't simply change the type, because 
x_dnd_disconnect_handler is also used in a call to longjmp, which expects an 
argument of type jmp_buf.

Ken



  reply	other threads:[~2022-05-26 16:34 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <165119574603.20331.4036366960372072878@vcs2.savannah.gnu.org>
     [not found] ` <20220429012907.00486C01683@vcs2.savannah.gnu.org>
2022-05-26 14:34   ` master 24d01af3dd: Cancel DND correctly during display disconnect on a different display Ken Brown
2022-05-26 16:34     ` Ken Brown [this message]
2022-05-27  0:10       ` Po Lu

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

  List information: https://www.gnu.org/software/emacs/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ba9d463d-8b15-3894-4504-ec1a5b2f9aa1@cornell.edu \
    --to=kbrown@cornell.edu \
    --cc=emacs-devel@gnu.org \
    --cc=luangruo@yahoo.com \
    /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.
Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/emacs.git

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).