On Fri, 2020-04-17 at 17:08 +0100, Sam Morris wrote:
> Anyway... I'll keep upgrading things and follow up if I figure out
> which package fixes things...
Upgrading libatspi2.0-0 improved things. Although gnome-shell still
hangs, the backtrace is now:
#0 0x00007ff4b79b8b4f in __GI___poll (fds=fds@entry=0x7fffc4ee7df0,
nfds=nfds@entry=1, timeout=timeout@entry=-1) at
../sysdeps/unix/sysv/linux/poll.c:29
#1 0x00007ff4b467ed97 in poll (__timeout=-1, __nfds=1,
__fds=0x7fffc4ee7df0) at /usr/include/x86_64-linux-gnu/bits/poll2.h:46
#2 read_block (len=8, buf=0x560e38cb2580, fd=53) at ../../src/xcb_in.c:388
#3 _xcb_in_read_block (c=c@entry=0x560e38cb77b0, buf=0x560e38cb2580,
len=len@entry=8) at ../../src/xcb_in.c:1075
#4 0x00007ff4b467cb31 in read_setup (c=0x560e38cb77b0) at
../../src/xcb_conn.c:177
#5 xcb_connect_to_fd (fd=fd@entry=53,
auth_info=auth_info@entry=0x7fffc4ee7f30) at ../../src/xcb_conn.c:359
#6 0x00007ff4b4680ac2 in xcb_connect_to_display_with_auth_info
(displayname=<optimized out>, auth=0x0, screenp=0x0) at
../../src/xcb_util.c:532
#7 0x00007ff48f204b26 in pa_client_conf_from_x11 () from
/usr/lib/x86_64-linux-gnu/pulseaudio/libpulsecommon-12.2.so
#8 0x00007ff48f1c64d8 in pa_client_conf_load () from
/usr/lib/x86_64-linux-gnu/pulseaudio/libpulsecommon-12.2.so
#9 0x00007ff48f24506b in pa_context_new_with_proplist () from
/usr/lib/x86_64-linux-gnu/libpulse.so.0
#10 0x00007ff48f2b2584 in gvc_mixer_new_pa_context
(self=self@entry=0x560e38004f80) at ../subprojects/gvc/gvc-mixer-control.c:3319
#11 0x00007ff48f2b3d0c in gvc_mixer_control_constructor (type=<optimized
out>, n_construct_properties=1, construct_params=0x560e38c76400)
at ../subprojects/gvc/gvc-mixer-control.c:3557
#12 0x00007ff4b87d95fc in g_object_new_with_custom_constructor (n_params=1,
params=0x7fffc4ee8810, class=0x560e38cb1600) at ../../../gobject/gobject.c:1855
#13 g_object_new_internal (class=class@entry=0x560e38cb1600,
params=params@entry=0x7fffc4ee8810, n_params=n_params@entry=1) at
../../../gobject/gobject.c:1935
#14 0x00007ff4b87dae0b in g_object_new_with_properties
(object_type=94619082364016, n_properties=<optimized out>, names=<optimized
out>, values=<optimized out>)
at ../../../gobject/gobject.c:2099
[lots of gjs and mozjs frames]
After upgrading libpulse0 (which pulled in upgrades of all the
installed binary packages from the pulseaudio source package) the
problem is fixed!
So I guess Breaks could be added to gnome-session to make sure
libatk1.0-0 and libpulse0 get pulled in when GNOME 3.6 is installed; on
the other hand, since so few people are likely to run into this problem
and as you say skewed versions aren't really supportable, I won't be
offended if you would prefer to close this bug without bloating gnome-
session's dependencies. :)
I remain faintly puzlled by the final mixture of X11 display sockets:
u_str LISTEN 0 1 /tmp/.X11-unix/X1024 1100515 * 0
users:(("Xwayland",pid=74039,fd=5))
u_str LISTEN 0 1 @/tmp/.X11-unix/X1024 1100514 * 0
users:(("Xwayland",pid=74039,fd=4))
u_str LISTEN 0 1 /tmp/.X11-unix/X1025 1100517 * 0
users:(("gnome-shell",pid=74013,fd=40))
u_str LISTEN 0 1 @/tmp/.X11-unix/X1025 1100516 * 0
users:(("Xwayland",pid=74039,fd=7))
... but things are definitely working again so I'm only faintly
puzzled. :)
Anyway, thanks for the help!
--
Sam Morris <s...@robots.org.uk>
