Project

General

Profile

Wiki » History » Version 20

Andrei Tatarnikov, 07/10/2013 05:49 PM

1 1 Andrei Tatarnikov
h1. Constraint Solver API
2
3 2 Andrei Tatarnikov
h2. Basic Concepts
4 1 Andrei Tatarnikov
5 2 Andrei Tatarnikov
Constraint Solver Java API provides a Java API for generating pseudorandom values that satisfy certain constraints. This feature is important for test generators that aim at creating directed tests. At logical level, a constraint is represented by a set of expressions that specify limitations for input values (assertions that must be hold for those values). If there are values satisfying all of the specified assertions they will be used a solution for the constraint. If there is a multitude of values satisfying the constraint, specific values will be selected from the range of possible solutions on random basis.
6
7 4 Andrei Tatarnikov
From an implementational point of view, the API represents a wrapper around some kind of an openly distributed SMT solver engine (in the current version, we use the Z3 solver by Microsoft Research). It can be extended to support other solver engines and it provides a possibility to interact with different solver engines in a uniform way. Also, it facilitates creating task-specific custom solvers and extending functionality of existing solver engines by adding custom operations (macros based on built-in operations).
8 2 Andrei Tatarnikov
9 3 Andrei Tatarnikov
h2. SMT-LIB
10 2 Andrei Tatarnikov
11
In SMT solvers, a special functional language is used to specify constraints. The constraint solver subsystem generates constructions in the SMT language and runs the engine to process them and produce the results (find values of unknown input variables).
12
13 1 Andrei Tatarnikov
h2. Constraints and SMT
14
15
Constrains specified as an SMT model are represented by a set of assertions (formulas) that must be satisfied. An SMT solver checks the satisfiability of the model and suggests a solution (variable values) that would satisfy the model. In the example below, we specify a model that should help us create a test that will cause a MIPS processor to generate an exception. We want to find values of the rs and rt general purpose registers that will cause the ADD instruction to raise an integer overflow exception. It should be correct 32-bit signed integers that are not equal to each other. Here is an SMT script:
16
17 18 Andrei Tatarnikov
<pre>
18 1 Andrei Tatarnikov
(define-sort        Int_t () (_ BitVec 64))
19
20
(define-fun      INT_ZERO () Int_t (_ bv0 64))
21
(define-fun INT_BASE_SIZE () Int_t (_ bv32 64))
22
(define-fun INT_SIGN_MASK () Int_t (bvshl (bvnot INT_ZERO) INT_BASE_SIZE))
23
24
(define-fun IsValidPos ((x!1 Int_t)) Bool (ite (= (bvand x!1 INT_SIGN_MASK) INT_ZERO) true false))
25
(define-fun IsValidNeg ((x!1 Int_t)) Bool (ite (= (bvand x!1 INT_SIGN_MASK) INT_SIGN_MASK) true false))
26
(define-fun IsValidSignedInt ((x!1 Int_t)) Bool (ite (or (IsValidPos x!1) (IsValidNeg x!1)) true false))
27
28
(declare-const rs Int_t)
29
(declare-const rt Int_t)
30
31
; rt and rs must contain valid sign-extended 32-bit values (bits 63..31 equal)
32
(assert (IsValidSignedInt rs))
33
(assert (IsValidSignedInt rt))
34
35
; the condition for an overflow: the summation result is not a valid sign-extended 32-bit value
36
(assert (not (IsValidSignedInt (bvadd rs rt))))
37
38
; just in case: rs and rt are not equal (to make the results more interesting)
39
(assert (not (= rs rt)))
40
41
(check-sat)
42
43
(echo "Values that lead to an overflow:")
44
(get-value (rs rt))
45 18 Andrei Tatarnikov
</pre>
46 1 Andrei Tatarnikov
47 5 Andrei Tatarnikov
In an ideal case, each run of an SMT solver should return random values from the set of possible solutions. This should improve test coverage. Unfortunately, the current implementation is limited to a single solution that is constant for all run. This should be improved in the final version.
48
49 9 Sergey Smolov
h3. SMT Limitations.
50 5 Andrei Tatarnikov
51 8 Andrei Tatarnikov
# *Recursion in not allowed* in SMT-LIB. At least, this applies to the Z3 implementation. In other words, code like provided below is not valid:
52 5 Andrei Tatarnikov
53
<pre>
54
(define-fun fact ((x Int)) Int (ite (= x 0) 1 (fact (- x 1))))
55
(simplify (fact 10))
56
</pre>
57 1 Andrei Tatarnikov
58 12 Andrei Tatarnikov
h3. Constraints in XML
59
60 20 Andrei Tatarnikov
Constraints can also be described in the XML format. The API provides functionality to load and save constraints in XML. Here is an example of an XML document describing a simple constraint.
61 19 Andrei Tatarnikov
62 15 Andrei Tatarnikov
<pre><code class="xml">
63 12 Andrei Tatarnikov
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
64 1 Andrei Tatarnikov
<Constraint version="1.0">
65 16 Andrei Tatarnikov
    <Name>SimpleBitVector</Name>
66
    <Description>SimpleBitVector constraint</Description>
67 13 Andrei Tatarnikov
    <Solver id="Z3_TEXT"/>
68 1 Andrei Tatarnikov
    <Signature>
69 16 Andrei Tatarnikov
        <Variable length="3" name="a" type="BIT_VECTOR" value=""/>
70
        <Variable length="3" name="b" type="BIT_VECTOR" value=""/>
71 13 Andrei Tatarnikov
    </Signature>
72 1 Andrei Tatarnikov
    <Syntax>
73 12 Andrei Tatarnikov
        <Formula>
74 13 Andrei Tatarnikov
            <Expression>
75 16 Andrei Tatarnikov
                <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="NOT"/>
76 1 Andrei Tatarnikov
                <Expression>
77 16 Andrei Tatarnikov
                    <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="EQ"/>
78
                    <VariableRef name="a"/>
79 1 Andrei Tatarnikov
                    <VariableRef name="b"/>
80
                </Expression>
81
            </Expression>
82 12 Andrei Tatarnikov
        </Formula>
83 1 Andrei Tatarnikov
        <Formula>
84
            <Expression>
85
                <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="EQ"/>
86 13 Andrei Tatarnikov
                <Expression>
87 16 Andrei Tatarnikov
                    <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="BVOR"/>
88
                    <VariableRef name="a"/>
89
                    <VariableRef name="b"/>
90 1 Andrei Tatarnikov
                </Expression>
91 16 Andrei Tatarnikov
                <Value length="3" type="BIT_VECTOR" value="111"/>
92 13 Andrei Tatarnikov
            </Expression>
93 12 Andrei Tatarnikov
        </Formula>
94 13 Andrei Tatarnikov
        <Formula>
95
            <Expression>
96 16 Andrei Tatarnikov
                <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="EQ"/>
97 12 Andrei Tatarnikov
                <Expression>
98 16 Andrei Tatarnikov
                    <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="BVLSHL"/>
99
                    <VariableRef name="a"/>
100
                    <Value length="3" type="BIT_VECTOR" value="011"/>
101 12 Andrei Tatarnikov
                </Expression>
102 16 Andrei Tatarnikov
                <Expression>
103
                    <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="BVSMOD"/>
104
                    <VariableRef name="a"/>
105
                    <Value length="3" type="BIT_VECTOR" value="010"/>
106
                </Expression>
107 13 Andrei Tatarnikov
            </Expression>
108
        </Formula>
109 12 Andrei Tatarnikov
        <Formula>
110 13 Andrei Tatarnikov
            <Expression>
111 16 Andrei Tatarnikov
                <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="EQ"/>
112
                <Expression>
113
                    <Operation family="ru.ispras.solver.core.syntax.EStandardOperation" id="BVAND"/>
114
                    <VariableRef name="a"/>
115
                    <VariableRef name="b"/>
116
                </Expression>
117
                <Value length="3" type="BIT_VECTOR" value="000"/>
118 12 Andrei Tatarnikov
            </Expression>
119 1 Andrei Tatarnikov
        </Formula>
120 12 Andrei Tatarnikov
    </Syntax>
121 14 Andrei Tatarnikov
</Constraint>
122 15 Andrei Tatarnikov
</code></pre>
123 1 Andrei Tatarnikov
124
h2. Tree Representation
125
126
In our system, we use context-independent syntax trees to represent constraints. These trees are then used to generate a representation that can be understood by a particular SMT solver. Generally, it is an SMT model that uses some limited set of solver features applicable to microprocessor verification. The syntax tree contains nodes of the following types:
127
# Constraint. This is the root node of the tree. It holds the list of unknown variables and the list of assertions (formulas) for these variables.
128
# Formula. Represents an assertion expression. Can be combined with other formulas to build a more complex expression (by applying logic OR, AND or NOT to it). The underlying expression must be a logic expression that can be solved to true or false.
129
# Operation. Represents an unary or binary operation with some unknown variable, some value or some expression as parameters.
130
# Variable.Represents an input variable. It can have an assigned value and, in such a case, will be treated as a value. Otherwise, it is an unknown variable. A variable includes a type as an attribute.
131
# Value. Specifies some known value of the specified type which can be accessed as an attribute.
132
133
Note: Operation, Variables and Value are designed to be treated polymorphically. This allows combining them to build complex expressions.
134
135
h2. Constraint Solver Java Library
136
137
The Constraint Solver subsystem is implemented in Java. The source code files are located in the "microtesk++/constraint-solver" folder. The Java classes are organized in the following packages:
138
# ru.ispras.microtesk.constraints - contains SMT model generation logic and solver implementations.
139
# ru.ispras.microtesk.constraints.syntax - contains classes implementing syntax tree nodes.
140
# ru.ispras.microtesk.constraints.syntax.types - contains code that specifies particular data types and operation types.
141
# ru.ispras.microtesk.constraints.tests - contains JUnit test cases.
142
143
h3. Core classes/interfaces
144
145
*Syntax Tree Implementation*
146
147
The syntax tree nodes are implemented in the following classes:
148
* Constraint. Parameterized by a collection of Variable objects and a collection of Formula objects.
149
* Formula. Parameterized by an Operation object.
150
* Operation. Implements SyntaxElement. Parameterized by operand objects implementing SyntaxElement and an operation type object implementing OperationType.
151
* Variable. Implements SyntaxElement. Parameterized by the variable name string, a data type object implemeting DataType and a BigInteger value object.   
152
* Value. Implements SyntaxElement. Parameterized a data type object implemeting DataType and a BigInteger value object.
153
154
The SyntaxElement interface provides the ability to combine different kinds of elements into expressions.
155
156 10 Alexander Kamkin
The current implementation supports operations with the following data types: (1) Bit vectors, (2) Booleans. They are implemented in the BitVector and LogicBoolean classes. The BitVectorOperation and LogicBooleanOperation classes specify supported operation with these types. For example, the LogicBooleanOperation class looks like this:
157 1 Andrei Tatarnikov
158
<pre><code class="java">
159
public final class LogicBooleanOperation extends OperationType
160
{
161
	private LogicBooleanOperation() {}
162
	
163
	/** Operation: Logic - Equality */
164
	public static final OperationType EQ = new LogicBooleanOperation();
165
	/** Operation: Logic - AND */
166
	public static final OperationType AND = new LogicBooleanOperation();
167
	/** Operation: Logic - OR */
168
	public static final OperationType OR  = new LogicBooleanOperation();
169
	/** Operation: Logic - NOT */
170
	public static final OperationType NOT = new LogicBooleanOperation();
171
	/** Operation: Logic - XOR */
172
	public static final OperationType XOR = new LogicBooleanOperation();
173
	/** Operation: Logic - Implication */
174 10 Alexander Kamkin
	public static final OperationType IMPL= new LogicBooleanOperation();
175 1 Andrei Tatarnikov
} 
176
</code></pre>
177
178 10 Alexander Kamkin
The code below demonstrates how we can build a syntax tree representation for the integer overflow constraint:
179 1 Andrei Tatarnikov
180
<pre><code class="java">
181
class BitVectorIntegerOverflowTestCase implements SolverTestCase
182
{
183
	private static final int      BIT_VECTOR_LENGTH = 64;
184
	private static final DataType   BIT_VECTOR_TYPE = DataType.getBitVector(BIT_VECTOR_LENGTH);
185
	private static final Value             INT_ZERO = new Value(new BigInteger("0"), BIT_VECTOR_TYPE);
186
	private static final Value        INT_BASE_SIZE = new Value(new BigInteger("32"), BIT_VECTOR_TYPE);
187
188
	private static final Operation    INT_SIGN_MASK =
189
		new Operation(BitVectorOperation.BVSHL, new Operation(BitVectorOperation.BVNOT, INT_ZERO, null), INT_BASE_SIZE);
190
	
191
	private Operation IsValidPos(SyntaxElement arg)
192
	{
193
		return new Operation(LogicBooleanOperation.EQ, new Operation(BitVectorOperation.BVAND, arg, INT_SIGN_MASK), INT_ZERO);
194
	}
195
	
196
	private Operation IsValidNeg(SyntaxElement arg)
197
	{
198
		return new Operation(LogicBooleanOperation.EQ, new Operation(BitVectorOperation.BVAND, arg, INT_SIGN_MASK), INT_SIGN_MASK);
199
	}
200
	
201
	private Operation IsValidSignedInt(SyntaxElement arg)
202
	{
203
		return new Operation(LogicBooleanOperation.OR, IsValidPos(arg), IsValidNeg(arg));
204
	}
205
	
206
	public Constraint getConstraint()
207
	{
208
		Constraint constraint = new Constraint();
209
		
210
		Variable rs = new Variable("rs", BIT_VECTOR_TYPE, null);
211
		constraint.addVariable(rs);
212
		
213
		Variable rt = new Variable("rt", BIT_VECTOR_TYPE, null);
214
		constraint.addVariable(rt);
215
		
216
		
217
		constraint.addFormula(
218
			new Formula(
219
				IsValidSignedInt(rs)
220
			)
221
		);
222
		
223
		constraint.addFormula(
224
			new Formula(
225
				IsValidSignedInt(rt)
226
			)
227
		);
228
229
		constraint.addFormula(
230
			new Formula(
231
				new Operation(
232
					LogicBooleanOperation.NOT,
233
					IsValidSignedInt(new Operation(BitVectorOperation.BVADD, rs, rt)),
234
					null
235
				) 
236
			)
237
		);
238
239
		constraint.addFormula(
240
			new Formula(
241
				new Operation(LogicBooleanOperation.NOT, new Operation(LogicBooleanOperation.EQ, rs, rt), null)
242
			)
243
		);
244
245
		return constraint;
246
	}
247
	
248
	public Vector<Variable> getExpectedVariables()	
249
	{
250
		Vector<Variable> result = new Vector<Variable>();
251
		
252
		result.add(new Variable("rs", BIT_VECTOR_TYPE, new BigInteger("000000009b91b193", 16)));
253
		result.add(new Variable("rt", BIT_VECTOR_TYPE, new BigInteger("000000009b91b1b3", 16)));
254
		
255
		return result;	
256 10 Alexander Kamkin
	}
257 1 Andrei Tatarnikov
}
258
</code></pre>
259
260
*Representation Translation*
261
262 10 Alexander Kamkin
The logic that translates a tree representation into an SMT representation is implemented in the following way: Methods of the Translator class traverse the constraint syntax tree and use methods of the RepresentationBuilder interface to translate information about its nodes into a representation that can be understood by a particular solver. The RepresentationBuilder interface looks like follows:
263 1 Andrei Tatarnikov
264
<pre><code class="java">
265
public interface RepresentationBuilder
266
{	
267
	public void addVariableDeclaration(Variable variable);
268
269
	public void beginConstraint();
270
	public void endConstraint();
271
272
	public void beginFormula();
273
	public void endFormula();
274
275
	public void beginExpression();
276
	public void endExpression();
277
278
	public void appendValue(Value value);
279
	public void appendVariable(Variable variable);
280 10 Alexander Kamkin
	public void appendOperation(OperationType type);
281 1 Andrei Tatarnikov
}
282
</code></pre>
283
284
*Solver Implementation*
285
286 10 Alexander Kamkin
Solvers use the Translator class and a specific implementation of the RepresentationBuilder interface to generate an SMT representation of a constraint. Then they run a solver engine to solve the constraint and produce the results. Solver implement a common interface called Solver that looks like this:
287 1 Andrei Tatarnikov
288
<pre><code class="java">
289
public interface Solver
290
{
291
	public boolean solveConstraint(Constraint constraint);
292
	
293
	public boolean isSolved();
294
	public boolean isSatisfiable();
295
	
296
	public int getErrorCount();
297
	public String getErrorText(int index);
298
	
299
	public int getVariableCount();
300 10 Alexander Kamkin
	public Variable getVariable(int index);
301 1 Andrei Tatarnikov
}
302
</code></pre>