Dev/ashish #753

xashisk · 2024-07-23T17:36:51Z

3 (out of previously 221) bugs remaining.
Code changes:

Made machineListByTypePerThread, machineSetPerThread and their 4 getter and setter functions, and getGlobalMachine() and addGlobalMachine() defined inside the abstract Scheduler.java i.e. local to each Scheduler object.
Add a replayScheduler field in PExplicitGlobal (default value is null). On replaying, we set it to the ReplayScheduler object. Changed PExplicitLogger.isReplaying() function to return true if replayScheduler field in PExplicitGlobal is not null.
Implemented PExplicitGlobal.getScheduler() which first checks isReplaying() or not. If yes, return replayScheduler. If not, get the current thread id, get corresponding localtID, and then return corresponding ExplicitSearchScheduler object. Change all calls to PexplicitGlobal.getMachineSet()/getGlobalMachine/addGlobalMachine to PexplicitGlobal.getScheduler().getMachineSet()/getGlobalMachine/addGlobalMachine
Retained mapping from tID to localtID in PExplicitGlobal, as well as localtID to ExplicitSearchScheduler object.
In class BugFoundException, added a local integer buggyLocalTID which is initialized in both constructors of BugFoundException. Then when a replayscheduler object is created in RunTimeExecutor, we add the following mapping to PExplicitGlobal.tID_to_localtID : [replaying thread ID] -> <local ID of buggy trace, found from e.buggyLocalTID, where e is an object of class BugFoundException>. Is required as many functions in Scheduler() use PExplicitGlobal.tID_to_localtID when replaying thread.

Current buggy benchmarks are:

P/Tst/RegressionTests/Integration/DynamicError/SEM_OneMachine_8
P/Tst/RegressionTests/Feature2Stmts/DynamicError/GotoStmt1
P/Tst/RegressionTests/Feature2Stmts/DynamicError/receive6

Old schedule/data choice changed to schedule/data SearchUnit Added new class for schedule/data Choice, which is just a wrapper around PMachineId/PValue<?> TODO: clean up Schedule with new class structure

This reverts commit 53c5d15.

[PEx] Make search strategies subclasses thread safe, minor refactoring

…hish

aman-goel · 2024-07-23T18:06:00Z

Src/PCompiler/CompilerCore/Backend/Java/MachineGenerator.cs

Please undo changes in this file. It is outside the scope of PExplicit-related code.

aman-goel · 2024-07-23T18:06:36Z

Src/PRuntimes/PExplicitRuntime/regressionTestScript.sh

Let's remove this file as well.

aman-goel · 2024-07-23T18:18:42Z

Now that there is a function PExplicitGlobal.getScheduler() to get the scheduler object, why is there PExplicitGlobal.getSchedulers()? That is redundant.

aman-goel · 2024-07-23T18:19:37Z

Src/PRuntimes/PExplicitRuntime/src/main/java/pexplicit/runtime/scheduler/Scheduler.java

+     * Mapping from machine type to list of all machine instances
+     */
+    @Getter
+    private final Map<Integer, Map<Class<? extends PMachine>, List<PMachine>>> machineListByTypePerThread = new ConcurrentHashMap<>(); // This is per thread; so make this map of tiD to same Map


Why is this per thread? Isn't each Scheduler object already per thread?

aman-goel · 2024-07-23T18:19:56Z

Src/PRuntimes/PExplicitRuntime/src/main/java/pexplicit/runtime/scheduler/Scheduler.java

+     * Set of machines
+     */
+    @Getter
+    private final Map<Integer, SortedSet<PMachine>> machineSetPerThread = new ConcurrentHashMap<>(); 


Same. Why is this per thread? Isn't each Scheduler object already per thread?

aman-goel · 2024-07-23T18:22:30Z

Src/PRuntimes/PExplicitRuntime/src/main/java/pexplicit/runtime/logger/PExplicitLogger.java

We should directly use PExplicit.getScheduler() throughout this and other files, instead of first getting localTID with PExplicit.getTId_toLocalTID() and then getting the Scheduler object using PExplicit.getSchedulers().get(localTID).

aman-goel · 2024-07-23T18:23:21Z

Src/PRuntimes/PExplicitRuntime/src/main/java/pexplicit/utils/exceptions/BugFoundException.java

    public BugFoundException(String message) {
        super(message);
+        buggyLocalTID = (PExplicitGlobal.getTID_to_localtID()).get(Thread.currentThread().getId());


Same. Directly use PExplicitGlobal.getScheduler().

aman-goel · 2024-07-23T18:23:29Z

Src/PRuntimes/PExplicitRuntime/src/main/java/pexplicit/utils/exceptions/BugFoundException.java

        PExplicitLogger.logBugFound(message);
    }

    public BugFoundException(String message, Throwable cause) {
        super(message, cause);
+        buggyLocalTID = (PExplicitGlobal.getTID_to_localtID()).get(Thread.currentThread().getId());


Same. Directly use PExplicitGlobal.getScheduler().

aman-goel · 2024-07-23T18:24:17Z

Src/PRuntimes/PJavaRuntime/src/main/java/prt/Monitor.java

Please remove all changes outside PExplicitRuntime codebase. They are not in scope here.

aman-goel · 2024-07-23T18:24:26Z

Src/PRuntimes/PJavaRuntime/src/test/java/MonitorTest.java

Same. Please remove all changes outside PExplicitRuntime codebase. They are not in scope here.

aman-goel · 2024-07-23T18:25:11Z

benchmarksRuns.sh

Let's remove this file. Keep it for your local experimentation only. No need to push upstream.

aman-goel

Shared a bunch of comments to address.

aman-goel · 2024-07-23T18:46:04Z

Looks like file PExplicitThreadLogger.java line 58 has a hard-coded path. That is making many CI tests fail.

        String filename = "/Users/xashisk/ashish-ws/SyncedForkedRepo/P/output/LogThread" + PExplicitGlobal.getTID_to_localtID().get(Thread.currentThread().getId()) + ".log";  // PIN: Configure output folder with CLI options.

aman-goel · 2024-07-23T21:10:38Z

Src/PRuntimes/PExplicitRuntime/src/main/java/pexplicit/commandline/PExplicitConfig.java

Looks like the default number of threads is set to 1. Can we try changing this to a higher number (say 4) and rerun the regression tests (i.e., mvn test).

…s are placed are inside the P folder (else can then change) 3. For nproc 4, added code for killing other threads when bugfoundException is raised

…rror: comment and uncomment L151 and L152, and run for both

aman-goel · 2024-08-05T22:31:06Z

Merging on dev branch of parallel PEx (dev/pex_parallel).

aman-goel and others added 30 commits June 13, 2024 18:50

[PEx] Change schedule choice from PMachine to PMachineId

122dc53

[PEx] Major revamping of Choice: 1/n

53c5d15

Old schedule/data choice changed to schedule/data SearchUnit Added new class for schedule/data Choice, which is just a wrapper around PMachineId/PValue<?> TODO: clean up Schedule with new class structure

Revert "[PEx] Major revamping of Choice: 1/n"

c6d87ed

This reverts commit 53c5d15.

[PEx] Separates unexplored choices from schedule

20f0216

[PEx] Cleanup and minor corrections to recent changes to SearchTask

9c9e7d0

[PEx] Minor correction

d8d9e7e

[PEx] Corrections to new backtracking logic

33c6ff3

Merge branch 'dev/pexplicit_checker' into dev/aman

82b298c

Ongoing Changes

ef2f17c

Week 6 tasks

673df13

discussed changes

b66d1f6

discussed changes

81815b4

Recent Changes

7c400c9

[PEx] Add thread-safe code for search strategies, minor cleanup

fdff598

[PEx] Refactoring and cleanup

f4c779c

Merge pull request #1 from xashisk/dev/aman

dafc672

[PEx] Make search strategies subclasses thread safe, minor refactoring

More changes

fd4e557

New changes

f8659a8

Changes

f0ae0d7

Changes

ec7e08c

Remove optional parameters in Monitor (p-org#747)

5da26ab

Changes

5ad287a

Merge branch 'p-org:master' into dev/ashish

6b405b4

Changes

611ea9c

Changes

37d329d

Merge branch 'dev/pex_parallel' into dev/ashish

19ff012

Local CHanges

b2fb517

All changes

ab8fa27

Merge branch 'dev/ashish' of https://github.com/xashisk/P into dev/as…

33d9cb7

…hish

Changes

5c0a9bf

aman-goel reviewed Jul 23, 2024

View reviewed changes

Src/PRuntimes/PExplicitRuntime/regressionTestScript.sh Outdated

Copy link

Contributor

aman-goel Jul 23, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let's remove this file as well.

Required changes

fb9b591

aman-goel reviewed Jul 23, 2024

View reviewed changes

benchmarksRuns.sh Outdated

Copy link

Contributor

aman-goel Jul 23, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let's remove this file. Keep it for your local experimentation only. No need to push upstream.

aman-goel requested changes Jul 23, 2024

View reviewed changes

aman-goel reviewed Jul 23, 2024

View reviewed changes

xashisk added 5 commits July 24, 2024 13:35

1. Changes made. 2. For hardcoded log path, need to know where script…

96f061f

…s are placed are inside the P folder (else can then change) 3. For nproc 4, added code for killing other threads when bugfoundException is raised

Hardcoded path changed

1a50e85

Added code for gracefully interrupting other threads, but confusing e…

4fc6683

…rror: comment and uncomment L151 and L152, and run for both

nproc: 4, bugs < 86

955b178

Changes

a87cbef

aman-goel merged commit 419f335 into p-org:dev/pex_parallel Aug 5, 2024
2 of 5 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Dev/ashish #753

Dev/ashish #753

xashisk commented Jul 23, 2024 •

edited

Loading

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel commented Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel left a comment

aman-goel commented Jul 23, 2024

aman-goel Jul 23, 2024

aman-goel commented Aug 5, 2024

Dev/ashish #753

Dev/ashish #753

Conversation

xashisk commented Jul 23, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aman-goel commented Jul 23, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aman-goel left a comment

Choose a reason for hiding this comment

aman-goel commented Jul 23, 2024

Choose a reason for hiding this comment

aman-goel commented Aug 5, 2024

xashisk commented Jul 23, 2024 •

edited

Loading