On 2018/10/03 01:05, Victor Kukshiev wrote:
> Hello, ports@!
>
> This diff fixes crash x11/sakura port with vte 0.54.
> upstream commit:
> https://bazaar.launchpad.net/~dabisu/sakura/sakura/revision/596
> ok?
I've included the diff below in a more standard format (the patch file
should be created with 'make update-patches' rather than come directly
from upstream repo, and have a bit of information about where it came
from).
I just tried sakura to test this, but it's not working at all well
for me, either with or without this diff. (And I don't think the diff
matters until the vte3 port is updated anyway).
About 75% of the time when I try to run it I briefly get a window
displayed, then it closes and the process exits, no crash/error message
(there is a warning about PCRE2_MULTILINE, but this happens whether it
opens the terminal or not).
If I am able to open a terminal, when I then open a new tab, there is a
high chance of getting the following
(sakura:58584): Gtk-WARNING **: 12:17:39.202: VteTerminal 0x973b4322390
reported min size 21 and natural size 2 in get_preferred_height(); natural size
must be >= min size
(sakura:58584): Gtk-WARNING **: 12:17:39.202: VteTerminal 0x973b4322390
reported min size 12 and natural size 2 in get_preferred_width(); natural size
must be >= min size
Segmentation fault (core dumped)
Program terminated with signal SIGSEGV, Segmentation fault.
#0 0x00000973bceee6ae in VteTerminalPrivate::insert_char(unsigned int, bool,
bool) ()
from /usr/local/lib/libvte-2.91.so.2.2
[Current thread is 1 (process 401647)]
(gdb) bt
#0 0x00000973bceee6ae in VteTerminalPrivate::insert_char(unsigned int, bool,
bool) ()
from /usr/local/lib/libvte-2.91.so.2.2
#1 0x00000973bceefc46 in VteTerminalPrivate::process_incoming() () from
/usr/local/lib/libvte-2.91.so.2.2
#2 0x00000973bcf000fa in VteTerminalPrivate::process(bool) () from
/usr/local/lib/libvte-2.91.so.2.2
#3 0x00000973bcf00e64 in update_repeat_timeout(void*) () from
/usr/local/lib/libvte-2.91.so.2.2
#4 0x00000973634c1433 in g_timeout_dispatch (source=0x973b4066e80,
callback=0x400000202000100, user_data=0x0)
at gmain.c:4649
#5 0x00000973634c5899 in g_main_dispatch (context=<optimized out>) at
gmain.c:3176
#6 g_main_context_dispatch (context=<optimized out>) at gmain.c:3829
#7 0x00000973634c5ca3 in g_main_context_iterate (context=<optimized out>,
block=<optimized out>, dispatch=<optimized out>,
self=<optimized out>) at gmain.c:3902
#8 0x00000973634c609f in g_main_loop_run (loop=0x973c33bfeb0) at gmain.c:4098
#9 0x000009739e203e58 in gtk_main () at gtkmain.c:1323
#10 0x00000970e9301a3c in main ()
It's not a general vte3 problem on this machine, for example terminator
runs just fine.
Index: Makefile
===================================================================
RCS file: /cvs/ports/x11/sakura/Makefile,v
retrieving revision 1.24
diff -u -p -r1.24 Makefile
--- Makefile 27 Jun 2018 21:04:22 -0000 1.24
+++ Makefile 3 Oct 2018 11:13:43 -0000
@@ -5,7 +5,7 @@ COMMENT = GTK and VTE based terminal emu
DISTNAME = sakura-${V}
EXTRACT_SUFX = .tar.bz2
CATEGORIES = x11
-REVISION = 1
+REVISION = 2
HOMEPAGE = http://www.pleyades.net/david/projects/sakura/
MAINTAINER = Victor Kukshiev <[email protected]>
Index: patches/patch-src_sakura_c
===================================================================
RCS file: patches/patch-src_sakura_c
diff -N patches/patch-src_sakura_c
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-src_sakura_c 3 Oct 2018 11:13:46 -0000
@@ -0,0 +1,66 @@
+$OpenBSD$
+
+Fix with newer vte.
+https://bugs.launchpad.net/sakura/+bug/1790317
+https://bazaar.launchpad.net/~dabisu/sakura/sakura/revision/596
+
+Index: src/sakura.c
+--- src/sakura.c.orig
++++ src/sakura.c
+@@ -339,6 +339,7 @@ struct terminal {
+ bool label_set_byuser;
+ GtkBorder padding; /* inner-property data */
+ int colorset;
++ gulong exit_handler_id;
+ };
+
+
+@@ -984,31 +985,6 @@ sakura_eof (GtkWidget *widget, void *data)
+ if (npages==1) {
+ sakura_config_done();
+ }
+-
+- /* Workaround for libvte strange behaviour. There is not child-exited
signal for
+- the last terminal, so we need to kill it here. Check with libvte
authors about
+- child-exited/eof signals */
+- if (gtk_notebook_get_current_page(GTK_NOTEBOOK(sakura.notebook))==0) {
+-
+- term = sakura_get_page_term(sakura, 0);
+-
+- if (option_hold==TRUE) {
+- SAY("hold option has been activated");
+- return;
+- }
+-
+- //SAY("waiting for terminal pid (in eof) %d", term->pid);
+- //waitpid(term->pid, &status, WNOHANG);
+- /* TODO: check wait return */
+- /* Child should be automatically reaped because we don't use
G_SPAWN_DO_NOT_REAP_CHILD flag */
+- g_spawn_close_pid(term->pid);
+-
+- sakura_del_tab(0);
+-
+- npages =
gtk_notebook_get_n_pages(GTK_NOTEBOOK(sakura.notebook));
+- if (npages==0)
+- sakura_destroy();
+- }
+ }
+
+ /* This handler is called when window title changes, and is used to change
window and notebook pages titles */
+@@ -3167,7 +3143,7 @@ sakura_add_tab()
+ g_signal_connect(G_OBJECT(term->vte), "bell", G_CALLBACK(sakura_beep),
NULL);
+ g_signal_connect(G_OBJECT(term->vte), "increase-font-size",
G_CALLBACK(sakura_increase_font), NULL);
+ g_signal_connect(G_OBJECT(term->vte), "decrease-font-size",
G_CALLBACK(sakura_decrease_font), NULL);
+- g_signal_connect(G_OBJECT(term->vte), "child-exited",
G_CALLBACK(sakura_child_exited), NULL);
++ term->exit_handler_id = g_signal_connect(G_OBJECT(term->vte),
"child-exited", G_CALLBACK(sakura_child_exited), NULL);
+ g_signal_connect(G_OBJECT(term->vte), "eof", G_CALLBACK(sakura_eof),
NULL);
+ g_signal_connect(G_OBJECT(term->vte), "window-title-changed",
G_CALLBACK(sakura_title_changed), NULL);
+ g_signal_connect_swapped(G_OBJECT(term->vte), "button-press-event",
G_CALLBACK(sakura_button_press), sakura.menu);
+@@ -3366,6 +3342,7 @@ sakura_del_tab(gint page)
+ }
+
+ gtk_widget_hide(term->hbox);
++ g_signal_handler_disconnect (term->vte, term->exit_handler_id);
+ gtk_notebook_remove_page(GTK_NOTEBOOK(sakura.notebook), page);
+
+ /* Find the next page, if it exists, and grab focus */
