lock up in screen drag

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

lock up in screen drag

Eliot Miranda-2
Hi ALl,

   I'm seeing a lock up that appears to be on window dragging.  This is on Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see this process when sending SIGUSR1 to the process:

Process  0x4800ad0 priority 40
0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
0xbffac340 M Delay>wait 0x4802858: a(n) Delay
0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag: 0x4802740: a(n) PluggableSystemWindow
0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel: 0x4802740: a(n) PluggableSystemWindow
0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n) EventHandler
0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n) EventHandler
0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n) MouseClickState
0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n) MorphicProject
0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure

_,,,^..^,,,_
best, Eliot


Reply | Threaded
Open this post in threaded view
|

Re: lock up in screen drag

marcel.taeumel
Eliot Miranda-2 wrote
Hi ALl,

   I'm seeing a lock up that appears to be on window dragging.  This is on
Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see
this process when sending SIGUSR1 to the process:

Process  0x4800ad0 priority 40
0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
0xbffac340 M Delay>wait 0x4802858: a(n) Delay
0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag:
0x4802740: a(n) PluggableSystemWindow
0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel:
0x4802740: a(n) PluggableSystemWindow
0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n)
EventHandler
0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n) EventHandler
0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n)
MouseClickState
0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n)
MorphicProject
0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure

_,,,^..^,,,_
best, Eliot
Hi Eliot,

since you are calling Rectangle >>#newRectFrom:, I suppose you have "Fast Morphic Drag" enabled. There, you continuously call Form >> #border:width:rule:fillColor:.

Hence, I think some latest changes in the VM's graphics backend for Mac OSX are buggy.

Btw: A while ago, Bert did observe that fast drag is faster in SqueakJS than it is in the Mac OSX build of the CogVM. :-D What's the current state of affairs?

Best,
Marcel
Reply | Threaded
Open this post in threaded view
|

Re: lock up in screen drag

Eliot Miranda-2


> On Jul 18, 2016, at 9:43 PM, marcel.taeumel <[hidden email]> wrote:
>
> Eliot Miranda-2 wrote
>> Hi ALl,
>>
>>   I'm seeing a lock up that appears to be on window dragging.  This is on
>> Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see
>> this process when sending SIGUSR1 to the process:
>>
>> Process  0x4800ad0 priority 40
>> 0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
>> 0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
>> 0xbffac340 M Delay>wait 0x4802858: a(n) Delay
>> 0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
>> 0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag:
>> 0x4802740: a(n) PluggableSystemWindow
>> 0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel:
>> 0x4802740: a(n) PluggableSystemWindow
>> 0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n)
>> EventHandler
>> 0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n)
>> EventHandler
>> 0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
>> 0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
>> 0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n)
>> MouseClickState
>> 0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
>> 0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
>> 0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
>> 0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
>> 0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
>> 0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
>> 0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
>> 0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
>> 0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n)
>> MorphicProject
>> 0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure
>>
>> _,,,^..^,,,_
>> best, Eliot
>
> Hi Eliot,
>
> since you are calling Rectangle >>#newRectFrom:, I suppose you have "Fast
> Morphic Drag" enabled. There, you continuously call Form >>
> #border:width:rule:fillColor:.
>
> Hence, I think some latest changes in the VM's graphics backend for Mac OSX
> are buggy.

Then should we revert?  I lost a days work did to this.  Tobias?

> Btw: A while ago, Bert did observe that fast drag is faster in SqueakJS than
> it is in the Mac OSX build of the CogVM. :-D

And this is a criticism of what?

> What's the current state of
> affairs?

That the Mac VM is now unstable, and as my main work platform that is very bad news.

>
> Best,
> Marcel
>
>
>
> --
> View this message in context: http://forum.world.st/lock-up-in-screen-drag-tp4907062p4907067.html
> Sent from the Squeak - Dev mailing list archive at Nabble.com.
>

Reply | Threaded
Open this post in threaded view
|

Re: lock up in screen drag

Tobias Pape
Hi,

On 19.07.2016, at 18:02, Eliot Miranda <[hidden email]> wrote:

>
>
>> On Jul 18, 2016, at 9:43 PM, marcel.taeumel <[hidden email]> wrote:
>>
>> Eliot Miranda-2 wrote
>>> Hi ALl,
>>>
>>>  I'm seeing a lock up that appears to be on window dragging.  This is on
>>> Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see
>>> this process when sending SIGUSR1 to the process:
>>>
>>> Process  0x4800ad0 priority 40
>>> 0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
>>> 0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
>>> 0xbffac340 M Delay>wait 0x4802858: a(n) Delay
>>> 0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
>>> 0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag:
>>> 0x4802740: a(n) PluggableSystemWindow
>>> 0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel:
>>> 0x4802740: a(n) PluggableSystemWindow
>>> 0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n)
>>> EventHandler
>>> 0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n)
>>> EventHandler
>>> 0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
>>> 0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
>>> 0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n)
>>> MouseClickState
>>> 0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
>>> 0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
>>> 0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
>>> 0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
>>> 0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
>>> 0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
>>> 0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
>>> 0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
>>> 0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n)
>>> MorphicProject
>>> 0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure
>>>
>>> _,,,^..^,,,_
>>> best, Eliot
>>
>> Hi Eliot,
>>
>> since you are calling Rectangle >>#newRectFrom:, I suppose you have "Fast
>> Morphic Drag" enabled. There, you continuously call Form >>
>> #border:width:rule:fillColor:.
>>
>> Hence, I think some latest changes in the VM's graphics backend for Mac OSX
>> are buggy.
>
> Then should we revert?  I lost a days work did to this.  Tobias?

try it, go ahead and revert
        a5f7c3e451b9fb643fb9d1e95819169394218082 (scale factor)
or
        df90dec754e10e5c57246e1ad8f75550f559f8cb (high-dpi)
(for example with
        git revert THE_HASH
or
        git revert -n THE_HASH
to just see if it works)

I've seen the screen go blank, too (White, in my case) but resizing the
window restored things, why so ever. However, I had it happen _before_ any
of my changes, too. (But not with the CGView based backend, IIRC)

>
>> Btw: A while ago, Bert did observe that fast drag is faster in SqueakJS than
>> it is in the Mac OSX build of the CogVM. :-D
>
> And this is a criticism of what?

I didn't know that criticism  starts with "By the way"…

>
>> What's the current state of
>> affairs?
>
> That the Mac VM is now unstable, and as my main work platform that is very bad news.

it has been before, too, just to say…

Best regards
        -Tobi

>
>>
>> Best,
>> Marcel
>>
>>
>>
>> --
>> View this message in context: http://forum.world.st/lock-up-in-screen-drag-tp4907062p4907067.html
>> Sent from the Squeak - Dev mailing list archive at Nabble.com.



Reply | Threaded
Open this post in threaded view
|

Re: lock up in screen drag

Eliot Miranda-2


On Tue, Jul 19, 2016 at 10:05 AM, Tobias Pape <[hidden email]> wrote:
Hi,

On 19.07.2016, at 18:02, Eliot Miranda <[hidden email]> wrote:

>
>
>> On Jul 18, 2016, at 9:43 PM, marcel.taeumel <[hidden email]> wrote:
>>
>> Eliot Miranda-2 wrote
>>> Hi ALl,
>>>
>>>  I'm seeing a lock up that appears to be on window dragging.  This is on
>>> Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see
>>> this process when sending SIGUSR1 to the process:
>>>
>>> Process  0x4800ad0 priority 40
>>> 0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
>>> 0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
>>> 0xbffac340 M Delay>wait 0x4802858: a(n) Delay
>>> 0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
>>> 0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag:
>>> 0x4802740: a(n) PluggableSystemWindow
>>> 0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel:
>>> 0x4802740: a(n) PluggableSystemWindow
>>> 0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n)
>>> EventHandler
>>> 0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n)
>>> EventHandler
>>> 0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
>>> 0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
>>> 0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n)
>>> MouseClickState
>>> 0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
>>> 0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
>>> 0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
>>> 0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
>>> 0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
>>> 0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
>>> 0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
>>> 0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
>>> 0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n)
>>> MorphicProject
>>> 0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure
>>>
>>> _,,,^..^,,,_
>>> best, Eliot
>>
>> Hi Eliot,
>>
>> since you are calling Rectangle >>#newRectFrom:, I suppose you have "Fast
>> Morphic Drag" enabled. There, you continuously call Form >>
>> #border:width:rule:fillColor:.
>>
>> Hence, I think some latest changes in the VM's graphics backend for Mac OSX
>> are buggy.
>
> Then should we revert?  I lost a days work did to this.  Tobias?

try it, go ahead and revert
        a5f7c3e451b9fb643fb9d1e95819169394218082 (scale factor)
or
        df90dec754e10e5c57246e1ad8f75550f559f8cb (high-dpi)
(for example with
        git revert THE_HASH
or
        git revert -n THE_HASH
to just see if it works)

I've seen the screen go blank, too (White, in my case) but resizing the
window restored things, why so ever. However, I had it happen _before_ any
of my changes, too. (But not with the CGView based backend, IIRC)

Ah brilliant!  Now I can wait until you fix it.  Thanks, Tobias, the resize is great.  But I'd really appreciate a fix reasonably prompty!!
 

>
>> Btw: A while ago, Bert did observe that fast drag is faster in SqueakJS than
>> it is in the Mac OSX build of the CogVM. :-D
>
> And this is a criticism of what?

I didn't know that criticism  starts with "By the way"…

>
>> What's the current state of
>> affairs?
>
> That the Mac VM is now unstable, and as my main work platform that is very bad news.

it has been before, too, just to say…

Best regards
        -Tobi

>
>>
>> Best,
>> Marcel
>>
>>
>>
>> --
>> View this message in context: http://forum.world.st/lock-up-in-screen-drag-tp4907062p4907067.html
>> Sent from the Squeak - Dev mailing list archive at Nabble.com.






--
_,,,^..^,,,_
best, Eliot


Reply | Threaded
Open this post in threaded view
|

Re: lock up in screen drag

Tobias Pape

On 19.07.2016, at 20:47, Eliot Miranda <[hidden email]> wrote:

> On Tue, Jul 19, 2016 at 10:05 AM, Tobias Pape <[hidden email]> wrote:
> Hi,
>
> On 19.07.2016, at 18:02, Eliot Miranda <[hidden email]> wrote:
>
> >
> >
> >> On Jul 18, 2016, at 9:43 PM, marcel.taeumel <[hidden email]> wrote:
> >>
> >> Eliot Miranda-2 wrote
> >>> Hi ALl,
> >>>
> >>>  I'm seeing a lock up that appears to be on window dragging.  This is on
> >>> Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see
> >>> this process when sending SIGUSR1 to the process:
> >>>
> >>> Process  0x4800ad0 priority 40
> >>> 0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
> >>> 0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
> >>> 0xbffac340 M Delay>wait 0x4802858: a(n) Delay
> >>> 0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
> >>> 0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag:
> >>> 0x4802740: a(n) PluggableSystemWindow
> >>> 0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel:
> >>> 0x4802740: a(n) PluggableSystemWindow
> >>> 0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n)
> >>> EventHandler
> >>> 0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n)
> >>> EventHandler
> >>> 0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
> >>> 0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
> >>> 0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n)
> >>> MouseClickState
> >>> 0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
> >>> 0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
> >>> 0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
> >>> 0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
> >>> 0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n)
> >>> MorphicProject
> >>> 0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure
> >>>
> >>> _,,,^..^,,,_
> >>> best, Eliot
> >>
> >> Hi Eliot,
> >>
> >> since you are calling Rectangle >>#newRectFrom:, I suppose you have "Fast
> >> Morphic Drag" enabled. There, you continuously call Form >>
> >> #border:width:rule:fillColor:.
> >>
> >> Hence, I think some latest changes in the VM's graphics backend for Mac OSX
> >> are buggy.
> >
> > Then should we revert?  I lost a days work did to this.  Tobias?
>
> try it, go ahead and revert
>         a5f7c3e451b9fb643fb9d1e95819169394218082 (scale factor)
> or
>         df90dec754e10e5c57246e1ad8f75550f559f8cb (high-dpi)
> (for example with
>         git revert THE_HASH
> or
>         git revert -n THE_HASH
> to just see if it works)
>
> I've seen the screen go blank, too (White, in my case) but resizing the
> window restored things, why so ever. However, I had it happen _before_ any
> of my changes, too. (But not with the CGView based backend, IIRC)
>
> Ah brilliant!  Now I can wait until you fix it.  Thanks, Tobias, the resize is great.  But I'd really appreciate a fix reasonably prompty!!

Sorry, ain't got no fix for that.
And since I have to always set up the Xcode-stuff manually, I cannot debug this reasonably.
To be frank, I'll actually only have time for VM stuff in September earliest.
Since this is obviously a tad late, feel free to revert my stuff; I cannot promise that this will fix
things, maybe things only break less often.

Best regards
        -Tobi

PS: The cocoa vm keycodes are off. Cmd-0 does not send cmd-0. Sligthly intoxicated tobi out

>  
>
> >
> >> Btw: A while ago, Bert did observe that fast drag is faster in SqueakJS than
> >> it is in the Mac OSX build of the CogVM. :-D
> >
> > And this is a criticism of what?
>
> I didn't know that criticism  starts with "By the way"…
>
> >
> >> What's the current state of
> >> affairs?
> >
> > That the Mac VM is now unstable, and as my main work platform that is very bad news.
>
> it has been before, too, just to say…
>
> Best regards
>         -Tobi
>
> >
> >>
> >> Best,
> >> Marcel
> >>
> >>
> >>




Reply | Threaded
Open this post in threaded view
|

Re: lock up in screen drag

marcel.taeumel
Tobias Pape wrote
On 19.07.2016, at 20:47, Eliot Miranda <[hidden email]> wrote:

> On Tue, Jul 19, 2016 at 10:05 AM, Tobias Pape <[hidden email]> wrote:
> Hi,
>
> On 19.07.2016, at 18:02, Eliot Miranda <[hidden email]> wrote:
>
> >
> >
> >> On Jul 18, 2016, at 9:43 PM, marcel.taeumel <[hidden email]> wrote:
> >>
> >> Eliot Miranda-2 wrote
> >>> Hi ALl,
> >>>
> >>>  I'm seeing a lock up that appears to be on window dragging.  This is on
> >>> Mac OS X with a new VM.  The symptom is that the screen goes blank.  I see
> >>> this process when sending SIGUSR1 to the process:
> >>>
> >>> Process  0x4800ad0 priority 40
> >>> 0xbffac304 M [] in Delay>wait 0x4802858: a(n) Delay
> >>> 0xbffac324 M BlockClosure>ifCurtailed: 0x4a920e0: a(n) BlockClosure
> >>> 0xbffac340 M Delay>wait 0x4802858: a(n) Delay
> >>> 0xbffac36c M Rectangle>newRectFrom: 0x4803df0: a(n) Rectangle
> >>> 0xbffac39c I PluggableSystemWindow(SystemWindow)>doFastFrameDrag:
> >>> 0x4802740: a(n) PluggableSystemWindow
> >>> 0xbffac3b8 M PluggableSystemWindow(SystemWindow)>startDragFromLabel:
> >>> 0x4802740: a(n) PluggableSystemWindow
> >>> 0xbffac3d8 M EventHandler>send:to:withEvent:fromMorph: 0x4803f30: a(n)
> >>> EventHandler
> >>> 0xbffac408 I EventHandler>startDrag:fromMorph: 0x4803f30: a(n)
> >>> EventHandler
> >>> 0xbffac428 M StringMorph(Morph)>startDrag: 0x4801458: a(n) StringMorph
> >>> 0xbffac44c I MouseClickState>drag: 0x4803fc8: a(n) MouseClickState
> >>> 0xbffac474 M MouseClickState>handleEvent:from: 0x4803fc8: a(n)
> >>> MouseClickState
> >>> 0xbffac498 M HandMorph>handleEvent: 0x5cef1a0: a(n) HandMorph
> >>> 0xbffac4c4 M HandMorph>processEvents 0x5cef1a0: a(n) HandMorph
> >>> 0xbffac4e0 M [] in WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac504 M Array(SequenceableCollection)>do: 0x4e6fa10: a(n) Array
> >>> 0xbffac520 M WorldState>handsDo: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac540 M WorldState>doOneCycleNowFor: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac55c M WorldState>doOneCycleFor: 0x5d1ba70: a(n) WorldState
> >>> 0xbffac578 M PasteUpMorph>doOneCycle 0x52d8780: a(n) PasteUpMorph
> >>> 0xbffac590 M [] in MorphicProject>spawnNewProcess 0x604fe08: a(n)
> >>> MorphicProject
> >>> 0xbffac5b0 I [] in BlockClosure>newProcess 0x48040b8: a(n) BlockClosure
> >>>
> >>> _,,,^..^,,,_
> >>> best, Eliot
> >>
> >> Hi Eliot,
> >>
> >> since you are calling Rectangle >>#newRectFrom:, I suppose you have "Fast
> >> Morphic Drag" enabled. There, you continuously call Form >>
> >> #border:width:rule:fillColor:.
> >>
> >> Hence, I think some latest changes in the VM's graphics backend for Mac OSX
> >> are buggy.
> >
> > Then should we revert?  I lost a days work did to this.  Tobias?
>
> try it, go ahead and revert
>         a5f7c3e451b9fb643fb9d1e95819169394218082 (scale factor)
> or
>         df90dec754e10e5c57246e1ad8f75550f559f8cb (high-dpi)
> (for example with
>         git revert THE_HASH
> or
>         git revert -n THE_HASH
> to just see if it works)
>
> I've seen the screen go blank, too (White, in my case) but resizing the
> window restored things, why so ever. However, I had it happen _before_ any
> of my changes, too. (But not with the CGView based backend, IIRC)
>
> Ah brilliant!  Now I can wait until you fix it.  Thanks, Tobias, the resize is great.  But I'd really appreciate a fix reasonably prompty!!

Sorry, ain't got no fix for that.
And since I have to always set up the Xcode-stuff manually, I cannot debug this reasonably.
To be frank, I'll actually only have time for VM stuff in September earliest.
Since this is obviously a tad late, feel free to revert my stuff; I cannot promise that this will fix
things, maybe things only break less often.

Best regards
        -Tobi

PS: The cocoa vm keycodes are off. Cmd-0 does not send cmd-0. Sligthly intoxicated tobi out

>  
>
> >
> >> Btw: A while ago, Bert did observe that fast drag is faster in SqueakJS than
> >> it is in the Mac OSX build of the CogVM. :-D
> >
> > And this is a criticism of what?
>
> I didn't know that criticism  starts with "By the way"…
>
> >
> >> What's the current state of
> >> affairs?
> >
> > That the Mac VM is now unstable, and as my main work platform that is very bad news.
>
> it has been before, too, just to say…
>
> Best regards
>         -Tobi
>
> >
> >>
> >> Best,
> >> Marcel
> >>
> >>
> >>
Hey Tobias,

we should fix the keyboard-event bugs on the Mac until the 5.1 release. Maybe we can pair-prog some afternoon to debug it. :-)

Best,
Marcel