bdemsky [Sat, 25 Jun 2011 00:48:40 +0000 (00:48 +0000)]
code change
bdemsky [Sat, 25 Jun 2011 00:16:36 +0000 (00:16 +0000)]
turn off debugging messages....another bug fix
bdemsky [Fri, 24 Jun 2011 23:39:08 +0000 (23:39 +0000)]
bug fix
david [Fri, 24 Jun 2011 23:12:24 +0000 (23:12 +0000)]
Updating shared values
david [Fri, 24 Jun 2011 23:09:19 +0000 (23:09 +0000)]
updated shared locations
bdemsky [Fri, 24 Jun 2011 00:37:51 +0000 (00:37 +0000)]
bug fixes
bdemsky [Wed, 22 Jun 2011 22:57:17 +0000 (22:57 +0000)]
remove some debug code...fix a bug
bdemsky [Wed, 22 Jun 2011 22:18:01 +0000 (22:18 +0000)]
more bug fixes
yeom [Wed, 22 Jun 2011 17:21:40 +0000 (17:21 +0000)]
small changes on test cases. disable -noloop option from makefile.
david [Wed, 22 Jun 2011 17:04:19 +0000 (17:04 +0000)]
First annoted version
david [Wed, 22 Jun 2011 17:03:25 +0000 (17:03 +0000)]
First annotated version
bdemsky [Wed, 22 Jun 2011 11:15:36 +0000 (11:15 +0000)]
bug fixes
bdemsky [Wed, 22 Jun 2011 08:08:00 +0000 (08:08 +0000)]
changes
yeom [Wed, 22 Jun 2011 00:27:05 +0000 (00:27 +0000)]
bug fixes + integrate loop termination analysis into ssjava checking
- add an interface to access loop invariant from LoopOptimize
- add "TERMINATE" annotation for method method declaration when developer wants to skip the loop termination analysis for an annotated method. There is no way to have direct annotations on the loop since Java only allows to have annotations on declaration nodes
bdemsky [Tue, 21 Jun 2011 22:59:36 +0000 (22:59 +0000)]
try to make update phase work...need to ask for permission to write into memory blocks
yeom [Tue, 21 Jun 2011 22:31:47 +0000 (22:31 +0000)]
have Javadoc-style method comments.
yeom [Tue, 21 Jun 2011 21:29:14 +0000 (21:29 +0000)]
reflects today's comments
yeom [Tue, 21 Jun 2011 20:04:39 +0000 (20:04 +0000)]
changes.
bdemsky [Tue, 21 Jun 2011 07:48:35 +0000 (07:48 +0000)]
changes
bdemsky [Tue, 21 Jun 2011 01:36:18 +0000 (01:36 +0000)]
bug fixes...still doesn't work
yeom [Tue, 21 Jun 2011 01:33:20 +0000 (01:33 +0000)]
convert it to the must analysis
bdemsky [Tue, 21 Jun 2011 00:48:57 +0000 (00:48 +0000)]
more changes...think first version of memory allocation exists now
jzhou [Tue, 21 Jun 2011 00:43:52 +0000 (00:43 +0000)]
Bug fix for cache adaption
yeom [Mon, 20 Jun 2011 22:01:38 +0000 (22:01 +0000)]
provides makefile and makes it compile for david
yeom [Mon, 20 Jun 2011 18:55:25 +0000 (18:55 +0000)]
changes.
bdemsky [Mon, 20 Jun 2011 10:22:44 +0000 (10:22 +0000)]
only call mark once...
bdemsky [Mon, 20 Jun 2011 10:14:43 +0000 (10:14 +0000)]
get it to compile again
bdemsky [Mon, 20 Jun 2011 09:38:02 +0000 (09:38 +0000)]
more changes
bdemsky [Mon, 20 Jun 2011 06:52:47 +0000 (06:52 +0000)]
get rid of manually created table...
bdemsky [Mon, 20 Jun 2011 02:20:55 +0000 (02:20 +0000)]
more changes
bdemsky [Sun, 19 Jun 2011 22:42:22 +0000 (22:42 +0000)]
more changes
bdemsky [Sun, 19 Jun 2011 01:38:33 +0000 (01:38 +0000)]
get it to compile again
bdemsky [Sun, 19 Jun 2011 01:34:26 +0000 (01:34 +0000)]
clean up stuff...will add back later...
yeom [Sun, 19 Jun 2011 01:12:23 +0000 (01:12 +0000)]
changes.
bdemsky [Sat, 18 Jun 2011 08:38:55 +0000 (08:38 +0000)]
get it to compile
bdemsky [Sat, 18 Jun 2011 08:21:32 +0000 (08:21 +0000)]
more changes
bdemsky [Sat, 18 Jun 2011 08:20:46 +0000 (08:20 +0000)]
code changes
bdemsky [Sat, 18 Jun 2011 04:50:11 +0000 (04:50 +0000)]
bug fixes
yeom [Sat, 18 Jun 2011 02:04:30 +0000 (02:04 +0000)]
start implementing loop termination analysis: for the initial version, checking whether induction variable is getting bigger and loop guard condition is composed only with induction variables and loop invariant.
bdemsky [Sat, 18 Jun 2011 01:17:53 +0000 (01:17 +0000)]
change objects
bdemsky [Sat, 18 Jun 2011 00:13:20 +0000 (00:13 +0000)]
more changes
bdemsky [Thu, 16 Jun 2011 21:21:01 +0000 (21:21 +0000)]
more of my changes
bdemsky [Thu, 16 Jun 2011 09:56:59 +0000 (09:56 +0000)]
code for generating mark bit tables
bdemsky [Thu, 16 Jun 2011 09:52:05 +0000 (09:52 +0000)]
code to do mark bits...
bdemsky [Wed, 15 Jun 2011 08:50:39 +0000 (08:50 +0000)]
changes to get rid of all type errors...
bdemsky [Wed, 15 Jun 2011 07:42:08 +0000 (07:42 +0000)]
bug fix...gcloads was used twice for different things with different types...
bdemsky [Wed, 15 Jun 2011 04:04:04 +0000 (04:04 +0000)]
changes
bdemsky [Wed, 15 Jun 2011 01:52:54 +0000 (01:52 +0000)]
more work towards type checking
bdemsky [Tue, 14 Jun 2011 23:08:04 +0000 (23:08 +0000)]
clean up typing mistakes
bdemsky [Tue, 14 Jun 2011 09:50:01 +0000 (09:50 +0000)]
fix mgchash table...isn't supposed to do allocation on the fly...nor use linked lists, etc...
bdemsky [Tue, 14 Jun 2011 09:17:55 +0000 (09:17 +0000)]
changes to support marked bit table...might work??
bdemsky [Tue, 14 Jun 2011 06:39:33 +0000 (06:39 +0000)]
changes
bdemsky [Tue, 14 Jun 2011 06:19:47 +0000 (06:19 +0000)]
more changes
bdemsky [Tue, 14 Jun 2011 06:15:32 +0000 (06:15 +0000)]
more changes...
yeom [Tue, 14 Jun 2011 00:01:51 +0000 (00:01 +0000)]
fixes on GLB and returnloc calculation, etc.
gets more class libraries
jzhou [Mon, 13 Jun 2011 17:53:50 +0000 (17:53 +0000)]
My changes
bdemsky [Sun, 12 Jun 2011 00:09:25 +0000 (00:09 +0000)]
remove extra locks and use atomics...hopefully i got this write...didn't test..
jjenista [Sat, 11 Jun 2011 23:14:01 +0000 (23:14 +0000)]
tinkering with debug stuff for barnes-hut
jjenista [Sat, 11 Jun 2011 23:12:33 +0000 (23:12 +0000)]
a mini version of barnes hut
jihoonl [Sat, 11 Jun 2011 22:56:19 +0000 (22:56 +0000)]
support multi threading
jjenista [Sat, 11 Jun 2011 18:41:55 +0000 (18:41 +0000)]
a test that actually worked fine, but shows bad convergence behavior
yeom [Sat, 11 Jun 2011 01:12:21 +0000 (01:12 +0000)]
fix bugs and changes on method checking
jjenista [Fri, 10 Jun 2011 18:26:28 +0000 (18:26 +0000)]
Bug fix: in some cases we really do want the heap region inherent reach set to have preds other than true, but when we allocate a new node, the state on that node should always have a true pred also.
bdemsky [Thu, 9 Jun 2011 16:42:58 +0000 (16:42 +0000)]
remove dependence on concat2...it isn't a standard string method
fix some bugs...
bdemsky [Thu, 9 Jun 2011 05:07:44 +0000 (05:07 +0000)]
many changes towards classpath...
bdemsky [Sun, 5 Jun 2011 08:23:44 +0000 (08:23 +0000)]
another bug fix
bdemsky [Sun, 5 Jun 2011 08:15:17 +0000 (08:15 +0000)]
code changes
bdemsky [Sun, 5 Jun 2011 08:07:45 +0000 (08:07 +0000)]
more changes towards jni
bdemsky [Sat, 4 Jun 2011 02:32:16 +0000 (02:32 +0000)]
add new file
bdemsky [Fri, 3 Jun 2011 23:50:21 +0000 (23:50 +0000)]
changes
bdemsky [Fri, 3 Jun 2011 23:40:22 +0000 (23:40 +0000)]
missing declaration
bdemsky [Fri, 3 Jun 2011 23:16:48 +0000 (23:16 +0000)]
file changes
yeom [Fri, 3 Jun 2011 23:03:56 +0000 (23:03 +0000)]
improves the strategy of checkings: starting from ssjava outermost loop, then only checking methods reachable from that loop. Also, check that every method which overrides an annotated method is also annotated.
jjenista [Fri, 3 Jun 2011 21:51:35 +0000 (21:51 +0000)]
fixed a bug where a single argument is passed in as more than one parameter, admitting only the last such edge
jjenista [Thu, 2 Jun 2011 23:43:16 +0000 (23:43 +0000)]
without this the call stie debug was capturing an immense amoutn of data
jjenista [Thu, 2 Jun 2011 18:38:24 +0000 (18:38 +0000)]
forgot to remove some commented out code
jjenista [Wed, 1 Jun 2011 23:50:01 +0000 (23:50 +0000)]
if a static field is encountered, ignore it gracefully
jjenista [Wed, 1 Jun 2011 23:48:35 +0000 (23:48 +0000)]
comment out import statements while we are between implementations of the import feature
jzhou [Wed, 1 Jun 2011 22:30:01 +0000 (22:30 +0000)]
Parallelize the cache adpation
jjenista [Wed, 1 Jun 2011 19:12:01 +0000 (19:12 +0000)]
check points-to module improvements along with a bug fix for disjoint call transfer func. The bug was when considering callee edges to bring to the caller, use the predicates of the edge only to decide
yeom [Wed, 1 Jun 2011 01:26:29 +0000 (01:26 +0000)]
having a location namespace. location names appeared in field lattice declaration are only valid within the declaration annotation. therefore a field location, which is an element of a composite location, will be specified by 'className.FieldLocation'. assuming that classname is unique right now but probably better to have a way to set each lattice with unique name.
yeom [Tue, 31 May 2011 23:48:40 +0000 (23:48 +0000)]
add one more checking, class inheritance: If a class B has a super class A, all relative orderings of class A must be respected by the ordering lattice of class B.
jjenista [Tue, 31 May 2011 18:05:19 +0000 (18:05 +0000)]
name flat IR graphs to include method signature to avoid losing some graphs
bdemsky [Fri, 27 May 2011 21:53:59 +0000 (21:53 +0000)]
add NOLOCK back in
adash [Thu, 26 May 2011 18:07:08 +0000 (18:07 +0000)]
enable error checking
Current the recovery SpamFilter version crashes when garbled characters are added to a filename when opening the file.
For example: email5 becomes email5% or email5+
The error checking identifies an incorrect filename passed to it....I think there may be a bug in the BufferedReader or
the FileInputStream that causes this problem and passes a wrong filename to the nativeopen().
adash [Thu, 26 May 2011 17:59:01 +0000 (17:59 +0000)]
some changes to allow running >1 thread per machine
yeom [Wed, 25 May 2011 23:52:18 +0000 (23:52 +0000)]
fixes on bugs and checking of method invocation:
when verifying callee and caller hierarchy matching,
caller has a constraint that output(return value) should be lower than inputs(parameters).
In regarding to this constraint, callee needs to return something lower than its parameters, which means there is relative ordering relations between parameters and return value in the callee hierarchy. If there is no relative odering relations between them, it is compile-time error.
yeom [Tue, 24 May 2011 00:45:34 +0000 (00:45 +0000)]
bug fixes and few extra things:
-static final field has TOP location by default
-do not allow to copy values from one class global variable to another class global variable since they're not comparable.
yeom [Mon, 23 May 2011 19:40:03 +0000 (19:40 +0000)]
having a new check that verifies the single reference constraint: only allows a single reference, not aliases. The reference location must be destroyed after a variable or field is read. plus, changes on ssjava class library and test cases according to new checking.
yeom [Fri, 20 May 2011 18:51:22 +0000 (18:51 +0000)]
fix bugs on the flow down rule and start annotating ssjava class library again
jjenista [Fri, 20 May 2011 18:26:31 +0000 (18:26 +0000)]
allow Pointer to run for usual compilation modes... added a todo that points-to checks at runtime need some extra modeling for string literals added to Pointer before that system will work.
stephey [Fri, 20 May 2011 09:42:02 +0000 (09:42 +0000)]
Fixed import bug but some other issues now pop up:
1) make rcrpointer now crashes at processNode in Pointer.java starting at line 439. The switch statement encounters a FlatLiteralNode and doesn't know what to do with it.
2) We fail to parse Hashtable.java in the gnu labrary when using the -jni option. The line of code it fails at is "Hashtable.this.clear();". I can't seem to find any resources telling me what having a "this" midline means, so I can't fix it myself.
yeom [Thu, 19 May 2011 22:17:28 +0000 (22:17 +0000)]
changes.
jjenista [Wed, 18 May 2011 23:47:59 +0000 (23:47 +0000)]
bug fix: heap contexts for call sites were propgated from caller to callee incorrectly. A heap context should propagate from the caller to each possible callee at virtual dispatch... how did this ever work??
adash [Wed, 18 May 2011 23:41:20 +0000 (23:41 +0000)]
changes
yeom [Wed, 18 May 2011 23:34:37 +0000 (23:34 +0000)]
more changes.
adash [Wed, 18 May 2011 20:09:21 +0000 (20:09 +0000)]
new changes
yeom [Wed, 18 May 2011 18:59:14 +0000 (18:59 +0000)]
add more test cases: delta func and local variable lattice
yeom [Wed, 18 May 2011 18:58:09 +0000 (18:58 +0000)]
changes.
yeom [Wed, 18 May 2011 01:43:55 +0000 (01:43 +0000)]
changes to reflect ssjava design changes and temporarily remove some of ssjava class libraries until code revision is done