using xrandr with external monitor - unhandled error

[Jehops]

1. Use an external monitor:
   xrandr --output VGA1 --auto --above LVDS1 --output LVDS1 --auto
2. Stop using external monitor:
   xrandr --output VGA1 --off

This results in the unhandled error pasted at the bottom. I recently updated to 0.9.8-14-g61a7cf2 compiled with sbcl 1.1.12,1 on FreeBSD 9-STABLE amd64 running Xorg-server 1.12.4_4,1. I didn't see this error in the past. If there is any other information I can offer, please let me know.

The value NIL is not of type STUMPWM::FRAME.
Backtrace for: #<SB-THREAD:THREAD "main thread" RUNNING {1003E84023}>
0: ((:METHOD STUMPWM::GROUP-CURRENT-WINDOW (STUMPWM::TILE-GROUP)) #<STUMPWM::TILE-GROUP {10054F7773}>) [fast-method]
1: (STUMPWM::HIDE-WINDOW #S(TILE-WINDOW "New Issue · stumpwm/stumpwm - Conkeror" #x1000070))
2: ((:METHOD STUMPWM::GROUP-REMOVE-HEAD (STUMPWM::TILE-GROUP T)) #<STUMPWM::TILE-GROUP {10054F7773}> #S(frame 1 NIL 0 0 1366 768)) [fast-method]
3: (STUMPWM::REMOVE-HEAD #S<screen #<XLIB:SCREEN :0.0 1366x768x24 TRUE-COLOR>> #S(frame 1 NIL 0 0 1366 768))
4: (STUMPWM::SCALE-SCREEN #S<screen #<XLIB:SCREEN :0.0 1366x768x24 TRUE-COLOR>> (#S(frame 0 NIL 0 0 1366 768)))
5: ((LABELS #:G110 :IN "/usr/home/jrm/scm/stumpwm.git/events.lisp") :STACK-MODE NIL :WINDOW #<XLIB:WINDOW :0 A7> :X 0 :Y 0 :WIDTH 1680 :HEIGHT 1818 :BORDER-WIDTH 0 :VALUE-MASK NIL)
6: (STUMPWM::HANDLE-EVENT :DISPLAY #<XLIB:DISPLAY :0 (The X.Org Foundation R11204000)> :EVENT-KEY :CONFIGURE-NOTIFY :EVENT-CODE 22 :SEND-EVENT-P NIL :SEQUENCE 29768 :EVENT-WINDOW #<XLIB:WINDOW :0 A7> :WINDOW #<XLIB:WINDOW :0 A7> :ABOVE-SIBLING NIL :X 0 :Y 0 :WIDTH 1680 :HEIGHT 1818 :BORDER-WIDTH 0 :OVERRIDE-REDIRECT-P NIL)
7: ((FLET SB-THREAD::WITH-RECURSIVE-LOCK-THUNK :IN XLIB:PROCESS-EVENT))
8: ((FLET SB-IMPL::TIMEOUT-BODY :IN XLIB:PROCESS-EVENT))
9: (XLIB:PROCESS-EVENT #<XLIB:DISPLAY :0 (The X.Org Foundation R11204000)> :HANDLER # :TIMEOUT 0 :PEEK-P NIL :DISCARD-P NIL :FORCE-OUTPUT-P T)
10: (STUMPWM::STUMPWM-INTERNAL-LOOP)
11: (STUMPWM::STUMPWM-INTERNAL ":0")
12: (STUMPWM ":0")
13: ((LAMBDA NIL :IN "/usr/home/jrm/scm/stumpwm.git/make-image.lisp"))
14: ((FLET #:WITHOUT-INTERRUPTS-BODY-54 :IN SB-EXT:SAVE-LISP-AND-DIE))
15: ((LABELS SB-IMPL::RESTART-LISP :IN SB-EXT:SAVE-LISP-AND-DIE))

[dmb2]

I tried to reproduce this, and I couldn't. It looks like something is nil when it should be defined, but without a way to reproduce it myself I can't handle the nil case properly. Is it possible this is a freebsd thing?

[Jehops]

dbjergaard notifications@github.com writes:

I tried to reproduce this, and I couldn't. It looks like something is
nil when it should be defined, but without a way to reproduce it
myself I can't handle the nil case properly. Is it possible this is a
freebsd thing?

I'm not sure. Do you haven any suggestions for pinning this down? Is
there a way to step through the code while the unhandled error occurs?

Joseph

[dmb2]

Please use this issue for further discussion

[theotherjimmy]

I don't think it is a freebsd issue, as this also has affected my archlinux box for something like 8 months. It showed up first when I started using an external monitor.

[erjoalgo]

I am affected by this on Debian. It seems to only happen when there are frames in the monitor that is turned off by xrandr.

[dmb2]

To be clear, if I run the steps to reproduce on an empty desktop, it will work as expected. But if I put windows on the external monitor, then run the xrandr commands, stump will crash?

[dmb2]

Put another way:
Can someone post their configuration ( stumpwmrc, xinitrc, Xsession, etc) with instructions on how X11 is configured to run stumpwm, as well as how they start stumpwm (startx, lightdm, etc). In addition, a set of commands that can be typed in the terminal that reproduces this behavior (included what windows are configured where, use xterm as dummy windows)? I can then replicate the configuration on my machine and (hopefully) reproduce the crash.

I know this seems like a lot to ask, but I can't for the life of me reproduce this on my machines.

[Jehops]

David Bjergaard notifications@github.com writes:

Can someone post their configuration ( stumpwmrc, xinitrc, Xsession, etc) with
instructions on how X11 is configured to run stumpwm, as well as how they start
stumpwm (startx, lightdm, etc). In addition, a set of commands that can be
typed in the terminal that reproduces this behavior (included what windows are
configured where, use xterm as dummy windows)? I can then replicate the
configuration on my machine and (hopefully) reproduce the crash.

Hi,

Whether I start X with startx or with xdm and if I move ~/.stumpwmrc out
of the way and start with a default configuration I can reproduce the
bug following these steps.

Connect an external VGA monitor to my laptop. Run

xrandr --output VGA1 --auto --above LVDS1 --output LVDS1 --auto

then run

xrandr --output VGA1 --off

then do a windowlist.

The error I get, which is different from the one I initially reported
(maybe because I now have no windows open) is:

The value NIL is not of type STUMPWM::FRAME.
Backtrace for: #<SB-THREAD:THREAD "main thread" RUNNING {1003EA4023}>
0: (STUMPWM::FRAME-HEAD #<STUMPWM::TILE-GROUP {1004909873}> NIL)
1: (STUMPWM::SETUP-WIN-GRAVITY #S<screen #<XLIB:SCREEN :0.0 1366x768x24 TRUE-COLOR>> #<XLIB:WINDOW :0 C00004> :TOP-RIGHT)
2: (STUMPWM::SETUP-MESSAGE-WINDOW #S<screen #<XLIB:SCREEN :0.0 1366x768x24 TRUE-COLOR>> 3 495)
3: (STUMPWM::ECHO-STRING-LIST #S<screen #<XLIB:SCREEN :0.0 1366x768x24 TRUE-COLOR>> ("^B^1_Error In Command '^bbanish^B': ^nThe value NIL" " is not of type" " STUMPWM::FRAME."))
4: (STUMPWM::MESSAGE-NO-TIMEOUT "~a" "^B^1_Error In Command '^bbanish^B': ^nThe value NIL
is not of type
STUMPWM::FRAME.")
5: ((LABELS #:G336 :IN "/usr/home/jrm/scm/stumpwm.git/events.lisp") :CODE 28 :STATE 4)
6: (STUMPWM::HANDLE-EVENT :DISPLAY #<XLIB:DISPLAY :0 (The X.Org Foundation R11204000)> :EVENT-KEY :KEY-PRESS :EVENT-CODE 2 :SEND-EVENT-P NIL :CODE 28 :SEQUENCE 577 :TIME 12992248 :ROOT #<XLIB:WINDOW :0 A7> :WINDOW #<XLIB:WINDOW :0 C00002> :EVENT-WINDOW #<XLIB:WINDOW :0 C00002> :CHILD NIL :ROOT-X 683 :ROOT-Y 767 :X 683 :Y 767 :STATE 4 :SAME-SCREEN-P T)
7: ((FLET SB-THREAD::WITH-RECURSIVE-LOCK-THUNK :IN XLIB:PROCESS-EVENT))
8: ((FLET SB-IMPL::TIMEOUT-BODY :IN XLIB:PROCESS-EVENT))
9: (XLIB:PROCESS-EVENT #<XLIB:DISPLAY :0 (The X.Org Foundation R11204000)> :HANDLER # :TIMEOUT 0 :PEEK-P NIL :DISCARD-P NIL :FORCE-OUTPUT-P T)
10: (STUMPWM::STUMPWM-INTERNAL-LOOP)
11: (STUMPWM::STUMPWM-INTERNAL ":0")
12: (STUMPWM ":0")
13: ((LAMBDA NIL :IN "/usr/home/jrm/scm/stumpwm.git/make-image.lisp"))
14: ((FLET #:WITHOUT-INTERRUPTS-BODY-54 :IN SB-EXT:SAVE-LISP-AND-DIE))
15: ((LABELS SB-IMPL::RESTART-LISP :IN SB-EXT:SAVE-LISP-AND-DIE))

The package versions I have running on the laptop are
http://slexy.org/view/s2dp0DoQhJ.

I also have two monitors connected to my desktop at work and I just
tried the same test, but there was no error. The desktop is running the
same stumpwm code, FreeBSD i386 with these packages
http://slexy.org/view/s21sstPKHb. These two monitor are identical. I
wonder if the difference in the aspect ratios between the laptop screen
and the external monitor is relevant?

Just let me know if there is anything else I can test.

Joseph

[dmb2]

Could you please post your .xinitrc or .xsession, as well as steps from turning the computer on to when the crash happens. I gather that this is what is happening:

1. Computer boots up
2. login without graphical login manager
3. type 'startx'
4. stumpwm starts
5. hotplug external monitor
6. type 'xrandr --output VGA1 --auto --above LVDS1 --output LVDS1 --auto' from an xterm
7. move a window onto the external screen
8. type 'xrandr --output VGA1 --off'
9. type 'C-t ; windowlist RET'
10. crash.

At this point, I'm much less interested in the stack trace as I am in reproducing the effect myself.

[Jehops]

David Bjergaard notifications@github.com writes:

Could you please post your .xinitrc or .xsession, ...

Here is /.xinitrc (/.xsession is a symbolic link to ~/.xinitrc):
http://slexy.org/view/s213yfbwGo.

as well as steps from turning
the computer on to when the crash happens. I gather that this is what is
happening:

1. Computer boots up
2. login without graphical login manager
3. type 'startx'

I can 1. log in with xdm or 2. without a login manger then type 'startx. It
makes no difference. The crash happens regardless.

1. stumpwm starts
2. hotplug external monitor

The external monitor is connected with a vga cable, so it makes no
difference if I have it plugged in before I turn the laptop on or after
Xorg has started. The crash happens regardless.

1. type 'xrandr --output VGA1 --auto --above LVDS1 --output LVDS1 --auto' from an xterm
2. move a window onto the external screen

I don't have to open any windows or even switch focus to the external
monitor. In other words, step 7. can be omitted.

1. type 'xrandr --output VGA1 --off'
2. type 'C-t ; windowlist RET'
3. crash.

That's it.

At this point, I'm much less interested in the stack trace as I am in
reproducing the effect myself.

Feel free to ask anything else. I'll do my best to help isolate the
problem.

Joseph

[girzel]

I get this every time, regardless of how many windows are open and where. I have no desktop environment. .xinitrc looks like this:

xsetroot -cursor_name left_ptr

export GTK_IM_MODULE=fcitx

export XMODIFIERS="@im=fcitx"

export QT_IM_MODULE=fcitx

exec /usr/local/bin/stumpwm

I start out with the external monitor plugged in (though it also happens if I plug it in later). I run "startx", X starts and my desktop is mirrored on the larger, external monitor. I open an xterm, and run the shell script ~/.screenlayout/two.sh, which contains:

#!/bin/sh
xrandr --output LVDS1 --mode 1280x800 --pos 1440x100 --rotate normal --output DP1 --off --output VGA1 --mode 1440x900 --pos 0x0 --rotate normal

Then immediately after I run ~/.screenlayout/one.sh, which contains:

#!/bin/sh
xrandr --output LVDS1 --mode 1280x800 --pos 0x0 --rotate normal --output DP1 --off --output VGA1 --off

And it crashes. I haven't opened any windows or even really issued any StumpWM commands at all, apart from that one xterm window.

[dmb2]

OK, I reproduced the issue! I was running unity-settings-daemon which is hijacking some of the XLIB events (it is also causing strange external monitor behavior). That's a separate issue though. For now, I have a procedure to reproduce the bug, and I have the region of the code. Now I just need some time to figure it out. Unless someone else can jump on the hacking wagon. Here's my minimal reproduction steps:

1. Get latest head of stumpwm (confirmed with sbcl 1.1.14.debian)
2. Make/build the head
3. echo "exec /path/to/stumpwm/stumpwm" > ~/.xinitrc #WARNING will remove existing .xinitrc
4. mv .stumpwmrc{,.bak}
5. log out, no external monitors attached (just to make sure lightdm doesn't mess with anything in ubuntu 14.04
6. log in
7. C-t c to open xterm
8. xrandr --output HDMI1 --auto --above LVDS1 --output LVDS1 --auto
9. C-t ; windowlist RET
10. xrandr --output HDMI1 --off
11. Enjoy the crash!

[Jehops]

David Bjergaard notifications@github.com writes:

Closed #71 via 3e9b24b.

Confirmed. Thanks!

Joseph