[4549] in Athena Bugs
mwm dumps core
daemon@ATHENA.MIT.EDU (John Carr)
Sat Mar 17 08:28:11 1990
To: bugs@ATHENA.MIT.EDU
Date: Sat, 17 Mar 90 08:28:00 EST
From: John Carr <jfc@ATHENA.MIT.EDU>
Note: the following only happens when I run mwm on an X server I am
developing. It happens whether the client machine is a vax or RT. I
suspect mwm is assuming certain things "can't happen" (for example, this X
server only implements a very small set of key codes at this time). I
fell this report is justified because an X client should not dump core
regardless of what the server sends it.
What I did:
mwm (in foreground)
^C (say "cancel" when mwm asks for confirmation)
^C
After the second ^C mwm dumped core.
This is the stack trace:
_StackWindow(b3c50,1,0) from _ReManageDialog+3e
_ReManageDialog(b1200) from 56ef
_ConfirmAction(2) from _F_Quit_Mwm+b
_F_Quit_Mwm(0,0,0) from _QuitWmSignalHandler+f
_QuitWmSignalHandler(2,0,7fffe374,12e54) from 7fffe87e
sigtramp(2,0,7fffe374,12e54) from _select+6
_select(4,7fffe444,7fffe424,7fffe404,0) from 4d005
__XtwaitForSomething(0,0,0,1,0,86400) from _XtAppNextEvent+89
_XtAppNextEvent(86400,7fffe4f8) from _XtNextEvent+12
_XtNextEvent(7fffe4f8) from _main+36
_main(1,7fffe590,7fffe598) from start+3d
From disassembling the code, I can tell that the fault occurs when testing
a variable in code that looks like:
if (some_memory_reference != 0)
FindTransientOnTop();
else
...
in the function StackWindow(). The problem is that the memory reference
(which looks like a pointer field in a structure) points to a bad address.
Core dumps or demos provided on request.
