AutoCAD again ! After finally R14 runs fine, i'm trying with AutoCAD 2000. Setup goes ok (besides usual VBA problem). Starting with 'wine acad' gives a very early error of unhandled access violation. I tried to start it with --debugmsg +snoop to see what happened and, surprise, app starts up quite good. It then hangs much later, but that's another problem. So the question : how it's possible that requesting snoop messages bypasses the initial access violation ? I think that snoop adds only a thunk to calls on native dlls.... Any suggestion appreciated !
Regards
Max
"Max" == Max max@veneto.com writes:
Max> AutoCAD again ! After finally R14 runs fine, i'm trying with Max> AutoCAD 2000. Setup goes ok (besides usual VBA problem). Starting Max> with 'wine acad' gives a very early error of unhandled access Max> violation. I tried to start it with --debugmsg +snoop to see what Max> happened and, surprise, app starts up quite good. It then hangs Max> much later, but that's another problem. So the question : how it's Max> possible that requesting snoop messages bypasses the initial access Max> violation ? I think that snoop adds only a thunk to calls on native Max> dlls.... Any suggestion appreciated !
If I remember right, this behaviour is a sign of "stack smashing", that means, the stack got overwritten in some place and causes a crash in some other , mostly unrelated place. Sometimes you can get some hints for the cause of that error by running with --debugmsg +heap. That makes things much slower, but checks the heap on every heap allocation and aborts on the first encountered error. Then the smashing must be in between the last allocation and the error. Running with +relay helps you pinpoint.
Bye
On Sun, Jul 07, 2002 at 10:52:30PM +0200, Max wrote:
AutoCAD again ! After finally R14 runs fine, i'm trying with AutoCAD 2000. Setup goes ok (besides usual VBA problem). Starting with 'wine acad' gives a very early error of unhandled access violation. I tried to start it with --debugmsg +snoop to see what happened and, surprise, app starts up quite good. It then hangs much later, but that's another problem. So the question : how it's possible that requesting snoop messages bypasses the initial access violation ? I think that snoop adds only a thunk to calls on native dlls....
No, AFAIK +snoop stuff does some additional "housekeeping" work, like initializing/correcting stack and stuff. This crash could probably be caused by - API functions with wrong number of parameters (in .spec etc) - buffers uninitialized by Wine's function implementations even though they should be - some other random stack trashing or so
-
Andreas Mohr -
Max -
Uwe Bonnes