[5169] in testers
sawfish crash on Sun 9.1
daemon@ATHENA.MIT.EDU (Bill Cattey)
Fri May 17 15:25:11 2002
From: Bill Cattey <wdc@MIT.EDU>
To: testers@mit.edu
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Date: 17 May 2002 15:25:08 -0400
Message-Id: <1021663508.9795.24.camel@tokata.mit.edu>
Mime-Version: 1.0
Often I try and log into my Sun 9.1 system sawfish dies and then the
panel gets rather upset.
The console sez:
Setting up environment...
Starting sawfish window manager...
Initializing settings...
Starting panel...
Creating initial xterm window...
Running standard startup activities...
You have xxx messages (yyyyyyyyy) bytes on PO12.MIT.EDU.
Running custom activities listed in ~/.startup.X...
Assertion failed: !WINDOW_IS_GONE_P (w), file stacking-list.c line 66
rep: received fatal signal: Abort
struct debug_buf common:
Backtrace in 'fatal_signal_handler':
<_setuid+96>
<abort+192>
<_assert+84>
<remove_from_stacking_list+48>
<remove_window+204>
<map_request+1608>
<deregister_event_handler+212>
<rep_call_with_barriar+304>
Lisp backtrace:
Session initialization completed.
----------------
Here is my .startup.X file:
# xsetroot -rv -bitmap ducts.bits
# xsetroot -bitmap bitmaps/pennyfar.xbm
set hostname = `hostname`
if ($host == "hanta-yo.mit.edu") then
alps_modmap
endif
config_console
------------------
I was gonna delete the "config_console" line from my init file, but as a
pre-test, I cannot seem to get the sawfish blow-out to happen again.
(It happened 3 times in a row, and now it's not happening.)
I think this is a slightly more toxic response to the known "Sometimes
your console window does not decorate because of a race condition."
I'm REACHING here, but I'll wager that when I do config_console,
sometimes the race condition bites HARD now.
-wdc
