jjenista [Thu, 27 Oct 2011 17:39:48 +0000 (17:39 +0000)]
exploring error injection space
jjenista [Wed, 26 Oct 2011 15:48:44 +0000 (15:48 +0000)]
the static inner classes are not being used, but Subu's work is seeing them and complaining, so I moved them out. Also twiddled the Player code to generate a normal and error output sampling for experiments, WITHOUT SSJAVA notation, so the analysis won't run on this benchmark at the moment
jjenista [Wed, 26 Oct 2011 15:46:28 +0000 (15:46 +0000)]
tweak scripts to take mp3 filename argument
jjenista [Wed, 26 Oct 2011 15:38:49 +0000 (15:38 +0000)]
restrict error injection to non-array primitive types so we can use a random value instead of zero to get a wackier program behavior deviation
yeom [Tue, 25 Oct 2011 23:41:16 +0000 (23:41 +0000)]
add longer mp3 file for test
jjenista [Tue, 25 Oct 2011 22:34:55 +0000 (22:34 +0000)]
def reach coming along
jjenista [Tue, 25 Oct 2011 17:42:33 +0000 (17:42 +0000)]
implementing def reach transfer funcs for R
spikeuci [Tue, 25 Oct 2011 00:03:29 +0000 (00:03 +0000)]
Checking in the test cases
spikeuci [Tue, 25 Oct 2011 00:01:10 +0000 (00:01 +0000)]
As discussed with Prof. Demsky/Jim, the constructor of inner class now by default takes the surrounding class arg based on the scope and the block statement for initializing is also set up. TODO: redirect inner class references to surrounding class. the test case substituting thisbash.outerclass works in the current build
jjenista [Sat, 22 Oct 2011 01:20:29 +0000 (01:20 +0000)]
decouple ssjava code gen to prevent crashes from the analysis, so we can turn it on just for error injection
jjenista [Sat, 22 Oct 2011 01:17:09 +0000 (01:17 +0000)]
automate trying different error probabilities and seeds
jjenista [Fri, 21 Oct 2011 23:55:04 +0000 (23:55 +0000)]
instrument an injected error and plot differences in output samples
jjenista [Fri, 21 Oct 2011 21:16:41 +0000 (21:16 +0000)]
hacks
jjenista [Fri, 21 Oct 2011 21:16:00 +0000 (21:16 +0000)]
some extras for multiviewmaps
jjenista [Fri, 21 Oct 2011 21:15:08 +0000 (21:15 +0000)]
Please don't check in debug lines, por favor :)
jjenista [Fri, 21 Oct 2011 18:59:13 +0000 (18:59 +0000)]
a start on error inject for SSJava
yeom [Fri, 21 Oct 2011 17:32:10 +0000 (17:32 +0000)]
moves MP3Decoder codes to Benchmark directory
jjenista [Fri, 21 Oct 2011 16:19:47 +0000 (16:19 +0000)]
going to start with just enough definite reach analysis to do a simple example
david [Fri, 21 Oct 2011 01:49:47 +0000 (01:49 +0000)]
added array access inference
spikeuci [Fri, 21 Oct 2011 01:32:58 +0000 (01:32 +0000)]
Some changes for signature of constructor of inner classes, need to add assignment
david [Fri, 21 Oct 2011 00:44:26 +0000 (00:44 +0000)]
Fixed a bug, added switch statement support
david [Fri, 21 Oct 2011 00:43:18 +0000 (00:43 +0000)]
changed treenode tag to blockstatementnode tag
david [Fri, 21 Oct 2011 00:42:25 +0000 (00:42 +0000)]
adds global, this, and static tags for var id
jjenista [Thu, 20 Oct 2011 18:11:24 +0000 (18:11 +0000)]
a short, clear example that definite reachability will improve the analysis of
spikeuci [Sun, 16 Oct 2011 08:23:43 +0000 (08:23 +0000)]
Step 1: Creating instances of all levels of enclosing classes in an inner class. TODO need to populate them while actually creating the inner class and indirect all accesses as necessay
jzhou [Thu, 13 Oct 2011 23:30:40 +0000 (23:30 +0000)]
More changes for Galois
jjenista [Thu, 13 Oct 2011 18:05:36 +0000 (18:05 +0000)]
OoOJava benchmarks, including barnes-hut NOT parallelizing, are in working order
jjenista [Thu, 13 Oct 2011 15:28:12 +0000 (15:28 +0000)]
Forgot to add the OoOJava-specific class directory
jjenista [Thu, 13 Oct 2011 02:34:32 +0000 (02:34 +0000)]
Bug fix, sometimes an edge matches except it has a different type of source node, so check for that instead of blindly casting and causing an exception
jjenista [Thu, 13 Oct 2011 02:33:14 +0000 (02:33 +0000)]
Allow buildscript user to set the compiler's JVM heap size, so you can set it differently for different hosts
jjenista [Thu, 13 Oct 2011 02:31:04 +0000 (02:31 +0000)]
move heap size of output program into the benchmark-specific makefile
spikeuci [Wed, 12 Oct 2011 01:46:07 +0000 (01:46 +0000)]
Adding support for default constructor to be created in case ctor is not found
spikeuci [Wed, 12 Oct 2011 01:44:26 +0000 (01:44 +0000)]
*** empty log message ***
jzhou [Sat, 8 Oct 2011 00:08:03 +0000 (00:08 +0000)]
More classes for galois
jjenista [Fri, 7 Oct 2011 21:16:48 +0000 (21:16 +0000)]
fix a bug for Taint similar to ReachState, when either is an element of an ExistPred, SCRUB the preds on the taint and reach state before adding it, and within ExistPred use the equalsIgnorePreds and hashCodeNoPreds variants
jjenista [Wed, 5 Oct 2011 19:18:51 +0000 (19:18 +0000)]
accidentally commited tinker code yesterday
bdemsky [Wed, 5 Oct 2011 17:44:18 +0000 (17:44 +0000)]
changes
bdemsky [Wed, 5 Oct 2011 17:40:41 +0000 (17:40 +0000)]
extend test
bdemsky [Wed, 5 Oct 2011 17:38:48 +0000 (17:38 +0000)]
inner class test case
jjenista [Tue, 4 Oct 2011 23:49:19 +0000 (23:49 +0000)]
Bug fix, subtle errors in exist pred hashcode, and a cycle pred.hashCode to state.hashCode and back
jjenista [Tue, 4 Oct 2011 23:43:02 +0000 (23:43 +0000)]
a test that exposed a cyclic hashcode bug and some related minor bugs
jjenista [Tue, 4 Oct 2011 15:46:35 +0000 (15:46 +0000)]
A test that exposed the bug that I just fixed
jjenista [Tue, 4 Oct 2011 15:43:46 +0000 (15:43 +0000)]
the bug I've been chasing, a critical one-letter typo, GAH
jjenista [Tue, 4 Oct 2011 15:42:45 +0000 (15:42 +0000)]
double-check the context conversions of out-of-context node names
jjenista [Tue, 4 Oct 2011 15:40:17 +0000 (15:40 +0000)]
slight code improvements
yeom [Tue, 4 Oct 2011 00:55:33 +0000 (00:55 +0000)]
SSJava runtime support
jzhou [Mon, 3 Oct 2011 23:31:14 +0000 (23:31 +0000)]
Changes for galois porting
jzhou [Mon, 3 Oct 2011 23:26:56 +0000 (23:26 +0000)]
More classes for galois
yeom [Mon, 3 Oct 2011 00:28:34 +0000 (00:28 +0000)]
annotated version.
yeom [Mon, 3 Oct 2011 00:28:08 +0000 (00:28 +0000)]
bug fix.
yeom [Sat, 1 Oct 2011 00:18:01 +0000 (00:18 +0000)]
add input sensor data file
yeom [Fri, 30 Sep 2011 23:14:15 +0000 (23:14 +0000)]
makefile
yeom [Fri, 30 Sep 2011 23:13:25 +0000 (23:13 +0000)]
start new benchmark JavaNator.
jjenista [Tue, 27 Sep 2011 23:16:50 +0000 (23:16 +0000)]
put unit tests here
jjenista [Tue, 27 Sep 2011 19:07:06 +0000 (19:07 +0000)]
just documenting what this is for
jjenista [Mon, 26 Sep 2011 22:20:07 +0000 (22:20 +0000)]
Incrementing on definite reach analysis
jjenista [Thu, 22 Sep 2011 23:47:25 +0000 (23:47 +0000)]
Be able to get def reach state out at a given program point
jjenista [Thu, 22 Sep 2011 22:09:39 +0000 (22:09 +0000)]
get hooked into disjoint
jjenista [Thu, 22 Sep 2011 18:19:01 +0000 (18:19 +0000)]
getting definite reach analysis set up as a fixed point piggy-back to disjoint
jjenista [Wed, 21 Sep 2011 23:32:41 +0000 (23:32 +0000)]
Starting implementation for definite reachability analysis
jjenista [Wed, 21 Sep 2011 23:25:41 +0000 (23:25 +0000)]
guard a Tilera-only line
jjenista [Wed, 21 Sep 2011 21:08:35 +0000 (21:08 +0000)]
A MultiViewMap is a generalization of the VarSrcTokTable in OoOJava. It is cumbersome to set up, but very powerful, and it will serve as the underlying implementation of the four maps in Definite Reachability Analysis. When that is implemented, check there for examples of MultiViewMap in the wild.
yeom [Wed, 21 Sep 2011 00:11:56 +0000 (00:11 +0000)]
it passes all checkings.
yeom [Tue, 20 Sep 2011 23:52:56 +0000 (23:52 +0000)]
take out the loop termination analysis from the java optimization phase. it will happens during the SSJava checking.
yeom [Tue, 20 Sep 2011 23:13:09 +0000 (23:13 +0000)]
bug fix on the Loop terminate analysis: as long as a field is not appeared in a flat set field node, it is suitable for the guard condition variable
yeom [Tue, 20 Sep 2011 21:11:08 +0000 (21:11 +0000)]
passes the flow down rule
yeom [Tue, 20 Sep 2011 01:41:12 +0000 (01:41 +0000)]
bug fixes + annotations
yeom [Sat, 17 Sep 2011 23:27:54 +0000 (23:27 +0000)]
1) it passes the linear type checking 2) start annotations 3) removes unnecessary class libraries
yeom [Fri, 16 Sep 2011 23:29:02 +0000 (23:29 +0000)]
EyeTracking has the same result of the original Java version.
yeom [Fri, 16 Sep 2011 21:16:58 +0000 (21:16 +0000)]
found compilation errors from david's recent commits . have interim fixes to make it compile.
yeom [Fri, 16 Sep 2011 18:03:21 +0000 (18:03 +0000)]
change a way to store image pixels as same as Java's PixelGrabber does
yeom [Fri, 16 Sep 2011 01:15:03 +0000 (01:15 +0000)]
more changes.
yeom [Thu, 15 Sep 2011 01:24:23 +0000 (01:24 +0000)]
changes toward compilation
yeom [Wed, 14 Sep 2011 01:40:58 +0000 (01:40 +0000)]
add eye tracking benchmark.
get rid of all XML serialization.
david [Wed, 7 Sep 2011 23:14:38 +0000 (23:14 +0000)]
added returnnode, loopnode, and opnode recognitino for inference
david [Wed, 7 Sep 2011 00:53:20 +0000 (00:53 +0000)]
Added implicit flow for if statements
david [Sat, 3 Sep 2011 00:55:04 +0000 (00:55 +0000)]
adding java files for inference engine
david [Sat, 3 Sep 2011 00:53:01 +0000 (00:53 +0000)]
adding test class for SSJava Inference Engine
yeom [Thu, 1 Sep 2011 23:07:45 +0000 (23:07 +0000)]
changes: 1) refactoring codes 2) forgot to implement ownsership checking in the flow-down rule. it checks that right hand side pointer variable should be owned by the current method.
yeom [Wed, 31 Aug 2011 01:26:46 +0000 (01:26 +0000)]
it passes the definite clearance analysis.
yeom [Tue, 30 Aug 2011 07:17:05 +0000 (07:17 +0000)]
changes.
yeom [Mon, 29 Aug 2011 18:28:58 +0000 (18:28 +0000)]
changes: have a better way to keep the set of reading shared variables to verify their clearance.
yeom [Mon, 29 Aug 2011 08:47:00 +0000 (08:47 +0000)]
fixes on the definite clearance for shared locations.
yeom [Mon, 29 Aug 2011 00:57:15 +0000 (00:57 +0000)]
changes on the definite clearance for shared locations analysis
yeom [Sat, 27 Aug 2011 05:02:41 +0000 (05:02 +0000)]
it passes the definitely written analysis.
the current implementation of the synthesis filter assumes to have the previous internal array values of the previous granule when it combines the subbands to the original PCM signal. So it calls for introducing the temporal data structure keeping the previous ones and making it fall down once per a loop. seems to sacrifice memory spaces for desireable properties but still looking for a better idea to handle it.
jzhou [Fri, 26 Aug 2011 18:29:56 +0000 (18:29 +0000)]
Add check to see if the msg buffer is full during msg sending
yeom [Fri, 26 Aug 2011 00:58:55 +0000 (00:58 +0000)]
changes: since the inverse-MDCT uses input samples from the previous output, make sure that every SSJava loop iteration drops the raw_output of IMDCT and the previous block buffer by one height.
yeom [Tue, 23 Aug 2011 23:19:05 +0000 (23:19 +0000)]
1) changes on the definitely written analysis: it only takes care about locations that are written to inside of the SSJava loop
2) bug fix on the definitely written analysis: static method doesn't have implicit 'this' argument.
3) add a set of static functions that initialize array elements
4) changes on mp3decoder: move init() method out of SSJava loop and start to use SSJava array init method
yeom [Tue, 23 Aug 2011 00:36:26 +0000 (00:36 +0000)]
bug fix
yeom [Tue, 23 Aug 2011 00:34:41 +0000 (00:34 +0000)]
having new variable 'inter' in-between "reorder/antialias" and "hybrid" in order to remove unnecessary the shared location. (haven't yet updated annotations)
yeom [Sat, 20 Aug 2011 10:56:25 +0000 (10:56 +0000)]
bug fix on the definitely written check: Field read does not need to remove OW set.
yeom [Sat, 20 Aug 2011 08:49:24 +0000 (08:49 +0000)]
changes.
yeom [Sat, 20 Aug 2011 00:57:51 +0000 (00:57 +0000)]
mp3decoder passes the loop termination analysis.
yeom [Sat, 20 Aug 2011 00:56:31 +0000 (00:56 +0000)]
bug fix on the loop termination analysis
yeom [Fri, 19 Aug 2011 18:03:48 +0000 (18:03 +0000)]
bug fix: the loop entrance node of DOWHILELOOP is the begin node, not the condition node.
yeom [Fri, 19 Aug 2011 02:30:46 +0000 (02:30 +0000)]
changes on the loop termination analysis: associate a labeled statement with a corresponding LoopNode rather than attaching it to a BlockNode. this is interim implementation until we have branching statements.
yeom [Fri, 19 Aug 2011 00:00:48 +0000 (00:00 +0000)]
mp3decoder finally passes the flow-down rule checking.
yeom [Thu, 18 Aug 2011 12:17:32 +0000 (12:17 +0000)]
more changes to pass the flow-down rule
yeom [Thu, 18 Aug 2011 01:50:57 +0000 (01:50 +0000)]
changes.
yeom [Wed, 17 Aug 2011 23:18:16 +0000 (23:18 +0000)]
bug fix on flow-down rule: check if all of assignments done by invoking method respect the current branch constraint