On 01/29/2017 07:48 AM, Ben Coman wrote:




On Sun, Jan 29, 2017 at 7:02 PM, Levente Uzonyi <leves@caesar.elte.hu> wrote:

Hi Dan,

The expected output is in the file you attached. There seem to be no active processes to run, so only the idle process is being executed:

All Smalltalk process stacks (active first):
Process          0x1c9a9c8 priority 10
    0x7ffc012e81c0 M ProcessorScheduler class>idleProcess 0x15d2a30: a(n) ProcessorScheduler class
    0x7ffc012e8200 I [] in ProcessorScheduler class>startUp 0x15d2a30: a(n) ProcessorScheduler class
    0x7ffc012e8240 I [] in BlockClosure>newProcess 0x1d1aa40: a(n) BlockClosure

Most recent primitives
relinquishProcessorForMicroseconds:
relinquishProcessorForMicroseconds:
relinquishProcessorForMicroseconds:
relinquishProcessorForMicroseconds:
relinquishProcessorForMicroseconds:
relinquishProcessorForMicroseconds:
...

Among the suspended processes is the UI process, which is waiting on a Delay. So, there may be issues on the image side with Delays.

Process          0x1c9a790 priority 40
    0x7ffc012e60d8 M [] in Delay>wait 0x1c9ab48: a(n) Delay
    0x7ffc012e6118 M BlockClosure>ifCurtailed: 0x11b1228: a(n) BlockClosure
    0x7ffc012e6150 M Delay>wait 0x1c9ab48: a(n) Delay
    0x7ffc012e6190 M WorldState>doOneCycleFor: 0x15cd6e0: a(n) WorldState

So as part of exploring the problem, an awful hack would be deleting that #wait from #dowOneCycleFor:.  

Hi Ben,

I started the image by doing this, in effect:

/home/dan/cuis/cogspur64/squeak /home/dan/cuis/Cuis-Smalltalk-Dev/a64.image > freeze.log 2>&1

After loading packages containing my code I deleted:

(Delay for seconds: ... ) wait.

and ran for a while, did the 'pkill -USR1 -n -x squeak';

ran some more until it froze, did another 'pkill'.

The stdout looked like the other runs. There was no sign of stderr output.

 - Dan


cheers -ben
 
    0x7ffc012e61c8 M PasteUpMorph>doOneCycle 0x165f890: a(n) PasteUpMorph
    0x7ffc012e6200 M [] in ProjectX class>spawnNewMorphicProcessFor: 0x15d7900: a(n) ProjectX class
    0x7ffc012e6240 I [] in BlockClosure>newProcess 0x1d1a940: a(n) BlockClosure

I don't see any sign of stack overflow you mentioned.

Levente