Michael Walker
4237055938
Make 'lineNum' produce a String.
2016-02-09 18:46:03 +00:00
Michael Walker
5acab89d6d
Add support for named threads in testing.
...
This is such a huge change because I had made a LOT of assumptions
about thread IDs being numbers. I also took the opportunity to do
a bit of refactoring with CRef and CVar identifiers.
2016-02-09 18:37:50 +00:00
Michael Walker
077650920e
Add functions to MonadConc and MonadSTM to name things
2016-02-09 17:15:06 +00:00
Michael Walker
00c5b22262
Bump to lts-4
2016-01-30 19:32:47 +00:00
Michael Walker
4ba1e19a53
Remove double traversal of BPOR
2015-12-04 01:12:37 +00:00
Michael Walker
9e21c0546e
Fix detection of aborted branches in toDotSmall
2015-12-04 00:00:33 +00:00
Michael Walker
4cef3c60b8
'grow' doesn't need to keep track of the crstate
2015-12-03 22:42:06 +00:00
Michael Walker
c1d4259178
Don't introduce a dependency for blocked actions
...
This is safe because there's no point pre-empting a block to cause a
context switch, as the context switch will happen immediately after
the block anyway.
This speeds up the Applicative composition test from over 30 minutes
to around 1.
2015-12-03 22:09:09 +00:00
Michael Walker
ef9a89ee9c
Simplify "ap (side effects)" test case
2015-12-03 21:36:07 +00:00
Michael Walker
7ebad3dd88
Store the 'CRState' inside 'BacktrackStep'.
2015-12-03 18:44:35 +00:00
Michael Walker
982ad6daae
Split up 'findBacktrack' a bit
2015-12-03 18:11:35 +00:00
Michael Walker
5217492c24
Remove _bignore. Sleep sets make it largely irrelevant now.
2015-12-02 17:24:13 +00:00
Michael Walker
b3b275cdb1
Add a 'noBounds' binding for convenience.
2015-12-02 14:40:05 +00:00
Michael Walker
a18a7d2fa5
Support unbounded DPOR ( closes #34 )
...
If no bounds are enabled, just backtrack to the point identified.
2015-12-02 14:35:22 +00:00
Michael Walker
c64ecbee94
Factor out common backtracking code
2015-12-02 04:43:19 +00:00
Michael Walker
cae5704ac9
Enable building tests with GHC <7.10
2015-12-01 05:13:47 +00:00
Michael Walker
37a5886c83
Make a separate package for the test suite. Closes #31 .
2015-12-01 05:07:56 +00:00
Michael Walker
bd340e6eb3
Fix for pre-FTP GHC
2015-12-01 04:29:46 +00:00
Michael Walker
0ff053eaae
Add a Dining Philosophers example
2015-12-01 04:27:58 +00:00
Michael Walker
cdcaebb26c
Add aborting predicates
2015-12-01 04:27:47 +00:00
Michael Walker
1e0eb2a172
Add typeclass laws test/example
2015-12-01 04:12:02 +00:00
Michael Walker
e75c1ee24a
Change test suite to use test-framework
2015-12-01 04:11:44 +00:00
Michael Walker
cb146a7769
Correctly count pre-emptions in a schedule
2015-12-01 04:10:23 +00:00
Michael Walker
acf2150788
Add a variant of 'toDot' which prunes aborted subtrees
2015-12-01 03:18:30 +00:00
Michael Walker
cb77ae234a
Correctly identify CVar read/write dependencies
2015-12-01 03:18:05 +00:00
Michael Walker
917c8a3954
Add missing symmetric cases to dependentActions
2015-11-30 22:08:44 +00:00
Michael Walker
efa65859b1
Add setNumCapabilities. Closes #23 .
2015-11-30 22:08:44 +00:00
Michael Walker
42f0132ff4
Add some more typeclass instances
2015-11-30 22:08:44 +00:00
Michael Walker
b84614b46f
Add a function to convert a 'BPOR' into a GraphViz graph
2015-11-30 22:08:44 +00:00
Michael Walker
e50152874b
Properly implement sleep sets.
...
- Don't make decisions in bporSched which are in the sleep set.
- Don't consider backtracking points from executions which terminated
due to every decision being in the sleep set.
2015-11-30 22:08:42 +00:00
Michael Walker
a47d8c00e7
Move willRelease to Deterministic.Internal.Common
2015-11-30 22:06:44 +00:00
Michael Walker
96d0964ea4
Return the most recent state when aborting.
2015-11-30 22:06:44 +00:00
Michael Walker
b13650b892
Don't pre-empt thread termination with exceptions.
...
If a thread is about to terminate gracefully, there's no point in
pre-empting the `Stop` action to throw an exception to it.
2015-11-30 22:06:43 +00:00
Michael Walker
5371e74ff0
Make PFL bounding the default
2015-11-30 22:06:43 +00:00
Michael Walker
3327358520
Introduce a dependency between the final action and everything.
...
Introducing length bounding failed to find a deadlock in the Dining
Philosophers! I believe that the typical DPOR algorithm assumes that
every thread terminates before the entire process terminates, which
simply isn't the case in Haskell.
To combat this, a dependency was introduced between the final action
in a trace and everything. Just introducing a dependency for (0, Stop)
isn't enough, as execution can be aborted due to the length bound on
any action.
2015-11-30 22:06:40 +00:00
Michael Walker
258d8dfc17
Add length bounding
2015-11-18 12:31:06 +00:00
Michael Walker
49a903c6e3
Add newtype for preemption bounds and fair bounds
2015-11-17 19:29:35 +00:00
Michael Walker
b6ceaede99
Abort SCT execution early if there are no decisions in bound. Closes #28 .
2015-11-17 19:16:19 +00:00
Michael Walker
68ef26727b
Give schedulers the trace so far
2015-11-17 19:16:13 +00:00
Michael Walker
d1643bb360
Enable building with Stackage lts-2
2015-11-17 19:15:56 +00:00
Michael Walker
f3be1d12ab
Replace type synonyms with newtypes. Closes #29 .
2015-11-17 19:15:52 +00:00
Michael Walker
c6a702f725
Use ThreadId in definition of Conc, not Int
2015-11-17 18:03:03 +00:00
Michael Walker
3d3119ffd6
Expand test suite
2015-11-16 02:48:54 +00:00
Michael Walker
bc04990dbd
Fix off-by-one in adding backtracking points
2015-11-16 02:46:22 +00:00
Michael Walker
17ddc217c4
Unsynchronised reads and writes ARE dependent
2015-11-16 02:45:53 +00:00
Michael Walker
4081a6d7d7
Fix error with stepCommit argument order
2015-11-16 01:49:35 +00:00
Michael Walker
74dce99d71
Make stepModRefCas strict in the new value
2015-11-15 16:53:10 +00:00
Michael Walker
64057590db
Merge branch 'primops-simp'. Closes #20 .
...
This only implements the atomic-primops operations which deal with CRefs,
as handling relaxed memory in the presence of MutVar#s is really hard.
2015-11-15 15:45:12 +00:00
Michael Walker
343e097f83
Implement casCRef and modifyCRefCAS
2015-11-15 15:42:45 +00:00
Michael Walker
12f0a118ef
Store threads in an IntMap
2015-11-15 15:42:45 +00:00
