(Normal) VM crash while debugging exception handling

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

(Normal) VM crash while debugging exception handling

Frank Shearar
 
I applied Andreas' Exception fix (Kernel-ar.540 in the Inbox), to
double-check Julian's counter-example. I used a virgin trunk updated to
10877.

1 I did a print-it on "[ Error signal ] on: UnhandledError do: [ :ex |
'ANSI says so' ]", and the debugger popped up just as Julian described.

2 I accidentally dropped too far down the stack into
Compiler>>evaluate:in:to:notifying:ifFail:logged:.

3 I hit "restart" and while stepping through execution I got as far as

     value := receiver
         withArgs: (context ifNil: [#()] ifNotNil: [{context}])
         executeMethod: method.

and when I hit "Over" the VM (Squeak4.0.01 from Mar 28 2010) crashed, saying

Exception code: C0000005
Exception address: 004096D1
Current byte code: 209
Primitive index: 258

Crash dump attached as crash-1.dmp.

Trying to reproduce the problem resulted in another crash:

Reason: Recursive not understood error encountered
Current byte code: 231
Primitive index: 199

Crash dump attached as crash-2.dmp

Subsequent attempts repeated the second error, as in crash-4.dmp.

Crash-3.dmp and crash-5.dmp happened while trying to reproduce the exact
crash.

So we have two errors:
* an access violation in crash-1, crash-3 and crash-5
* a recursive MNU in crash-2, crash-4

All appear somewhere during the execution of MethodContext>printDetails:.

frank

crash-1.dmp (2K) Download Attachment
crash-2.dmp (2K) Download Attachment
crash-5.dmp (2K) Download Attachment
crash-3.dmp (2K) Download Attachment
crash-4.dmp (3K) Download Attachment