61936 - Dropping items from other screen causes Mozilla to "crash"...

Status: RESOLVED WONTFIX

(Core :: XUL, defect, P3)

Description

[Roland Mainz]

On an Unix system (Solaris+Linux) with multihead (two screens) Mozilla M18
"exists" (crash-like) if I drag&drop something from another application (for
example from Netscape 4.x or a selected text within a Motif app. dragged with
the middle mouse).
Last message of Mozilla looks like this:
-- snip --
Enabling Quirk StyleSheet
Enabling Quirk StyleSheet
Document http://puck/medmikrobio/service/ loaded successfully

Gdk-WARNING **: Invalid target index in TOP_LEVEL_ENTER MESSAGE

Gdk-WARNING **: Invalid target index in TOP_LEVEL_ENTER MESSAGE
X Error of failed request:  BadAtom (invalid Atom parameter)
  Major opcode of failed request:  23 (X_GetSelectionOwner)
  Atom id in failed request:  0x0
  Serial number of failed request:  151947
  Current serial number in output stream:  151947
-- snip --

Looks like the usual "GTK+ cannot handle multiple screens"-issue... ;-(

xdpyinfo output:
-- snip --
% xdpyinfo
name of display:    :0.1
version number:    11.0
vendor string:    Sun Microsystems, Inc.
vendor release number:    3610
maximum request size:  262140 bytes
motion buffer size:  256
bitmap unit, bit order, padding:    32, MSBFirst, 32
image byte order:    MSBFirst
number of supported pixmap formats:    4
supported pixmap formats:
    depth 1, bits_per_pixel 1, scanline_pad 32
    depth 8, bits_per_pixel 8, scanline_pad 32
    depth 16, bits_per_pixel 16, scanline_pad 32
    depth 24, bits_per_pixel 32, scanline_pad 32
keycode range:    minimum 8, maximum 132
focus:  window 0x680000b, revert to Parent
number of extensions:    28
    AccessX
    Adobe-DPS-Extension
    DOUBLE-BUFFER
    DPMS
    DPSExtension
    Extended-Visual-Information
    GLX
    LBX
    MIT-SCREEN-SAVER
    MIT-SHM
    MIT-SUNDRY-NONSTANDARD
    Multi-Buffering
    RECORD
    SECURITY
    SHAPE
    SUN_ALLPLANES
    SUN_DGA
    SUN_OVL
    SUN_SME
    SYNC
    SolarisIA
    TOG-CUP
    XC-APPGROUP
    XC-MISC
    XIE
    XInputDeviceEvents
    XInputExtension
    XTEST
default screen number:    1
number of screens:    2

screen #0:
  dimensions:    1152x900 pixels (344x269 millimeters)
  resolution:    85x85 dots per inch
  depths (2):    1, 24
  root window id:    0x2c
  depth of root window:    24 planes
  number of colormaps:    minimum 1, maximum 1
  default colormap:    0x21
  default number of colormap cells:    256
  preallocated pixels:    black 0, white 16777215
  options:    backing-store YES, save-unders YES
  largest cursor:    unlimited
  current input event mask:    0x78203f
    KeyPressMask             KeyReleaseMask           ButtonPressMask          
    ButtonReleaseMask        EnterWindowMask          LeaveWindowMask          
    ButtonMotionMask         SubstructureNotifyMask   SubstructureRedirectMask 
    FocusChangeMask          PropertyChangeMask       
  number of visuals:    1
  default visual id:  0x22
  visual:
    visual id:    0x22
    class:    TrueColor
    depth:    24 planes
    available colormap entries:    256 per subfield
    red, green, blue masks:    0xff0000, 0xff00, 0xff
    significant bits in color specification:    8 bits

screen #1:
  dimensions:    1152x900 pixels (344x269 millimeters)
  resolution:    85x85 dots per inch
  depths (2):    1, 8
  root window id:    0x2e
  depth of root window:    8 planes
  number of colormaps:    minimum 1, maximum 1
  default colormap:    0x23
  default number of colormap cells:    256
  preallocated pixels:    black 1, white 0
  options:    backing-store YES, save-unders YES
  largest cursor:    unlimited
  current input event mask:    0x78203f
    KeyPressMask             KeyReleaseMask           ButtonPressMask          
    ButtonReleaseMask        EnterWindowMask          LeaveWindowMask          
    ButtonMotionMask         SubstructureNotifyMask   SubstructureRedirectMask 
    FocusChangeMask          PropertyChangeMask       
  number of visuals:    6
  default visual id:  0x24
  visual:
    visual id:    0x24
    class:    PseudoColor
    depth:    8 planes
    available colormap entries:    256
    red, green, blue masks:    0x0, 0x0, 0x0
    significant bits in color specification:    8 bits
  visual:
    visual id:    0x25
    class:    DirectColor
    depth:    8 planes
    available colormap entries:    8 per subfield
    red, green, blue masks:    0x7, 0x38, 0xc0
    significant bits in color specification:    8 bits
  visual:
    visual id:    0x26
    class:    GrayScale
    depth:    8 planes
    available colormap entries:    256
    red, green, blue masks:    0x0, 0x0, 0x0
    significant bits in color specification:    8 bits
  visual:
    visual id:    0x27
    class:    StaticColor
    depth:    8 planes
    available colormap entries:    256
    red, green, blue masks:    0x7, 0x38, 0xc0
    significant bits in color specification:    8 bits
  visual:
    visual id:    0x28
    class:    TrueColor
    depth:    8 planes
    available colormap entries:    8 per subfield
    red, green, blue masks:    0x7, 0x38, 0xc0
    significant bits in color specification:    8 bits
  visual:
    visual id:    0x29
    class:    StaticGray
    depth:    8 planes
    available colormap entries:    256
    red, green, blue masks:    0x0, 0x0, 0x0
    significant bits in color specification:    8 bits
-- snip --

Comment 1

[Asa Dotzler [:asa]]

over to toolkit.

Comment 2

[Peter Trudelle]

->pinkerton, cc blizzard

Comment 3

[Mike Pinkerton (not reading bugmail)]

does it only crash when the other app is on a second screen, or does it crash
regardless, even if both are on the same screen? what about dragging from, say,
bookmarks to navigator (within the same app)?

Comment 4

[Roland Mainz]

AFAIK the same GTK+ application cannot span over two logical _screens_ (except
with Xinerama, but that would one _one_ logical screen).
It "crashes" (exit with X error) when I drag&drop items from one screen to
another.

Testing:
- In theory you may test this with Xnest (if someone fixes GTK+ core to avoid
killing Xnest by sending rubbish in the X layer) using the -scrns option
(binaries for Solaris SPARC on demand)...
- Has anyone here an UltraSPARC 5 or UltraSPARC10 ? Both can get up-to four
heads with cheap m64 cards (~$20)), details on demand...

Comment 5

[Roland Mainz]

Update:
Now I am totally confused...
a) this bug seems to happen on the 2nd screen (DISPLAY=:0.1) but not on the 1st
screen (DISPLAY=:0.0)
b) dragging something from :0.0 to :0.1 causes the described exit due X error;
dragging from :0.1 to :0.0 does not
c) sometimes the Zilla dies like this after :0.1-->:0.0 drag&drop:
-- snip --
Gdk-WARNING **: shmat failed!
nsWidget::~nsWidget() of toplevel: 11 widgets still exist.
WEBSHELL- = 4
X Error of failed request:  BadWindow (invalid Window parameter)
  Major opcode of failed request:  20 (X_GetProperty)
  Resource id in failed request:  0x3802b9b
  Serial number of failed request:  14325
  Current serial number in output stream:  14325
Bus Error - core dumped
-- snip --

-- snip --
(gdb) where
#0  0xfec461dc in t_splay () from /usr/lib/libc.so.1
#1  0xfec4604c in t_delete () from /usr/lib/libc.so.1
#2  0xfec45c64 in realfree () from /usr/lib/libc.so.1
#3  0xfec4648c in _free_unlocked () from /usr/lib/libc.so.1
#4  0xfec463dc in free () from /usr/lib/libc.so.1
#5  0xfd5b1464 in _XFreeDisplayStructure () from /usr/openwin/lib/libX11.so.4
#6  0xfd5aa614 in XCloseDisplay () from /usr/openwin/lib/libX11.so.4
#7  0xfd7262b8 in gdk_exit_func () from /usr/local/lib/libgdk-1.2.so.0
#8  0xfec20260 in _exithandle () from /usr/lib/libc.so.1
#9  0xfec983e8 in exit () from /usr/lib/libc.so.1
#10 0xfd5b1688 in _XDefaultError () from /usr/openwin/lib/libX11.so.4
#11 0xfdc50340 in XIMErrorHandler (dpy=0x9c8e0, event=0xffbed8b0) at
../../../../widget/src/gtk/nsGtkIMEHelper.cpp:1107
#12 0xfd5a75fc in _XError () from /usr/openwin/lib/libX11.so.4
#13 0xfd5999e4 in _XReply () from /usr/openwin/lib/libX11.so.4
#14 0xfd59e098 in XGetWindowProperty () from /usr/openwin/lib/libX11.so.4
#15 0xfd733468 in get_client_window_at_coords_recurse () from
/usr/local/lib/libgdk-1.2.so.0
#16 0xfd733890 in get_client_window_at_coords () from
/usr/local/lib/libgdk-1.2.so.0
#17 0xfd7393f0 in gdk_drag_find_window () from /usr/local/lib/libgdk-1.2.so.0
#18 0xfd96e070 in gtk_drag_update () from /usr/local/lib/libgtk-1.2.so.0
#19 0xfd96e528 in gtk_drag_motion_cb () from /usr/local/lib/libgtk-1.2.so.0
#20 0xfd9d98b0 in gtk_marshal_BOOL__POINTER () from
/usr/local/lib/libgtk-1.2.so.0
#21 0xfda50740 in gtk_handlers_run () from /usr/local/lib/libgtk-1.2.so.0
#22 0xfda4e8cc in gtk_signal_real_emit () from /usr/local/lib/libgtk-1.2.so.0
#23 0xfda49ac4 in gtk_signal_emit () from /usr/local/lib/libgtk-1.2.so.0
#24 0xfdacf7f8 in gtk_widget_event () from /usr/local/lib/libgtk-1.2.so.0
#25 0xfd9d9584 in gtk_propagate_event () from /usr/local/lib/libgtk-1.2.so.0
#26 0xfd9d70f4 in gtk_main_do_event () from /usr/local/lib/libgtk-1.2.so.0
#27 0xfdc3ad04 in handle_gdk_event (event=0x1da658, data=0x0) at
../../../../widget/src/gtk/nsGtkEventHandler.cpp:948
#28 0xfd741478 in gdk_event_dispatch () from /usr/local/lib/libgdk-1.2.so.0
#29 0xfd6b6dd0 in g_main_dispatch () from /usr/local/lib/libglib-1.2.so.0
#30 0xfd6b7bd4 in g_main_iterate () from /usr/local/lib/libglib-1.2.so.0
#31 0xfd6b7f6c in g_main_run () from /usr/local/lib/libglib-1.2.so.0
#32 0xfd9d60a8 in gtk_main () from /usr/local/lib/libgtk-1.2.so.0
#33 0xfdc2ea18 in nsAppShell::Run (this=0x97be8) at
../../../../widget/src/gtk/nsAppShell.cpp:350
#34 0xfe92d378 in nsAppShellService::Run (this=0x95ce8) at
../../../../xpfe/appshell/src/nsAppShellService.cpp:407
#35 0x1b450 in main1 (argc=1, argv=0xffbeea24, nativeApp=0x0) at
../../../xpfe/bootstrap/nsAppRunner.cpp:1016
#36 0x1bea8 in main (argc=1, argv=0xffbeea24) at
../../../xpfe/bootstrap/nsAppRunner.cpp:1260
-- snip --

Other info:
% gtk-config --version
1.2.8
% glib-config --version
1.2.8

Comment 6

[Mike Pinkerton (not reading bugmail)]

ok, this really sounds like some kind of gtk problem. do any other apps have
this problem, or is it just us?

--> blizzard, who will probably have a better idea of the gtk guts and better
access to hardware to test/debug this.

Comment 7

[Roland Mainz]

Please correct me anyone if I am wrong, but...
...GTK+ (1.2.8) does not support multihead (better: working with multiple
DISPLAYs at the same time), trying to do this results in all kinds of _bad_
problems...

Comment 8

[Christopher Blizzard (:blizzard)]

I don't know what happens when you try to DND across screens.  I also don't have
the hardware to test it out.

Comment 9

[Roland Mainz]

To blizzard: Which hardware do you have access to ?
One possibility is (which I use): get an UltraSPARC 5 and plug a ATI RagePro
(compatible cards are listed in
http://www.belwue.de/aktivitaeten/projekte/SUN-SH-2-SUN-DH.html (yes, I know...
german... sorry)).

I found an older URL covering the theme "does GTK support multiple screens":
http://mail.gnome.org/archives/gtk-list/1998-June/msg00662.html
Question is if this has been changed yet (I am afraid noone touched that issue
yet)... ;-((

Comment 10

[Christopher Blizzard (:blizzard)]

I don't have access to any dual head systems.

Comment 11

[Christopher Blizzard (:blizzard)]

This is a gtk bug.

Comment 12

[Roland Mainz]

Will this ever be fixed on GTK+ library side ?