Smalltalk › Squeak › Squeak VM

Cog VMMaker.oscog-eem.114/r2485

‹ Previous Topic Next Topic ›
 
classic Classic list List threaded Threaded
♦
♦
 8 messages Options
Eliot Miranda-2
Reply | Threaded
Open this post in threaded view
♦
♦
|

Cog VMMaker.oscog-eem.114/r2485

Eliot Miranda-2
 
Hi All,

    please find new VMs in http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a couple of long-standing bugs.  One is a rare crash in GC on copying a compiled method, caused by the VM not making sure that a copy of a method that has been compiled to machine code is an independent copy.  The other is a potential bug in GC with become where a become causes some machine code methods to gain references to young objects but the VM forgets to include these methods in the young referrers (those machine code methods that refer to young objects).
--
best,
Eliot
stephane ducasse-2
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

stephane ducasse-2

thanks eliot
looks like a vicious one to spot.

On Aug 8, 2011, at 7:55 AM, Eliot Miranda wrote:

> Hi All,
>
>     please find new VMs in http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a couple of long-standing bugs.  One is a rare crash in GC on copying a compiled method, caused by the VM not making sure that a copy of a method that has been compiled to machine code is an independent copy.  The other is a potential bug in GC with become where a become causes some machine code methods to gain references to young objects but the VM forgets to include these methods in the young referrers (those machine code methods that refer to young objects).
> --
> best,
> Eliot
>

Eliot Miranda-2
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

Eliot Miranda-2
In reply to this post by Eliot Miranda-2
 
Hi All,

    and now http://www.mirandabanda.org/files/Cog/VM/VM.r2486 includes Igor's rule 41 code in the BitBltPlugin, and includes a real multi-threaded linux VM.  The one in r2485 is only single-threaded due to a regression in a build script.

On Sun, Aug 7, 2011 at 10:55 PM, Eliot Miranda <[hidden email]> wrote:
Hi All,

    please find new VMs in http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a couple of long-standing bugs.  One is a rare crash in GC on copying a compiled method, caused by the VM not making sure that a copy of a method that has been compiled to machine code is an independent copy.  The other is a potential bug in GC with become where a become causes some machine code methods to gain references to young objects but the VM forgets to include these methods in the young referrers (those machine code methods that refer to young objects).
--
best,
Eliot




--
best,
Eliot
Igor Stasenko
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

Igor Stasenko

On 8 August 2011 22:49, Eliot Miranda <[hidden email]> wrote:
>
> Hi All,
>     and now http://www.mirandabanda.org/files/Cog/VM/VM.r2486 includes Igor's rule 41 code in the BitBltPlugin, and includes a real multi-threaded linux VM.  The one in r2485 is only single-threaded due to a regression in a build script.

Correction:
rule 41 code in the BitBltPlugin
is not mine. :)
I just took an effort to integrate already existing work made by other(s).
I had to change the methods in order to replace _ with := for
assignments and also use annotations <..foo.. > instead of 'self
foo:..'
that's why those methods stamped by my name.
But i didn't changed any semantics, just made sure it works.

>
> On Sun, Aug 7, 2011 at 10:55 PM, Eliot Miranda <[hidden email]> wrote:
>>
>> Hi All,
>>     please find new VMs in http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a couple of long-standing bugs.  One is a rare crash in GC on copying a compiled method, caused by the VM not making sure that a copy of a method that has been compiled to machine code is an independent copy.  The other is a potential bug in GC with become where a become causes some machine code methods to gain references to young objects but the VM forgets to include these methods in the young referrers (those machine code methods that refer to young objects).
>> --
>> best,
>> Eliot
>
>
>
> --
> best,
> Eliot
>
>



--
Best regards,
Igor Stasenko AKA sig.
Igor Stasenko
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

Igor Stasenko
In reply to this post by Eliot Miranda-2

On 8 August 2011 07:55, Eliot Miranda <[hidden email]> wrote:
>
> Hi All,
>     please find new VMs in http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a couple of long-standing bugs.  One is a rare crash in GC on copying a compiled method, caused by the VM not making sure that a copy of a method that has been compiled to machine code is an independent copy.  The other is a potential bug in GC with become where a become causes some machine code methods to gain references to young objects but the VM forgets to include these methods in the young referrers (those machine code methods that refer to young objects).

Yes, and i think this would be a first version(s) of Cog VMs, which
could be titled as 'stable'.
Because the bug, which Eliot fixed are really vicious and were plagued
us for more than half of year.

(i preparing an update on jenkins build server, we will see how well
it goes after the fix).
Because now we having a test jobs on jenkins which running tests for
newly built VMs (but i will write more details in separate post).



--
Best regards,
Igor Stasenko AKA sig.
Juan Vuletich-4
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

Juan Vuletich-4
In reply to this post by Eliot Miranda-2
 
Hi Eliot,

Both these give "You cannot use this version of the application Cog.app
with this version of Mac OS X." on 10.5.8. It looks like Apple wants to
force everybody to upgrade... :(

Cheers,
Juan Vuletich

Eliot Miranda wrote:

>  
>
> ------------------------------------------------------------------------
>
> Hi All,
>
>     and now http://www.mirandabanda.org/files/Cog/VM/VM.r2486 includes
> Igor's rule 41 code in the BitBltPlugin, and includes a real
> multi-threaded linux VM.  The one in r2485 is only single-threaded due
> to a regression in a build script.
>
> On Sun, Aug 7, 2011 at 10:55 PM, Eliot Miranda
> <[hidden email] <mailto:[hidden email]>> wrote:
>
>     Hi All,
>
>         please find new VMs in
>     http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a
>     couple of long-standing bugs.  One is a rare crash in GC on
>     copying a compiled method, caused by the VM not making sure that a
>     copy of a method that has been compiled to machine code is an
>     independent copy.  The other is a potential bug in GC with become
>     where a become causes some machine code methods to gain references
>     to young objects but the VM forgets to include these methods in
>     the young referrers (those machine code methods that refer to
>     young objects).
>     --
>     best,
>     Eliot
>
>
>
>
> --
> best,
> Eliot
>
>
> ------------------------------------------------------------------------
>
> No virus found in this message.
> Checked by AVG - www.avg.com <http://www.avg.com>
> Version: 10.0.1391 / Virus Database: 1518/3810 - Release Date: 08/04/11
> Internal Virus Database is out of date.
>

Eliot Miranda-2
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

Eliot Miranda-2
 


On Thu, Aug 11, 2011 at 1:42 PM, Juan Vuletich <[hidden email]> wrote:

Hi Eliot,

Both these give "You cannot use this version of the application Cog.app with this version of Mac OS X." on 10.5.8. It looks like Apple wants to force everybody to upgrade... :(

That's odd.  I've been compiling using the 10.5 sdk for a long time and that hasn't changed.  Here are extracts from my build log:

    setenv HEADER_SEARCH_PATHS "/Users/eliot/Cog/oscogvm/macbuild/build/Development/include /Developer/SDKs/MacOSX10.5.sdk/Developer/Headers/FlatCarbon /Users/eliot/Cog/oscogvm/macbuild"
    setenv PATH_PREFIXES_EXCLUDED_FROM_HEADER_DEPENDENCIES "/usr/include /usr/local/include /System/Library/Frameworks /System/Library/PrivateFrameworks /Developer/Headers /Developer/SDKs /Developer/Platforms"
    setenv PLATFORM_DEVELOPER_SDK_DIR /Developer/SDKs
    setenv SDKROOT /Developer/SDKs/MacOSX10.5.sdk
    setenv SDK_DIR /Developer/SDKs/MacOSX10.5.sdk
    setenv SDK_NAME macosx10.5
 
I wonder if I recently updated Xcode and that broke things :(

I'll take a look.

 

Cheers,
Juan Vuletich

Eliot Miranda wrote:
 
------------------------------------------------------------------------

Hi All,

   and now http://www.mirandabanda.org/files/Cog/VM/VM.r2486 includes Igor's rule 41 code in the BitBltPlugin, and includes a real multi-threaded linux VM.  The one in r2485 is only single-threaded due to a regression in a build script.

On Sun, Aug 7, 2011 at 10:55 PM, Eliot Miranda <[hidden email] <mailto:[hidden email]>> wrote:

   Hi All,

       please find new VMs in
   http://www.mirandabanda.org/files/Cog/VM/VM.r2485/.  These fix a
   couple of long-standing bugs.  One is a rare crash in GC on
   copying a compiled method, caused by the VM not making sure that a
   copy of a method that has been compiled to machine code is an
   independent copy.  The other is a potential bug in GC with become
   where a become causes some machine code methods to gain references
   to young objects but the VM forgets to include these methods in
   the young referrers (those machine code methods that refer to
   young objects).
   --     best,
   Eliot




--
best,
Eliot


------------------------------------------------------------------------

No virus found in this message.
Checked by AVG - www.avg.com <http://www.avg.com>
Version: 10.0.1391 / Virus Database: 1518/3810 - Release Date: 08/04/11
Internal Virus Database is out of date.





--
best,
Eliot
Juan Vuletich-4
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Cog VMMaker.oscog-eem.114/r2485

Juan Vuletich-4
 
Eliot Miranda wrote:

>
> On Thu, Aug 11, 2011 at 1:42 PM, Juan Vuletich <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>
>     Hi Eliot,
>
>     Both these give "You cannot use this version of the application
>     Cog.app with this version of Mac OS X." on 10.5.8. It looks like
>     Apple wants to force everybody to upgrade... :(
>
>
> That's odd.  I've been compiling using the 10.5 sdk for a long time
> and that hasn't changed.  Here are extracts from my build log:
>
>     setenv HEADER_SEARCH_PATHS
> "/Users/eliot/Cog/oscogvm/macbuild/build/Development/include
> /Developer/SDKs/MacOSX10.5.sdk/Developer/Headers/FlatCarbon
> /Users/eliot/Cog/oscogvm/macbuild"
>     setenv PATH_PREFIXES_EXCLUDED_FROM_HEADER_DEPENDENCIES
> "/usr/include /usr/local/include /System/Library/Frameworks
> /System/Library/PrivateFrameworks /Developer/Headers /Developer/SDKs
> /Developer/Platforms"
>     setenv PLATFORM_DEVELOPER_SDK_DIR /Developer/SDKs
>     setenv SDKROOT /Developer/SDKs/MacOSX10.5.sdk
>     setenv SDK_DIR /Developer/SDKs/MacOSX10.5.sdk
>     setenv SDK_NAME macosx10.5
>  
> I wonder if I recently updated Xcode and that broke things :(
>
> I'll take a look.
>
>  

I checked more carefully. The last version that runs is 2460. Starting
at 2462 all versions give the same message and refuse to run on 10.5.8.

Both 2460 and 2462 were built on 7/15. So it is something that changed
that very day.

Thanks,
Juan Vuletich
Free forum by Nabble Edit this page