Distributed Verification of Multithreaded C Programs

1
Distributed Verification of Multi-threaded C
Programs
Stefan Edelkamp joint work with Damian
Sulewski and Shahid Jabbar
2
Motivation IO-HSF-SPIN
Same states in both parts
Arrives at the final state
Large jumps due to 2nd heuristic
Current state
Already seen final state
Arrives again at same final state
2.9 TB 20 days 1 node ---- 8 days 3 nodes
3
Overview

• Software Checking in StEAM

• Externalization

• Virtual Addresses

• Parallelization

4
Overview

• Software Checking in StEAM

• Externalization

• Virtual Addresses

• Parallelization

5
Software Checking

• Advantages

Building a model unnecessary
Learning specification language unnecessary
Checking can be done more often

• Disadvantages

- Code has to be executed
- Huge number of states
- Huge states
6
StEAM

• Can check concurrent C programs

• Uses a virtual machine for execution

• supports BFS, DFS, Best-First, A, IDA

• finds
• Deadlocks
• Assertion Violations
• Segmentation Faults

7
StEAM - Checking a C Program
Model checker
igcc Compiler
Virtual Machine
char globalChar int globalBlocksize 7 int
main() allocateBlock(blocksize) void
allocateBlock(int size) void
memBlock memBlock (void ) malloc(size)
Objectcode
8
StEAM - Interpreting the Object Code
ICVM Virtual Machine
Register
char globalChar int globalBlocksize 7 int
main() allocateBlock(blocksize) void
allocateBlock(int size) void
memBlock memBlock (void ) malloc(size)
Text Section
Objectcode
BSS Section
Data Section
Stack
Memory Pool
9
StEAM Generating States
ICVM Virtual Machine
Register
Text Section
BSS Section
Data Section
Stack
Memory Pool
10
Overview

• Software Checking in StEAM

• Externalization

• Virtual addresses

• Parallelization

11
Externalization - Motivation
time
Internal
External
problem size
12
Externalization Mini States
EJMRS 06

• pointer to a state in RAM or on Disk

• pointer to the predecessor mini state

• constant size

13
Externalization Expanding a State
Cache
14
Externalization Flushing the Cache
Cache
Mini States
15
Externalization Collapse Compression
State
Caches
Files on Disk
Register
Text Section
BSS Section
Data Section
Stack
Memory Pool
16
Overview

• Software Checking in StEAM

• Externalization

• Virtual Addresses

• Parallelization

17
Virtual Addresses

• programs request memory

• memory assignment done by system

• moving program between nodes impossible

• two possible strategies

• converting the addresses before executing

• using virtual addresses

18
Virtual Addresses Memory Management
Memory pool
19
Virtual Addresses - Overhead
time
virtual
real
nodes
20
Overview

• Software Checking in StEAM

• Externalization

• Virtual Addresses

• Parallelization

21
Parallelization Motivation

• Distributed (Shared) Memory
• ? MPI channels/shared RAM communication
• Sending full states too expensive (if not used
  for expansion)
• ?Exploit externalization
• ? DualChannel (Speedup vs. Load Balance)
• ?Appropriate State Space Partitioning

22
Parallelization Dual Channel Communication
23
Parallelization Hash Partitioning

• Partitioning by hashing full state
• Problem Successors often not in same partition ?
  high communication overhead
• Partitioning by hashing partial state,
• e.g. memory pool
• Problem Too many states map to one hash value ?
  Load balancing

24
Parallelization Incremental Tree Hashing
EM05
h(s) (Si si 3i) mod 17
h(1,2,3,1,2,2,1,2) 4132 93(22) mod 17
11
h(3,1) 3319 mod 17 1
h(2,2,1,2) 9 6h(2,1,2)31 613 mod 17
h(1,2) 1329 mod 17 4
h(2) 231 mod 17 6
25
Parallelization Search Partitioning
horizontal slices
vertical slices
DFS Holzman Bosnacki 2006
Best-First, A
26
Parallelization - Hardware

• Cluster Vision System (PBS)
• Linux Suse 10.0
• MPI via infiniband
• Files via GBit Ethernet
• 224 nodes (464 procs), lt 15 used
• AMD Opteron DP 50 (2.4 GHz)

27
Experiments 15-Puzzle Partial Hash
speedup
time
nodes
28
Experiments Depth-First Slicing 200 Philosophers
time
Top Result 600 Phils / 6 nodes 97 KB /state
Ex Collapse Compression Distribution 16GB ? 1.5
GB per node
processors
29
Experiments - Bath-Tub Effect (50 phils-avg.)
Time
validates Holzmann Bosnacki
Size of Depth Layer
30
Experiment - Shared Memory Bakery (pthread)

• 4 Opteron MP 852 (2.6 GHZ)

speedup
time
nodes
31
Conclusion

• Preceeding Work Full Externalization of States,
  inIO-HSF-SPIN ? Constant-Size RAM, e.g. 1.8 GB
  RAM, 20 days 1 proc, 8 days 4 procs, 2.9TB disk
  EJ06, Distribution via (gh)-Value
• Problem Huge Highly Dynamic States
• Solution Mini States as Constant Size Finger
  Prints of States in RAM for Dual-Channel
  Communication to combine External and Parallel
  Search with Memory-Pool, Best-First Slicing
  Partitioning