xcb_xlib_lock assertion fails when using DMX from X11R7.2-RC3

[Andrew Shewmaker]

I tested Xdmx from R7.2-RC3 with the following command
(after starting another X server on display :1).

Xdmx :2 -display :1 -input :0

This command works with the R6.8.2 included in RHEL4 Update 4,
but it aborts with Xdmx from R7.2-RC3.

(II) dmx[o0/:1]: DPMS 1.1 (on, enabled, 0 0 0)
(II) dmx[o0/:1]: (request) s=1920x1200+0+0 r=1920x1200+0+0 @0,0 (0)
(be=1920x1200 depth=24 bpp=32)
(II) dmx[o0/:1]: s=1920x1200+0+0 r=1920x1200+0+0 @0,0 (be=1920x1200
depth=24 bpp=32)
(II) dmx: Using 1920x1200 as global bounding box
(II) dmx: XSync batching with 100 ms interval
(II) dmx: Shadow framebuffer support disabled
(..) dmx: No atom on :0
(II) dmx[i0/:0]: Using console input from :0 (with windows)
Xdmx: xcb_xlib.c:41: xcb_xlib_lock: Assertion `!c->xlib.lock' failed.
Aborted

Configs that are more realistic behave similarly.

# xdmx.conf
virtual screen4 {
  option +xinerama;
  option -fontpath /usr/X11R6/lib/X11/fonts/misc;
  option -input :0;
  wall 4x1 1600x1200
       dq014:0 dq015:0  dq016:0 dq017:0;
}
# end xdmx.conf

$ Xdmx :1 -configfile xdmx.conf -config screen4

...
II) dmx: Using 6400x1200 as global bounding box
(II) dmx: XSync batching with 100 ms interval
(II) dmx: Shadow framebuffer support disabled
(..) dmx: No atom on :0
(..) dmx: No atom on :0
(..) dmx: No atom on :0
(..) dmx: No atom on :0
(II) dmx[i0/:0]: Using console input from :0 (with windows)
Xdmx.stat: xcb_xlib.c:41: xcb_xlib_lock: Assertion `!c->xlib.lock' failed.
Aborted


Output from the working version looks like:

(II) dmx: Using 6400x1200 as global bounding box
(II) dmx: XSync batching with 100 ms interval
(II) dmx: Shadow framebuffer support disabled
(II) dmx: screen0 FBConfig 0x79 matched to  0x79 on screen#1
(II) dmx: screen0 FBConfig 0x79 matched to  0x79 on screen#2
(II) dmx: screen0 FBConfig 0x79 matched to  0x79 on screen#3
...
(..) dmx: No atom on :0
(..) dmx: No atom on :0
(..) dmx: No atom on :0
(..) dmx: No atom on :0
(II) dmx[i0/:0]: Using console input from :0 (with windows)
(II) dmx[i0/:0]: Locating devices on :0 (XInputExtension version 1.3)
(II) dmx[i0/:0]:    0 DevInputMi XExtensionDevice
(II) dmx[i0/:0]:    1 Keyboard0  XKeyboard
(II) dmx[i0/:0]:    2 Mouse0     XPointer
(II) dmx[i0/:0]:    3 NVIDIA Eve XExtensionDevice
(II) dmx[i0/:0]: Added console-mou as pointer device called Mouse0 [core]
(II) dmx[i0/:0]: Added console-kbd as keyboard device called Keyboard0 [core]
(II) dmx[i0/:0]: Added common-oth as extension device called Other0
(II) dmx[i0/:0]: Added common-oth as extension device called Other1
(II) dmx[i0/:0]: Using Keyboard0 and Mouse0 as true core devices
(II) dmx[i0/:0]: XKEYBOARD: keycodes = xfree86+aliases(qwerty)
(II) dmx[i0/:0]: XKEYBOARD: symbols  = pc/pc(pc105)+pc/us
(II) dmx[i0/:0]: XKEYBOARD: geometry = (null)
(II) dmx[i0/:0]: XKEYBOARD: From device: xfree86 pc/pc(pc105)+pc/us
(II) dmx: ===== Start of Summary =====
(II) dmx: 4 screens configured with Xinerama (6400 1200)
(II) dmx: 88 visuals at depth 24:
(II) dmx:   0x23 TrueColor   24b 10b/rgb 256 0xff0000 0xff00 0x00ff
(II) dmx:   0x24 TrueColor   24b 10b/rgb 256 0xff0000 0xff00 0x00ff
(II) dmx:   0x25 TrueColor   24b 10b/rgb 256 0xff0000 0xff00 0x00ff
...
(II) dmx:   0x4f DirectColor 24b 10b/rgb 256 0xff0000 0xff00 0x00ff
(II) dmx:   0x50 DirectColor 24b 10b/rgb 256 0xff0000 0xff00 0x00ff
(II) dmx:   0x51 DirectColor 24b 10b/rgb 256 0xff0000 0xff00 0x00ff
...
(II) dmx: 0 visuals at depth 1:
(II) dmx: 0 visuals at depth 4:
(II) dmx: 0 visuals at depth 8:
(II) dmx: 0 visuals at depth 15:
(II) dmx: 0 visuals at depth 16:
(II) dmx: 0 visuals at depth 32:
(II) dmx: 4 devices:
(II) dmx:   Id  Name                 Classes
(II) dmx:    0  Mouse0               val btn fb/ptr     [i0/:0/id2=Mouse0] core
(II) dmx:    1  Keyboard0            key foc fb/kbd
[i0/:0/id1=Keyboard0] core
(II) dmx:    2  Other0               val btn fb/ptr
[i0/:0/id0=DevInputMice] extension
(II) dmx:    3  Other1               (none)     [i0/:0/id3=NVIDIA
Event Handler] extension
(II) dmx: ===== End of Summary =====


Thanks,

Andrew Shewmaker