[Vm-dev] Re: 32bits squeak.stack.spur failures

[Eliot Miranda]

> On Mar 18, 2016, at 5:46 PM, Nicolas Cellier <nicolas.cellier.aka.nice at gmail.com> wrote:
> 
> More exactly this happens from:
> ...
> processWeakSurvivor: 
> fireFinalization
> queueMourner:
> in
>      self deny: (self is: anEphemeronOrWeakArray onObjStack: mournQueue).

If mournQueue is not initialized, and that's expected, then the assert needs to read 

    self deny: (mournQueue ~= 0 and: [self is: anEphemeronOrWeakArray onObjStack: mournQueue]).

Check where the mournQueue is initialized.  I may have written the code so that there is no mournQueue (mournQueue = 0) until the new finalization mechanism is enabled.  If so, the assert needs updating.

If I write the code so that there's always a mournQueue (mournQueue is initialized in or from initializeObjectMemory:) then the asset is correct and mournQueue got zapped by mistake.


> 
> 2016-03-19 1:36 GMT+01:00 Nicolas Cellier <nicolas.cellier.aka.nice at gmail.com>:
>> Hi, this question is mainly for Eliot I guess...
>> 
>> I have regular failures with a 32 bits squeak.stack.spur
>> The VM crashes in garbage collect when running all tests.
>> 
>> When i simulate, I've got an assert: failure in
>> 
>> Spur32BitMMLESimulator>>is: oop onObjStack: objStack
>>     | index nextPage |
>>     self assert: (self numSlotsOfAny: objStack) = ObjStackPageSlots.
>> 
>> 0 = 4092 fails...
>> 
>> Does it mean the image got corrupted?
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20160318/77051e86/attachment-0001.htm