[Soot-list] How to obtain the topological graph in the transactional transformation phase (wjtp.tn)?

Richard L. Halpert richardlhalpert at gmail.com
Wed Dec 3 20:22:47 EST 2008


SynchronizedRegionFinder is definitely too strict regarding what bytecode
patterns it can work with.  This is something I hope to someday remedy...

-Rich

On Wed, Dec 3, 2008 at 1:02 PM, Eric Bodden <eric.bodden at mail.mcgill.ca>wrote:

> I am unable to reproduce the behavior you are describing.
>
> I do see an exception but not the kind of exception you reported when
> I run the current version of Soot on your LockGraphTest class.
>
> Soot started on Wed Dec 03 15:52:16 GMT-05:00 2008
> No main class given. Inferred 'LockGraphTest' as main class.
> [Call Graph] For information on where the call graph may be
> incomplete, use the verbose option to the cg phase.
> [Spark] Pointer Assignment Graph in 6.0 seconds.
> [Spark] Type masks in 0.7 seconds.
> [Spark] Pointer Graph simplified in 0.0 seconds.
> [Spark] Propagation in 88.9 seconds.
> [Spark] Solution found in 89.0 seconds.
> [wjtp.tn] *** Build May-Happen-in-Parallel Info *** Wed Dec 03
> 15:54:01 GMT-05:00 2008
> [mhp]
> [mhp] Main Thread (Single, ---,---):
> [mhp] [<LockGraphTest: void main(java.lang.String[])>
> [mhp]   <LockGraphTest: void setI(int)>
> [mhp]   <LockGraphTest: void <init>()>
> [mhp]   <LockGraphTest: void setF(LockGraphTest)>
> [mhp]   <LockGraphTest: float getF()>]
> [mhp]
> [wjtp.tn] *** Find and Name Transactions *** Wed Dec 03 15:54:01 GMT-05:00
> 2008
> Warning: using default implementation of addAll. You should implement
> a faster specialized implementation.
> this is of type soot.jimple.spark.sets.HashPointsToSet
> other is of type soot.jimple.spark.sets.HybridPointsToSet
> exclude is null
> Exception in thread "main" java.lang.RuntimeException: Unknown
> bytecode pattern: exitmonitor not followed by return, exitmonitor,
> goto, or throw
>        at
> soot.jimple.toolkits.thread.synchronization.SynchronizedRegionFinder.flowThrough(SynchronizedRegionFinder.java:270)
>        at
> soot.jimple.toolkits.thread.synchronization.SynchronizedRegionFinder.flowThrough(SynchronizedRegionFinder.java:18)
>        at
> soot.toolkits.scalar.ForwardFlowAnalysis.doAnalysis(ForwardFlowAnalysis.java:162)
>        at
> soot.jimple.toolkits.thread.synchronization.SynchronizedRegionFinder.<init>(SynchronizedRegionFinder.java:75)
>        at
> soot.jimple.toolkits.thread.synchronization.LockAllocator.internalTransform(LockAllocator.java:168)
>         at soot.SceneTransformer.transform(SceneTransformer.java:39)
>        at soot.Transform.apply(Transform.java:89)
>         at soot.ScenePack.internalApply(ScenePack.java:43)
>         at soot.Pack.apply(Pack.java:114)
>         at soot.PackManager.runWholeProgramPacks(PackManager.java:407)
>         at soot.PackManager.runPacks(PackManager.java:327)
>        at soot.Main.run(Main.java:202)
>        at soot.Main.main(Main.java:145)
>         at MyMain.main(MyMain.java:45)
>
>
> This exception only points to an apparent shortcoming of the current
> implementation of SynchronizedRegionFinder.
>
> Eric
>
> 2008/12/1 Marco Bakera <marco.bakera at tu-dortmund.de>:
> > You are right, Eric. The problem does not occur in any case. The
> following
> > piece of code provides such an example that runs into problems.
> >
> > public class LockGraphTest {
> >        private final Object lockI = new Object();
> >        private final Object lockF = new Object();
> >        private int iVal;
> >        private float fVal;
> >
> >        public LockGraphTest() {
> >        }
> >
> >        public void setI(int newValue) {
> >                synchronized(lockI) {
> >                        iVal = newValue;
> >                }
> >        }
> >
> >        public void setF(LockGraphTest lgt) {
> >                synchronized(lockF) {
> >                        fVal = (int) lgt.getF();
> >                }
> >        }
> >
> >        public float getF() {
> >                synchronized(lockF) {
> >                        return fVal;
> >                }
> >        }
> >
> >
> >        public static void main(String[] args) {
> >                LockGraphTest lgtF = new LockGraphTest();
> >                lgtF.setI(7);
> >
> >                LockGraphTest lgtI = new LockGraphTest();
> >                lgtI.setI(5);
> >                lgtI.setI(6);
> >                lgtI.setF(lgtF);
> >        }
> > }
> >
> > On Thursday 27 November 2008 15:39:35 Eric Bodden wrote:
> >> Marco does this happen when you apply Soot to any application or only
> >> if you apply it to one special program? In the latter case it would
> >> help to know what program that would be.
> >>
> >> Cheers,
> >> Eric
> >>
> >> 2008/11/27 Marco Bakera <marco.bakera at tu-dortmund.de>:
> >> > On Monday 10 November 2008 19:42:52 Richard L. Halpert wrote:
> >> >> 1) The LockAllocator uses a may-alias relationship to group locks
> >> >> together - so if lock1 in *may* be the same as lock2, they would be
> >> >> treated as a single lock (and the two critical sections would be put
> in
> >> >> the same CriticalSectionGroup).  The Lock allocator constructs this
> >> >> graph for the sake of deadlock detection.  You can use the deadlock
> >> >> detection graph by calling
> >> >>
> soot.jimple.toolkits.thread.synchronization.LockAllocation.getDeadlockGr
> >> >>aph (). The nodes are CriticalSectionGroups, and the edges represent
> >> >> calls from a critical section in one group to a critical section in
> >> >> another group.  Be sure to call Soot with "-p wjtp.tn
> >> >> avoid-deadlock:false", or else the deadlock detector may stop
> >> >> constructing the graph early if it finds potential deadlock.
> >> >
> >> > That sounds great and I will use this. :) Thanks for that hint.
> However
> >> > when trying the following lines of code.
> >> >
> >> >                String args =
> >> >                        "-w --app " +
> >> >                        // options for call graph creation
> >> >                        "-p cg.cha enabled:false " +
> >> >                        "-p cg.spark enabled:true " +
> >> >                        // options for whole jimple transformation pack
> >> >                        "-p wjtp.tn enabled:true " +
> >> >                        "-p wjtp.tn avoid-deadlock:false " +
> >> >                        "-p wjtp.tn do-tlo:false " +
> >> >                        "-p wjtp.tn do-mhp:true " +
> >> >                        // the main application class
> >> >                        mainClass.getName();
> >> >
> >> >                // start soot
> >> >                LOG.debug("Starting Soot with command line: "+args);
> >> >                Main.main(args.split(" "));
> >> >
> >> > I get the following exception from the main method:
> >> >
> >> > java.lang.RuntimeException: trying to hash null value.
> >> >        at soot.util.HashChain.getPredOf(HashChain.java:338)
> >> >        at soot.PatchingChain.getPredOf(PatchingChain.java:225)
> >> >        at
> >> >
> soot.toolkits.graph.ExceptionalUnitGraph.buildExceptionDests(ExceptionalU
> >> >nitGraph.java:337) at
> >> >
> soot.toolkits.graph.ExceptionalUnitGraph.initialize(ExceptionalUnitGraph.
> >> >java:263) at
> >> >
> soot.toolkits.graph.ExceptionalUnitGraph.<init>(ExceptionalUnitGraph.java
> >> >:149) at
> >> >
> soot.toolkits.graph.ExceptionalUnitGraph.<init>(ExceptionalUnitGraph.java
> >> >:182) at
> >> >
> soot.toolkits.scalar.LocalSplitter.internalTransform(LocalSplitter.java:7
> >> >7) at soot.BodyTransformer.transform(BodyTransformer.java:51) at
> >> > soot.Transform.apply(Transform.java:104)
> >> >        at
> soot.JimpleBodyPack.applyPhaseOptions(JimpleBodyPack.java:57)
> >> >        at soot.JimpleBodyPack.internalApply(JimpleBodyPack.java:89)
> >> >        at soot.Pack.apply(Pack.java:124)
> >> >        at
> >> > soot.coffi.CoffiMethodSource.getBody(CoffiMethodSource.java:117) at
> >> > soot.SootMethod.getBodyFromMethodSource(SootMethod.java:82) at
> >> > soot.SootMethod.retrieveActiveBody(SootMethod.java:329) at
> >> >
> soot.jimple.toolkits.callgraph.OnFlyCallGraphBuilder.processNewMethod(OnF
> >> >lyCallGraphBuilder.java:181) at
> >> >
> soot.jimple.toolkits.callgraph.OnFlyCallGraphBuilder.processReachables(On
> >> >FlyCallGraphBuilder.java:81) at
> >> > soot.jimple.spark.solver.OnFlyCallGraph.build(OnFlyCallGraph.java:55)
> at
> >> >
> soot.jimple.spark.solver.PropWorklist.handleVarNode(PropWorklist.java:123
> >> >) at
> soot.jimple.spark.solver.PropWorklist.propagate(PropWorklist.java:53)
> >> > at
> >> >
> soot.jimple.spark.SparkTransformer.internalTransform(SparkTransformer.jav
> >> >a:151) at soot.SceneTransformer.transform(SceneTransformer.java:39) at
> >> > soot.Transform.apply(Transform.java:89)
> >> >        at soot.RadioScenePack.internalApply(RadioScenePack.java:57)
> >> >        at
> >> >
> soot.jimple.toolkits.callgraph.CallGraphPack.internalApply(CallGraphPack.
> >> >java:47) at soot.Pack.apply(Pack.java:114)
> >> >        at soot.PackManager.runWholeProgramPacks(PackManager.java:406)
> >> >        at soot.PackManager.runPacks(PackManager.java:327)
> >> >        at soot.Main.run(Main.java:202)
> >> >        at soot.Main.main(Main.java:145)
> >> >
> >> > The problem is only present when using the development version of Soot
> >> > from the SVN repository. Release 2.3.0 has no problems. However, the
> >> > method to obtain the Deadlock Graph is not present there. :)
> >> >
> >> > Thanks for your help.
> >> >
> >> >
> >> >
> >> > Greetings,
> >> > Marco.
> >
>
>
>
> --
> Eric Bodden
> Sable Research Group, McGill University, Montréal, Canada
> Got an interesting job offer? http://www.bodden.de/hire-me/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.CS.McGill.CA/pipermail/soot-list/attachments/20081203/e0c579c5/attachment.htm


More information about the Soot-list mailing list