On Sun, Aug 04, 2019 at 12:32:12PM +0200, Klemens Nanni wrote: > Here is a reproducer as well as a complete backtrace.
Did you try with the updates to libX11 and libXft that I sent to tech@ a few weeks ago ? There is one change in libXft which may be relevant. > > My current setup consists of a X230 on a docking station with one > external monitor connected via DVI-D (shows up as HDMI-2 in xrandr). > > This monitor is a bit special since it lacks proper EDID, so I have to > add modelines manually. The two one-liners are included here for the > sake of completeness, but keep in mind that this crash also happens with > a completely different setup with completely distinct monitors which do > provide EDID and work out of the box. > > Running snapshot from > > OpenBSD 6.5-current (GENERIC.MP) #170: Sat Aug 3 09:50:56 MDT 2019 > [email protected]:/usr/src/sys/arch/amd64/compile/GENERIC.MP > > with `vm.malloc_conf=CFGSU' and cwm, libX11, libXft built with `-g3 -O0'. > > Steps to reproduce: > > 1. add manualmodeline > > xrandr --newmode "2560x1440_30.00" 146.27 2560 2680 2944 3328 1440 > 1441 1444 1465 -HSync +Vsync > xrandr --addmode HDMI-2 2560x1440_30.00 > > 2. enable external monitor, keep internal one as secondary > > xrandr --output HDMI-2 --primary --rotation inverted --above LVDS-1 > --auto --output LVDS-1 > > 3. run a fullscreen program (starts on primary) > > 0ad > > 4. while in fullscreen, press M-/ (Alt slash with default cwm) > > cwm crashes and I drop to the xenodm login. > > Without restarting X itself, the external monitor seemingly keeps its > modelines configured since it continues to show all black, but it is not > configured as part of my screen any longer. > > But strangely, I can see the half of my mouse cursor showing up on it > when moving along. Simply rerunning step 2. to configure the screen > fails: > > xrandr: Output HDMI-2 is not disconnected but has no modes > xrandr: cannot find preferred mode > > Restarting X (read: xenodm) resets everything and I can run steps 1. and > 2. to get everything working again. > > My other setups include two external monitors with the internal one > turned off, I have yet to find a reproducer that does not include a > fullscreen program like 0ad, but it the same crash appeared there as > well. > > See backtraced below and a full version attached. > > [New process 212650] > Core was generated by `cwm'. > Program terminated with signal SIGBUS, Bus error. > #0 0x00000bbbe340a01d in XAddExtension (dpy=0xdfdfdfdfdfdfdfdf) at > /usr/xenocara/lib/libX11/src/InitExt.c:73 > 73 LockDisplay (dpy); > #0 0x00000bbbe340a01d in XAddExtension (dpy=0xdfdfdfdfdfdfdfdf) at > /usr/xenocara/lib/libX11/src/InitExt.c:73 > #1 0x00000bbbd2ff48f9 in _XftDisplayInfoGet (dpy=0xdfdfdfdfdfdfdfdf, > createIfNecessary=1) at /usr/xenocara/lib/libXft/src/xftdpy.c:91 > #2 0x00000bbbd2ff6a19 in XftDrawSrcPicture (draw=0xbbb46fd7600, > color=0xbbbed59cb28) at /usr/xenocara/lib/libXft/src/xftdraw.c:300 > #3 0x00000bbbd2ff6fdb in XftDrawGlyphs (draw=0xbbb46fd7600, > color=0xbbbed59cb28, pub=0xbbbfd21f000, x=0, y=13, glyphs=0x7f7fffff8ad0, > nglyphs=8) at /usr/xenocara/lib/libXft/src/xftdraw.c:484 > #4 0x00000bbbd2ff78ec in XftDrawStringUtf8 (draw=0xbbb46fd7600, > color=0xbbbed59cb28, pub=0xbbbfd21f000, x=0, y=13, string=0x7f7fffff9eeb "", > len=0) at /usr/xenocara/lib/libXft/src/xftdraw.c:621 > #5 0x00000bb9389bdfdd in menu_draw (mc=0x7f7fffff9df0, menuq=0x7f7fffffa1c8, > resultq=0x7f7fffff9de0) at /x/app/cwm/menu.c:399 > #6 0x00000bb9389bd12f in menu_filter (sc=0xbbbed59ca00, > menuq=0x7f7fffffa1c8, prompt=0xbb9389b349f "window", initial=0x0, flags=0, > match=0xbb9389bef00 <search_match_client>, print=0xbb9389bfb20 > <search_print_client>) at /x/app/cwm/menu.c:153 > #7 0x00000bb9389c7a16 in kbfunc_menu_client (ctx=0xbbbed59ca00, > cargs=0xbbb9b36c780) at /x/app/cwm/kbfunc.c:484 > #8 0x00000bb9389c3ffa in xev_handle_keypress (ee=0x7f7fffffa2a8) at > /x/app/cwm/xevents.c:336 > #9 0x00000bb9389c519d in xev_process () at /x/app/cwm/xevents.c:491 > #10 0x00000bb9389b763f in main (argc=0, argv=0x7f7fffffa450) at > /x/app/cwm/calmwm.c:114 -- Matthieu Herrb
