VM Crash?

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

VM Crash?

Jan Kurš
Hi,

I am running the latest moose, with VM from May 4th on Linux and I consistently experience (over a period of several weeks) occasional crashes.

My rough observation is that this happens when working with visualization.

Do I do something wrong, or are there some issues?

The crash dump is attached. 

Cheers,
Jan

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev

crash.dmp (25K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

Peter Uhnak
I consistently experience (over a period of several weeks) occasional crashes.

If it makes you feel any better, I experience it several times a day…
on Monday it got to a point that I took several days (three and counting) break from Pharo because I was so utterly frustrated by this…


My rough observation is that this happens when working with visualization.

Do I do something wrong, or are there some issues?

There are (at least) three different crashes that happen often… two related to Athens/Roassal it seems (one you reported is one of them) and one related to fonts.
Of course I reported all of them, but with no known way to reproducibly crash it, I seriously doubt it will be fixed anytime soon.

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

abergel
In reply to this post by Jan Kurš
> I am running the latest moose, with VM from May 4th on Linux and I consistently experience (over a period of several weeks) occasional crashes.
>
> My rough observation is that this happens when working with visualization.
>
> Do I do something wrong, or are there some issues?
>
> The crash dump is attached.

You are not the only one. In your trace:

Smalltalk stack dump:
0xff9f8564 M GrafPort>copyBits 0x95f2390: a(n) GrafPort
0xff9f857c M GrafPort>image:at:sourceRect:rule: 0x95f2390: a(n) GrafPort
0xff9f85a4 M FormCanvas>image:at:sourceRect:rule: 0x95f2428: a(n) FormCanvas
0xff9f85d0 M TRMorph>drawOn: 0x95edd40: a(n) TRMorph

I have no idea what is wrong with the form canvas and this GraphPort.
I have added several guards in TRMorph>>drawOn: they help, but they reduce the user experience.

It would be help to have that issue fixed. Drawing primitives graphical elements should not make the crash the VM crash

Alexandre

>
> Cheers,
> Jan
> <crash.dmp>_______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.



_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

jfabry
In reply to this post by Peter Uhnak
I have the same problem and I am also frustrated. But to fix it is very difficult.

Fundamentally, the problem with this issue is to reliably reproduce it. I talked to Esteban about it, and he needs an image that will reliably crash to be able to debug it. He has no such image. I have one that almost crashes reliably, and I will give it to him tomorrow.

If you have an image that always crashes quickly, this is GOLD! Please please please end it to Esteban so he can debug!

--
Does this mail seem too brief? Sorry for that, I don’t mean to be rude! Please see http://emailcharter.org .

Johan Fabry   -   http://pleiad.cl/~jfabry
PLEIAD and RyCh labs  -  Computer Science Department (DCC)  -  University of Chile

On Jul 7, 2016, at 05:27, Peter Uhnák <[hidden email]> wrote:

I consistently experience (over a period of several weeks) occasional crashes.

If it makes you feel any better, I experience it several times a day…
on Monday it got to a point that I took several days (three and counting) break from Pharo because I was so utterly frustrated by this…


My rough observation is that this happens when working with visualization.

Do I do something wrong, or are there some issues?

There are (at least) three different crashes that happen often… two related to Athens/Roassal it seems (one you reported is one of them) and one related to fonts.
Of course I reported all of them, but with no known way to reproducibly crash it, I seriously doubt it will be fixed anytime soon.
_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev


_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

Stephan Eggermont-3
First thing is to upgrade the vm to stable. There have been interesting bug fixes

Stephan

Verstuurd vanaf mijn iPhone

Op 7 jul. 2016 om 13:27 heeft Johan Fabry <[hidden email]> het volgende geschreven:

I have the same problem and I am also frustrated. But to fix it is very difficult.

Fundamentally, the problem with this issue is to reliably reproduce it. I talked to Esteban about it, and he needs an image that will reliably crash to be able to debug it. He has no such image. I have one that almost crashes reliably, and I will give it to him tomorrow.

If you have an image that always crashes quickly, this is GOLD! Please please please end it to Esteban so he can debug!

--
Does this mail seem too brief? Sorry for that, I don’t mean to be rude! Please see http://emailcharter.org .

Johan Fabry   -   http://pleiad.cl/~jfabry
PLEIAD and RyCh labs  -  Computer Science Department (DCC)  -  University of Chile

On Jul 7, 2016, at 05:27, Peter Uhnák <[hidden email]> wrote:

I consistently experience (over a period of several weeks) occasional crashes.

If it makes you feel any better, I experience it several times a day…
on Monday it got to a point that I took several days (three and counting) break from Pharo because I was so utterly frustrated by this…


My rough observation is that this happens when working with visualization.

Do I do something wrong, or are there some issues?

There are (at least) three different crashes that happen often… two related to Athens/Roassal it seems (one you reported is one of them) and one related to fonts.
Of course I reported all of them, but with no known way to reproducibly crash it, I seriously doubt it will be fixed anytime soon.
_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

abergel
In reply to this post by Jan Kurš
One way to investigate where the problem comes from, is to understand what keeps some references of AthensCairoSurface
These objects contains some external resources, and my images are full of them. No idea why they are kept along. I will investigate this soon, but if you want to give a try, please do.

Alexandre


> On Jul 7, 2016, at 9:46 AM, Jan Kurš <[hidden email]> wrote:
>
> Hi,
>
> I am running the latest moose, with VM from May 4th on Linux and I consistently experience (over a period of several weeks) occasional crashes.
>
> My rough observation is that this happens when working with visualization.
>
> Do I do something wrong, or are there some issues?
>
> The crash dump is attached.
>
> Cheers,
> Jan
> <crash.dmp>_______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.



_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

jfabry

Put an image on dropbox so Esteban can look at it as well!!

--
Does this mail seem too brief? Sorry for that, I don’t mean to be rude! Please see http://emailcharter.org .

Johan Fabry   -   http://pleiad.cl/~jfabry
PLEIAD and RyCh labs  -  Computer Science Department (DCC)  -  University of Chile

> On Jul 7, 2016, at 12:43, Alexandre Bergel <[hidden email]> wrote:
>
> One way to investigate where the problem comes from, is to understand what keeps some references of AthensCairoSurface
> These objects contains some external resources, and my images are full of them. No idea why they are kept along. I will investigate this soon, but if you want to give a try, please do.
>
> Alexandre
>
>
>> On Jul 7, 2016, at 9:46 AM, Jan Kurš <[hidden email]> wrote:
>>
>> Hi,
>>
>> I am running the latest moose, with VM from May 4th on Linux and I consistently experience (over a period of several weeks) occasional crashes.
>>
>> My rough observation is that this happens when working with visualization.
>>
>> Do I do something wrong, or are there some issues?
>>
>> The crash dump is attached.
>>
>> Cheers,
>> Jan
>> <crash.dmp>_______________________________________________
>> Moose-dev mailing list
>> [hidden email]
>> https://www.list.inf.unibe.ch/listinfo/moose-dev
>
> --
> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
> Alexandre Bergel  http://www.bergel.eu
> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>
>
>
> _______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

Henrik Nergaard
In reply to this post by abergel
That dump looks familiar..

----
| surface morph formCanvas |

surface := AthensCairoSurface extent: 100@100.

surface drawDuring: [ :athCan |
 athCan
  setShape: (0 asPoint corner: 100 asPoint);
  setPaint: Color orange;
  draw.
].

morph := surface asForm asMorph.

"comment out these two lines, and it draws fine"
surface := nil.
Smalltalk garbageCollect.

formCanvas := FormCanvas extent: 100 asPoint.

morph fullDrawOn: formCanvas .
morph debugDrawError
---

AthensCairoCanvas >>#asForm do not actually return a "valid" form, but an "external" that holds an index to a Surface, instead of the actual bits, and somewhere during drawing there is a machinery which finds the surface based on this index, and copy those bits. That way #asForm do not need to allocate any bits itself, but at the expense that the lifetime of the surface must be longer then the form.

I guess you can try to use AthensCairoSurface>>#displayOnMorphicCanvas:at: in TRMorph>>#drawOn: instead of how it is done now, this should assure that the Surface is still alive when drawing.

I also think that changing the extent of the surface should be moved from #layoutChanged to the start of #drawOn:. The layoutChanged/ morph extent change could happen more than once in between draw calls,  which could cause unnecessary recreations of the surface. It should also ensure that the surface cannot be recreated while inside #drawOn:.

Hope this helps.

Best regards,
Henrik





 

Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

abergel
Hi Henrik,

thank thank you to look into that.

Indeed, that error looks familiar

> I guess you can try to use AthensCairoSurface>>#displayOnMorphicCanvas:at:
> in TRMorph>>#drawOn: instead of how it is done now, this should assure that
> the Surface is still alive when drawing.


In TRMorph>>drawOn:, replacing

        aCanvas image: f at: self bounds origin sourceRect: (0 @ 0 extent: surface extent) rule: 34

by
        surface displayOnMorphicCanvas: aCanvas at: 0 @ 0

does not draw anything. Did I do something wrong?


> I also think that changing the extent of the surface should be moved from
> #layoutChanged to the start of #drawOn:. The layoutChanged/ morph extent
> change could happen more than once in between draw calls,  which could cause
> unnecessary recreations of the surface. It should also ensure that the
> surface cannot be recreated while inside #drawOn:.

Thanks! I have committed this
Running all the Roassal2- tests leaves 5 instances of AthensCairoSurface that cannot be garbage collected.
With the change you are suggesting, it is reduced to 4. Still something!

Cheers,
Alexandre

>
> Hope this helps.
>
> Best regards,
> Henrik
>
>
>
>
>
>
>
>
>
>
>
> --
> View this message in context: http://forum.world.st/VM-Crash-tp4905304p4905497.html
> Sent from the Moose mailing list archive at Nabble.com.
> _______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.



_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

Henrik Nergaard
You need to draw it at the correct location in the world ;).

surface displayOnMorphicCanvas: aCanvas at: self position

Best regards,
Henrik

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf Of Alexandre Bergel
Sent: Friday, July 8, 2016 11:20 AM
To: Moose-related development <[hidden email]>
Subject: [Moose-dev] Re: VM Crash?

Hi Henrik,

thank thank you to look into that.

Indeed, that error looks familiar

> I guess you can try to use AthensCairoSurface>>#displayOnMorphicCanvas:at:
> in TRMorph>>#drawOn: instead of how it is done now, this should assure
> that the Surface is still alive when drawing.


In TRMorph>>drawOn:, replacing

        aCanvas image: f at: self bounds origin sourceRect: (0 @ 0 extent: surface extent) rule: 34

by
        surface displayOnMorphicCanvas: aCanvas at: 0 @ 0

does not draw anything. Did I do something wrong?


> I also think that changing the extent of the surface should be moved
> from #layoutChanged to the start of #drawOn:. The layoutChanged/ morph
> extent change could happen more than once in between draw calls,  
> which could cause unnecessary recreations of the surface. It should
> also ensure that the surface cannot be recreated while inside #drawOn:.

Thanks! I have committed this
Running all the Roassal2- tests leaves 5 instances of AthensCairoSurface that cannot be garbage collected.
With the change you are suggesting, it is reduced to 4. Still something!

Cheers,
Alexandre

>
> Hope this helps.
>
> Best regards,
> Henrik
>
>
>
>
>
>
>
>
>
>
>
> --
> View this message in context:
> http://forum.world.st/VM-Crash-tp4905304p4905497.html
> Sent from the Moose mailing list archive at Nabble.com.
> _______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.



_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev
Reply | Threaded
Open this post in threaded view
|

Re: VM Crash?

abergel
Okay! It works now.
thanks

Alexandre


> On Jul 8, 2016, at 11:24 AM, Henrik Nergaard <[hidden email]> wrote:
>
> You need to draw it at the correct location in the world ;).
>
> surface displayOnMorphicCanvas: aCanvas at: self position
>
> Best regards,
> Henrik
>
> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf Of Alexandre Bergel
> Sent: Friday, July 8, 2016 11:20 AM
> To: Moose-related development <[hidden email]>
> Subject: [Moose-dev] Re: VM Crash?
>
> Hi Henrik,
>
> thank thank you to look into that.
>
> Indeed, that error looks familiar
>
>> I guess you can try to use AthensCairoSurface>>#displayOnMorphicCanvas:at:
>> in TRMorph>>#drawOn: instead of how it is done now, this should assure
>> that the Surface is still alive when drawing.
>
>
> In TRMorph>>drawOn:, replacing
>
> aCanvas image: f at: self bounds origin sourceRect: (0 @ 0 extent: surface extent) rule: 34
>
> by
> surface displayOnMorphicCanvas: aCanvas at: 0 @ 0
>
> does not draw anything. Did I do something wrong?
>
>
>> I also think that changing the extent of the surface should be moved
>> from #layoutChanged to the start of #drawOn:. The layoutChanged/ morph
>> extent change could happen more than once in between draw calls,  
>> which could cause unnecessary recreations of the surface. It should
>> also ensure that the surface cannot be recreated while inside #drawOn:.
>
> Thanks! I have committed this
> Running all the Roassal2- tests leaves 5 instances of AthensCairoSurface that cannot be garbage collected.
> With the change you are suggesting, it is reduced to 4. Still something!
>
> Cheers,
> Alexandre
>
>>
>> Hope this helps.
>>
>> Best regards,
>> Henrik
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> --
>> View this message in context:
>> http://forum.world.st/VM-Crash-tp4905304p4905497.html
>> Sent from the Moose mailing list archive at Nabble.com.
>> _______________________________________________
>> Moose-dev mailing list
>> [hidden email]
>> https://www.list.inf.unibe.ch/listinfo/moose-dev
>
> --
> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
> Alexandre Bergel  http://www.bergel.eu
> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>
>
>
> _______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev
> _______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.list.inf.unibe.ch/listinfo/moose-dev

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.



_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.list.inf.unibe.ch/listinfo/moose-dev