Updating 4.5 or getting a recent build of 4.6

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

Updating 4.5 or getting a recent build of 4.6

Jeff Gonis-2
Hi Folks,

I have had some time recently and decided to get back into Squeak. I grabbed Eliot's latest cog vm for windows and went to update my 4.5 image from the trunk. I hit a VM crash after a minute or two of processing updates.  Trying several other vintages of Eliot's vm did not help, and I also went to the Squeak ftp site and grabbed the image in the 4.6 folder and attempted to update that and ran into the same problem. I am pretty sure this crash is the crash that was discussed on the list a while ago with a corrupted mcm/mcz file. I thought that it had been resolved, but maybe I misunderstood, or maybe I am missing a step here.

Anyway, the TL;DR of this mail is: How do I avoid crashing the vm when updating from the trunk on windows, or is there a place to grab a more recent vm that can be updated safely.  Figuring out either of these will work for me. I am happy to post the crash dump if that will help.

Thanks for any and all help!
Jeff


Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

marcel.taeumel (old)
Did you try updating from the trunk image instead of the 4.5 image?

http://ftp.squeak.org/trunk/

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

Re: Updating 4.5 or getting a recent build of 4.6

marcel.taeumel (old)
Oops, directly jumped to the "TL;DR" part. :D Okay, we need to fix the process of entering the trunk path... Downloading the trunk image and updating it should work fine.

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

Re: Updating 4.5 or getting a recent build of 4.6

Jeff Gonis-2
Hi Marcel,

I was pretty sure that I had tried that file, but grabbed it from that link just to be sure and tried to update again, and unfortunately no luck.  Here is a link to the crash dump file that is generated: https://www.dropbox.com/s/tu8wmro478xmt1t/crash.dmp?dl=0

At least for myself, all I do to reproduce is open the image using the latest Cog, go to the world menu and hit update.

Thanks for any and all help,
Jeff


Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

Jeff Gonis-2
Hi Folks,

Just wanted to PING the list again to see if anyone had any feedback.
Am I just seeing a crash that no one else is seeing and it is
something with my setup, is there a safe VM version that people are
using to update from the trunk image, and is anyone else able to grab
the image that Marcel posted and do a successful update on the trunk?

I am just playing around with the image Marcel posted and it is just
fine for getting work done, but my OCD tendencies flare up a bit
knowing that I am not working with fully updated code.

Any help or at least just letting me know that this is a problem on my
end rather than the trunk would be greatly appreciated.

Thanks,
Jeff

Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

Eliot Miranda-2
Hi Jeff,

On Wed, Oct 29, 2014 at 1:06 PM, Jeff Gonis <[hidden email]> wrote:
Hi Folks,

Just wanted to PING the list again to see if anyone had any feedback.
Am I just seeing a crash that no one else is seeing and it is
something with my setup, is there a safe VM version that people are
using to update from the trunk image, and is anyone else able to grab
the image that Marcel posted and do a successful update on the trunk?

I am just playing around with the image Marcel posted and it is just
fine for getting work done, but my OCD tendencies flare up a bit
knowing that I am not working with fully updated code.

Any help or at least just letting me know that this is a problem on my
end rather than the trunk would be greatly appreciated.

I'm not sure whats going on.  Perhaps your image has got into a strange state.  I have no problems updating the 4.5 image from the all-in-one to trunk as part of the Spur image bootstrap.  See  http://www.squeakvm.org/svn/squeak/branches/Cog/image/buildspurtrunkimage.sh/  You might try checking-out  http://www.squeakvm.org/svn/squeak/branches/Cog/image and running the script to check.  The script downloads an up-to-date VM, the All-in-one, changes the update repository, updates, then builds a VMMaker image and then uses that to convert the updated trunk to Spur.  This has been working well for a couple of months.

HTH
--
best,
Eliot


Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

David T. Lewis
In reply to this post by Jeff Gonis-2
Hi Jeff,

Nice to see you on the list, and sorry for not responding earlier.

I am fairly sure that we had some things screwed up in the update stream
(since resolved) and you are experiencing the effects of that. I have a
feeling that most of us working with trunk images we are keeping up to
date with the stream, and we may have neglected to go back and fix up
whatever update maps needed to be patched in order to work around the
problem.

I don't know if this will make any difference, but the latest version of
the traditional interpreter VM for Windows is at

http://www.squeakvm.org/win32/
http://www.squeakvm.org/win32/release/Squeak4.10.2-2612.win32-i386.zip


> Hi Folks,
>
> Just wanted to PING the list again to see if anyone had any feedback.
> Am I just seeing a crash that no one else is seeing and it is
> something with my setup, is there a safe VM version that people are
> using to update from the trunk image, and is anyone else able to grab
> the image that Marcel posted and do a successful update on the trunk?
>
> I am just playing around with the image Marcel posted and it is just
> fine for getting work done, but my OCD tendencies flare up a bit
> knowing that I am not working with fully updated code.
>
> Any help or at least just letting me know that this is a problem on my
> end rather than the trunk would be greatly appreciated.
>
> Thanks,
> Jeff
>



Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

David T. Lewis
Sorry, I hit "send" by accident. What I meant to say is that the
interpreter VM may get you past the crash that you experienced, or it
might just crash differently. If you have the patience you can give it a
try and see if it helps.

Either way, I'm pretty sure that the problem is in the update stream as
opposed to the VM.

Dave

> Hi Jeff,
>
> Nice to see you on the list, and sorry for not responding earlier.
>
> I am fairly sure that we had some things screwed up in the update stream
> (since resolved) and you are experiencing the effects of that. I have a
> feeling that most of us working with trunk images we are keeping up to
> date with the stream, and we may have neglected to go back and fix up
> whatever update maps needed to be patched in order to work around the
> problem.
>
> I don't know if this will make any difference, but the latest version of
> the traditional interpreter VM for Windows is at
>
> http://www.squeakvm.org/win32/
> http://www.squeakvm.org/win32/release/Squeak4.10.2-2612.win32-i386.zip
>
>
>> Hi Folks,
>>
>> Just wanted to PING the list again to see if anyone had any feedback.
>> Am I just seeing a crash that no one else is seeing and it is
>> something with my setup, is there a safe VM version that people are
>> using to update from the trunk image, and is anyone else able to grab
>> the image that Marcel posted and do a successful update on the trunk?
>>
>> I am just playing around with the image Marcel posted and it is just
>> fine for getting work done, but my OCD tendencies flare up a bit
>> knowing that I am not working with fully updated code.
>>
>> Any help or at least just letting me know that this is a problem on my
>> end rather than the trunk would be greatly appreciated.
>>
>> Thanks,
>> Jeff
>>
>
>
>



Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

douglas mcpherson
In reply to this post by Eliot Miranda-2

---- Eliot Miranda <[hidden email]> wrote:

> Hi Jeff,
>
> On Wed, Oct 29, 2014 at 1:06 PM, Jeff Gonis <[hidden email]> wrote:
>
> > Hi Folks,
> >
> > Just wanted to PING the list again to see if anyone had any feedback.
> > Am I just seeing a crash that no one else is seeing and it is
> > something with my setup, is there a safe VM version that people are
> > using to update from the trunk image, and is anyone else able to grab
> > the image that Marcel posted and do a successful update on the trunk?
> >
> > I am just playing around with the image Marcel posted and it is just
> > fine for getting work done, but my OCD tendencies flare up a bit
> > knowing that I am not working with fully updated code.
> >
> > Any help or at least just letting me know that this is a problem on my
> > end rather than the trunk would be greatly appreciated.
> >
>
> I'm not sure whats going on.  Perhaps your image has got into a strange
> state.  I have no problems updating the 4.5 image from the all-in-one to
> trunk as part of the Spur image bootstrap.  See
> http://www.squeakvm.org/svn/squeak/branches/Cog/image/buildspurtrunkimage.sh/
>  You might try checking-out
> http://www.squeakvm.org/svn/squeak/branches/Cog/image and running the
> script to check.  The script downloads an up-to-date VM, the All-in-one,
> changes the update repository, updates, then builds a VMMaker image and
> then uses that to convert the updated trunk to Spur.  This has been working
> well for a couple of months.
>
> HTH

Just a caveat about the scripts Eliot mentions. Due to recent flux in the layout of the All-in-One, [at least some of] the scripts stopped working. See

http://lists.squeakfoundation.org/pipermail/vm-dev/2014-October/016726.html

Doug

> --
> best,
> Eliot

Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

Levente Uzonyi-2
In reply to this post by Jeff Gonis-2
On Wed, 29 Oct 2014, Jeff Gonis wrote:

> Hi Folks,
>
> Just wanted to PING the list again to see if anyone had any feedback.
> Am I just seeing a crash that no one else is seeing and it is
> something with my setup, is there a safe VM version that people are
> using to update from the trunk image, and is anyone else able to grab
> the image that Marcel posted and do a successful update on the trunk?

I've checked your crash.dmp, and it looks interesting, especially this
part:

Stack backtrace:
  [80126560] ??? + 0 in (null)
  [8008DE63] next: + 107 in CogCode
  [7FFF07C8] ceReturnToInterpreterTrampoline + 0 in CogCode
  [7FFF06F8] ceBaseFrameReturnTrampoline + 0 in CogCode

Smalltalk stack dump:
   0x99553c M SocketStream>receiveData: 0x812fa8ac: a(n) SocketStream
   0x99555c M SocketStream>next: 0x812fa8ac: a(n) SocketStream
   0x9955ac I HTTPSocket class>httpRequest:url:headers:content:response: 0x80513ad0: a(n) HTTPSocket class
   0x995cc0 I HTTPSocket class>httpGetDocument:args:accept:request: 0x80513ad0: a(n) HTTPSocket class
   0x995cec M HTTPSocket class>httpGet:args:accept:request: 0x80513ad0: a(n) HTTPSocket class
   0x995d24 I HTTPSocket class>httpGet:args:user:passwd: 0x80513ad0: a(n) HTTPSocket class
   0x995d54 M [] in MCHttpRepository>readStreamForFileNamed:do: 0x80dfe9d8: a(n) MCHttpRepository
   0x995d70 M BlockClosure>on:do: 0x812e22e4: a(n) BlockClosure
   0x995d98 M [] in MCHttpRepository>displayProgress:during: 0x80dfe9d8: a(n) MCHttpRepository
   0x995dc0 M [] in MorphicUIManager>displayProgress:at:from:to:during: 0x8057fb00: a(n) MorphicUIManager
0x812e9b08 w BlockClosure>on:do:

I can't tell what went wrong, but it doesn't seem like it's related to the
ZipPlugin changes.
Can you tell how far the update process can go? (which update map, and
which package is being installed when the crash happens)
Does it always crash at the same point?


Levente


>
> I am just playing around with the image Marcel posted and it is just
> fine for getting work done, but my OCD tendencies flare up a bit
> knowing that I am not working with fully updated code.
>
> Any help or at least just letting me know that this is a problem on my
> end rather than the trunk would be greatly appreciated.
>
> Thanks,
> Jeff
>
>

Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

Jeff Gonis-2
On Wed, Oct 29, 2014 at 3:24 PM, Levente Uzonyi <[hidden email]> wrote:

>
>
> I've checked your crash.dmp, and it looks interesting, especially this part:
>
> Stack backtrace:
>         [80126560] ??? + 0 in (null)
>         [8008DE63] next: + 107 in CogCode
>         [7FFF07C8] ceReturnToInterpreterTrampoline + 0 in CogCode
>         [7FFF06F8] ceBaseFrameReturnTrampoline + 0 in CogCode
>
> Smalltalk stack dump:
>   0x99553c M SocketStream>receiveData: 0x812fa8ac: a(n) SocketStream
>   0x99555c M SocketStream>next: 0x812fa8ac: a(n) SocketStream
>   0x9955ac I HTTPSocket class>httpRequest:url:headers:content:response:
> 0x80513ad0: a(n) HTTPSocket class
>   0x995cc0 I HTTPSocket class>httpGetDocument:args:accept:request:
> 0x80513ad0: a(n) HTTPSocket class
>   0x995cec M HTTPSocket class>httpGet:args:accept:request: 0x80513ad0: a(n)
> HTTPSocket class
>   0x995d24 I HTTPSocket class>httpGet:args:user:passwd: 0x80513ad0: a(n)
> HTTPSocket class
>   0x995d54 M [] in MCHttpRepository>readStreamForFileNamed:do: 0x80dfe9d8:
> a(n) MCHttpRepository
>   0x995d70 M BlockClosure>on:do: 0x812e22e4: a(n) BlockClosure
>   0x995d98 M [] in MCHttpRepository>displayProgress:during: 0x80dfe9d8: a(n)
> MCHttpRepository
>   0x995dc0 M [] in MorphicUIManager>displayProgress:at:from:to:during:
> 0x8057fb00: a(n) MorphicUIManager
> 0x812e9b08 w BlockClosure>on:do:
>
> I can't tell what went wrong, but it doesn't seem like it's related to the
> ZipPlugin changes.
> Can you tell how far the update process can go? (which update map, and which
> package is being installed when the crash happens)
> Does it always crash at the same point?
>
>
> Levente
>
>
>

Hi Levente,

So I might not be able to recreate that exact crash because following
David's advice and using the Interpreter VM caused the update to
succeed.  Prior to that however, it appeared to always crash in the
same location in a repeatable fashion.  Interestingly, after
completing the update with the Interpreter VM and quitting without
saving, the update proceeds using the same image and the cog VM. Could
some sort of caching of packages be the problem here and completing
the update successfully changes this?

Of course this fact makes it hard to reproduce the issue using the
image I was talking about, as it now works, but I will attempt to see
if I have any other images that I can reproduce the issue with and I
will post another crash dump and details if I am successful.

Thanks to everyone for your help!
Jeff

Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

Eliot Miranda-2
Hi Jeff,

    whenever you, or anyone, has an mage that repeatably crashes the VM, I'd appreciate you keeping a copy.  I may not have time to look at it, it may not be a VM bug, but if the copy isn't taken we'll never know...

On Wed, Oct 29, 2014 at 3:03 PM, Jeff Gonis <[hidden email]> wrote:
On Wed, Oct 29, 2014 at 3:24 PM, Levente Uzonyi <[hidden email]> wrote:
>
>
> I've checked your crash.dmp, and it looks interesting, especially this part:
>
> Stack backtrace:
>         [80126560] ??? + 0 in (null)
>         [8008DE63] next: + 107 in CogCode
>         [7FFF07C8] ceReturnToInterpreterTrampoline + 0 in CogCode
>         [7FFF06F8] ceBaseFrameReturnTrampoline + 0 in CogCode
>
> Smalltalk stack dump:
>   0x99553c M SocketStream>receiveData: 0x812fa8ac: a(n) SocketStream
>   0x99555c M SocketStream>next: 0x812fa8ac: a(n) SocketStream
>   0x9955ac I HTTPSocket class>httpRequest:url:headers:content:response:
> 0x80513ad0: a(n) HTTPSocket class
>   0x995cc0 I HTTPSocket class>httpGetDocument:args:accept:request:
> 0x80513ad0: a(n) HTTPSocket class
>   0x995cec M HTTPSocket class>httpGet:args:accept:request: 0x80513ad0: a(n)
> HTTPSocket class
>   0x995d24 I HTTPSocket class>httpGet:args:user:passwd: 0x80513ad0: a(n)
> HTTPSocket class
>   0x995d54 M [] in MCHttpRepository>readStreamForFileNamed:do: 0x80dfe9d8:
> a(n) MCHttpRepository
>   0x995d70 M BlockClosure>on:do: 0x812e22e4: a(n) BlockClosure
>   0x995d98 M [] in MCHttpRepository>displayProgress:during: 0x80dfe9d8: a(n)
> MCHttpRepository
>   0x995dc0 M [] in MorphicUIManager>displayProgress:at:from:to:during:
> 0x8057fb00: a(n) MorphicUIManager
> 0x812e9b08 w BlockClosure>on:do:
>
> I can't tell what went wrong, but it doesn't seem like it's related to the
> ZipPlugin changes.
> Can you tell how far the update process can go? (which update map, and which
> package is being installed when the crash happens)
> Does it always crash at the same point?
>
>
> Levente
>
>
>

Hi Levente,

So I might not be able to recreate that exact crash because following
David's advice and using the Interpreter VM caused the update to
succeed.  Prior to that however, it appeared to always crash in the
same location in a repeatable fashion.  Interestingly, after
completing the update with the Interpreter VM and quitting without
saving, the update proceeds using the same image and the cog VM. Could
some sort of caching of packages be the problem here and completing
the update successfully changes this?

Of course this fact makes it hard to reproduce the issue using the
image I was talking about, as it now works, but I will attempt to see
if I have any other images that I can reproduce the issue with and I
will post another crash dump and details if I am successful.

Thanks to everyone for your help!
Jeff




--
best,
Eliot


Reply | Threaded
Open this post in threaded view
|

Re: Updating 4.5 or getting a recent build of 4.6

Jeff Gonis-2
Hi Eliot,

I will try to get some images together for you that are reproducible,
but in the meantime I have to head out for the evening.  One thing I
have noticed, looking through the crash dumps generated for all of the
crashes I have seen is that either at the very end of the primitive
trace, or 2 or 3 calls from the end of the primitive trace in every
single crash has been **CompactCode**

This is the only similarity between the cases in terms of the
primitive trace that I can see so far, because the stack dump has had
a variety of different methods being processed at the time.  I will
work to get a reproducible image for you, but in the meantime I
thought that might twig something for you.

Thanks for your time,
Jeff

On Wed, Oct 29, 2014 at 4:44 PM, Eliot Miranda <[hidden email]> wrote:
> Hi Jeff,
>
>     whenever you, or anyone, has an mage that repeatably crashes the VM, I'd
> appreciate you keeping a copy.  I may not have time to look at it, it may
> not be a VM bug, but if the copy isn't taken we'll never know...
>