[Vm-dev] Trying to make systemTracer write a CogImage

Wed Apr 25 18:04:59 UTC 2012

HAhaha, experte?  Nope, I'm learning :).  And I have to thank Mariano
because he was sitting next to me and telling me lots of gorgeous tips to
debug the vm :P.

On Wed, Apr 25, 2012 at 7:55 PM, Mariano Martinez Peck <
marianopeck at gmail.com> wrote:

>
>
>
> On Wed, Apr 25, 2012 at 7:27 PM, Eliot Miranda <eliot.miranda at gmail.com>wrote:
>
>>
>>
>>
>> On Wed, Apr 25, 2012 at 4:10 AM, Guillermo Polito <
>> guillermopolito at gmail.com> wrote:
>>
>>>
>>>
>>>
>>> On Wed, Apr 25, 2012 at 1:07 PM, Guillermo Polito <
>>> guillermopolito at gmail.com> wrote:
>>>
>>>>
>>>>
>>>> On Wed, Apr 25, 2012 at 12:04 PM, Guillermo Polito <
>>>> guillermopolito at gmail.com> wrote:
>>>>
>>>>> Ok, I've managed to get the cause of the process failing :).
>>>>>
>>>>> I'm having a primitive failure in
>>>>>
>>>>> VirtualMachine class>setGCBiasToGrowGCLimit: 2002796048: a(n)
>>>>> VirtualMachine class
>>>>>
>>>>> when trying to startup the vm, and therefore the ui process is never
>>>>> set up :/.  But I do not know why it should be failing :(.  I'm still
>>>>> debugging.
>>>>>
>>>>
>>>> And this is because I have to put in 0 the index of the cache in named
>>>> primitives literal :).
>>>>
>>>> Still not working, but better than yesterday :).
>>>>
>>>>
>>> And now it works! :D  I was setting wrongly the endiannes flag in the vm
>>> header, and the floats were not well read.
>>>
>>> Next step, polish and publish new SystemTracer :).
>>>
>>
>> Brilliant!  Well done for pushing through.  Now you're the expert!!
>>
>>
>
>
> The good thing is that he is learning a lot about the VM but at the same
> time delivering something useful for the community.
>
> Cheers
>
>
>>
>>> Guille
>>>
>>>
>>>>
>>>>
>>>>>
>>>>> Guille
>>>>>
>>>>> On Tue, Apr 24, 2012 at 11:08 PM, Eliot Miranda <
>>>>> eliot.miranda at gmail.com> wrote:
>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Tue, Apr 24, 2012 at 9:46 AM, Guillermo Polito <
>>>>>> guillermopolito at gmail.com> wrote:
>>>>>>
>>>>>>>
>>>>>>> Ok, I'm writing the image header as with the extra words.  But my
>>>>>>> image is kind of not living.  It seems the Display is not waking up.
>>>>>>>
>>>>>>> When I open my new image with a cog vm, I get an image that does not
>>>>>>> break, but comes up with a window in black. I sent a SIGUSR1 to the vm
>>>>>>> Process and get the following.
>>>>>>>
>>>>>>> All Smalltalk process stacks (active first):
>>>>>>> Process 0x776c9498 priority 10
>>>>>>> 0xbfc2fbf0 M ProcessorScheduler class>idleProcess 2003034484: a(n)
>>>>>>> ProcessorScheduler class
>>>>>>> 2006384288 s [] in ProcessorScheduler class>startUp
>>>>>>> 2008868524 s [] in BlockClosure>newProcess
>>>>>>>
>>>>>>> Process 0x77636ce8 priority 60
>>>>>>> 2003063132 s SmalltalkImage>lowSpaceWatcher
>>>>>>> 2003240984 s [] in SmalltalkImage>installLowSpaceWatcher
>>>>>>> 2003695368 s [] in BlockClosure>newProcess
>>>>>>>
>>>>>>> Process 0x77638668 priority 50
>>>>>>> 2003074716 s WeakArray class>finalizationProcess
>>>>>>> 2003263900 s [] in WeakArray class>restartFinalizationProcess
>>>>>>> 2003718900 s [] in BlockClosure>newProcess
>>>>>>>
>>>>>>> Process 0x779860e4 priority 80
>>>>>>> 2009271856 s Delay class>handleTimerEvent
>>>>>>> 2010697732 s Delay class>runTimerEventLoop
>>>>>>> 2011549844 s [] in Delay class>startTimerEventLoop
>>>>>>> 2012016284 s [] in BlockClosure>newProcess
>>>>>>>
>>>>>>> Process 0x78809050 priority 40
>>>>>>> 0xbfc2eac4 M [] in Semaphore>critical: 2006475044: a(n) Semaphore
>>>>>>> 0xbfc2eae4 M BlockClosure>ensure: 2021697252: a(n) BlockClosure
>>>>>>> 0xbfc2eb04 M Semaphore>critical: 2006475044: a(n) Semaphore
>>>>>>> 0xbfc2eb28 I Delay>schedule 2021697000: a(n) Delay
>>>>>>> 0xbfc2eb48 I Delay>wait 2021697000: a(n) Delay
>>>>>>> 0xbfc2eb70 I WorldState>interCyclePause: 2004530692: a(n) WorldState
>>>>>>> 0xbfc2eb94 I WorldState>doOneCycleFor: 2004530692: a(n) WorldState
>>>>>>> 0xbfc2ebb8 I PasteUpMorph>doOneCycle 2003571264: a(n) PasteUpMorph
>>>>>>> 0xbfc2ebd0 M [] in MorphicUIManager>spawnNewProcess 2006644196: a(n)
>>>>>>> MorphicUIManager
>>>>>>> 0xbfc2ebf0 I [] in BlockClosure>newProcess 2021691252: a(n)
>>>>>>> BlockClosure
>>>>>>>
>>>>>>> Most recent primitives
>>>>>>> relinquishProcessorForMicroseconds:
>>>>>>> relinquishProcessorForMicroseconds:
>>>>>>> ...
>>>>>>> (more relinquishProcessorForMicroseconds:)
>>>>>>>
>>>>>>>
>>>>>>> Now, looking at the primitives called, It seems the idle process is
>>>>>>> living but not giving any place to the other processes...  any ideas?  Does
>>>>>>> cog do something different with processes at system startup?
>>>>>>>
>>>>>>
>>>>>> Of course not :)  It's a Squeak VM with a JIT.
>>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>> Thanks!
>>>>>>> Guille
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> best,
>>>>>> Eliot
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>>
>>
>>
>> --
>> best,
>> Eliot
>>
>>
>>
>
>
> --
> Mariano
> http://marianopeck.wordpress.com
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20120425/e91dd3bc/attachment-0001.htm