[Soot-list] Release class

Eric Bodden eric.bodden at mail.mcgill.ca
Wed Jul 30 14:59:35 EDT 2008


Ok in that case you can actually restrict Soot so that it only
processed a subset of your classes. See
http://www.sable.mcgill.ca/soot/tutorial/usage/index.html

Also -Xms does not really help with anything; you need to set the -Xmx
option, preferably to something more than 512MB, depending on what
exactly you are trying to do.

Hope that helps.

Eric

2008/7/30 Silviu ANDRICA <silviu.andrica at epfl.ch>:
> Hello,
>   I'm not trying to do full program analysis. I'm trying to do an
> intra-procedural analysis for a subset of the methods of each class. For a
> method, my analysis needs to build a CFG and a StrongLocalMustAliasAnalysis.
>   Regarding the heap space, I run java with -Xms512m.
>
> Silviu
> Eric Bodden wrote:
>>
>> Hi Silviu.
>>
>> This really depend on what exactly you are trying to do with Soot. If
>> you are doing a whole-program analysis then there is little hope;
>> otherwise there may be a chance. So I guess we need a little more
>> information about your plans in order to help you out. Also, how much
>> heap space do you give to the VM?
>>
>> Eric
>>
>> 2008/7/30 Silviu ANDRICA <silviu.andrica at epfl.ch>:
>>
>>>
>>> Hello all,
>>>  I am trying to analyze JBoss Application Server. This application has
>>> around 16,300 classes. The problem is that I load all of them at once and
>>> then proceed to analyze them. The analysis build Control Flow Graphs and
>>> some other big objects so, before I finish the whole application, I run
>>> out
>>> of memory, the GC quits. Is there a way I can "unload" unneeded JBoss
>>> classes? Or circumvent the need for loading library classes?
>>>
>>> Thank you,
>>>  Silviu
>>> _______________________________________________
>>> Soot-list mailing list
>>> Soot-list at sable.mcgill.ca
>>> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>>>
>>>
>>
>>
>>
>> --
>> Eric Bodden
>> Sable Research Group
>> McGill University, Montréal, Canada
>>
>
>



-- 
Eric Bodden
Sable Research Group
McGill University, Montréal, Canada


More information about the Soot-list mailing list