Code Analysis-run time error prediction

Overview
• Introduction
• Existing solutions
• Run time errors
• Design
• Implementation
• Future Work

Code Analysis
Difference between project success & failure.

• If there's going to be a program, there has to be
construction.
• Code is often the only accurate description of the
software available.
• Code must follow coding standards and code
conventions.

Source code Conventions
• 80% of the lifetime cost of a piece of software goes to
maintenance.
• Hardly any software is maintained for its whole life by
the original author.
• Code conventions improve the readability of the
software.
• Source code like any other product should be well
packaged

Code optimization based analysis
• Code Verification and Run-Time Error prediction at
compile time using syntax directed translation.
• Predict run time errors without program execution or
test cases.
• Uses Intermediate Code

Possible Run time Errors
1) Detecting uninitialized Variables

Using variables before they have been initialized by the
program can cause unpredictable results

2) Detecting Overflows, Underflows, and Divide by Zeros

Consider pseudo-code:

X=X/(X-Y)

Identifying all possible causes for error on the
operation:

o X and Y may not be initialized

X-Y may overflow or underflow

- X and Y may be equal and cause a division by

zero

e X/(X–Y) may overflow or underflow

All possible values of x & y in
program p

If the value of x & y both fall on the black line, there is a
divide by zero error.

3) Detecting incorrect argument data types and incorrect
number of arguments

• Checking of arguments for type and for the correct order of
occurrence.

• Requires both the calling program and the called program
to be compiled with a special compiler option.

• Checks can be made to determine if the number and types
of arguments in function (and subroutine) calls are consistent
with the actual function definitions.

4) Detecting errors with strings at run-time

• A string must have a null terminator at the end of the
meaningful data in the string. A common mistake is to not
allocate room for this extra character.

This can also be a problem with dynamic allocation.

char * copy_str = malloc( strlen(orig_str) + 1);

strcpy(copy_str, orig_str);

• The strlen() function returns a count of the data characters
which does not include the null terminator.

• In the case of dynamic allocation, it might corrupt the heap

a. Detecting Out-of-bounds indexing of statically and
dynamically allocated arrays

A common run-time error is the reading and writing of arrays
outside of their declared bounds.

b. Detecting Out-of-Bounds Pointer References

A common run-time error for C and C++ programs occurs
when a pointer points to memory outside its associated
memory block.

Pseudo code for out of bound
references
for(i=0;i<5;i++)

A[i]=i;

p=A;

for(i=0;i<=5;i++)

p++;

a=*p;

/* out-of-bounds reading using pointers */

5) Detecting Memory Allocation and Deallocation Errors

• A memory deallocation error occurs when a portion of
memory is deallocated more than once.

• Another common source of errors in C and C++ programs is
an attempt to use a dangling pointer. A dangling pointer is a
pointer to storage that is no longer allocated.

6) Detecting Memory Leaks

• A program has a memory leak if during execution the
program loses its ability to address a portion of memory
because of a programming error;

• A pointer points to a location in memory and then all the
pointers pointing to this location are set to point somewhere
else

• A function/subroutine is called, memory is allocated during
execution of the function/subroutine, and then the memory
is not deallocated upon exit and all pointers to this memory
are destroyed

Source code analyzer predicates

Reliable: Proven free of run-
time errors and under all
operating conditions within
the scope
Faulty: Proven faulty each
time the operation is
executed.
Dead: Proven unreachable
(may indicate a functional
issue)
Unproven: Unproven code
section or beyond the scope of
the analyzer.

Specifications

•Why Java for developing
analyser?

Specifications

•Why C/C++ as input language?

Design for Code Analyzer
Input program

(C File)

S
Lexical Analyzer y
m
b
o
l
T
a
b
l
Parser e

IC(SDT)

Generation

Run Time
Error Predictions

Analysis of Code

Input Program

Lexical Analysis-Stream Tokenizer

Parser-
Condition = "(" Expression ("=="|"!="|">"|"<"|">="|"<=")
Expression ")"
Expression = Term {("+"|"-") Term}
Term = Factor {("*"|"/") Factor}
Factor = number |
identifier |

Intermediate code generation: Postfix Evaluation

3 address code generation
Target Source File:
argument operator operand operand result
Test(n){ 1 2

int b,a,n,j; 0 < j n

if(j<n) 1 if 0 gotol0
{
2 + a b
a=a+b;}
3 = a 2
}
l0:

Further Work

• Evaluation of intermediate code for performing data
flow and control flow analysis.
• Prediction of run time errors using intermediate code.
• Using code optimization techniques such as constant
folding to predict code behavior

REFERENCES
• A V. Aho, R Sethi, J D. Ullman., Compilers: Principles, Techniques and
Tools, 2nd ed. , Addison-Wesley Pub. Co.
• G R. Luecke, J Coyle, J Hoekstra “A Survey of Systems for Detecting
Serial Run-Time Errors”, The Iowa State University's High Performance
Computing Group, Concurrency and Computation. : Practice and
Experience. 18, 15(Dec. 2006), 1885-1907.
• T Erkkinen, C Hote “Code Verification and Run-Time Error Detection
Through Abstract Interpretation”, AIAA Modeling and Simulation
Technologies Conference and Exhibit ,21 - 24 Aug 2006, Keystone,
Colorado.
• PolySpace Client for C/C++ 6 datasheet. Available HTTP:
http://www.mathworks.com/products/polyspaceclientc.html.
• D.M. Dhamdhere, Compiler Construction, Tata McGraw-Hill.
• Semantic designs, “Flow analysis for control and data”, Available
HTTP: http://www.semdesigns.com/Products/DMS/FlowAnalysis.html.

Code Analysis-run time error prediction

More Related Content

What's hot

Viewers also liked

Similar to Code Analysis-run time error prediction

Recently uploaded

Code Analysis-run time error prediction