# HG changeset patch
# User Christian Humer
+ * A parameter annotated with {@link Cached} in a {@link Specialization} refers to a cached
+ * value of a specialization instance. A cached parameter value is initialized once using the
+ * initializer expression at specialization instantiation. For each call of the specialization
+ * method the cached value is provided by using the annotated parameter from the method body. Cache
+ * initializers are potentially executed before guard expressions declared in
+ * {@link Specialization#guards()}.
+ *
+ * A typical specialization may define multiple dynamic and multiple cached parameters. Dynamic
+ * parameter values are typically provided by executing child nodes of the operation. Cached
+ * parameters are initialized and stored once per specialization instantiation. Cached parameters
+ * are always constant at compile time. You may verify this by invoking
+ * {@link CompilerAsserts#compilationConstant(Object)} on any cached parameter. For consistency
+ * between specialization declarations cached parameters must be declared last in a specialization
+ * method.
+ *
+ * The initializer expression of a cached parameter is defined using a subset of Java. This subset
+ * includes field/parameter accesses, function calls, type exact infix comparisons (==, !=, <, <=,
+ * >, >=) and integer literals. The return type of the initializer expression must be assignable to
+ * the parameter type. If the annotated parameter type is derived from {@link Node} then the
+ * {@link Node} instance is allowed to use the {@link Node#replace(Node)} method to replace itself.
+ * Bound elements without receivers are resolved using the following order:
+ *
+ *
+ *
+ * The following examples explain the intended use of the {@link Cached} annotation. All of the
+ * examples have to be enclosed in the following node declaration:
+ * new
keyword
+ * as method name.
+ * @NodeChild("operand") + * abstract TestNode extends Node { + * abstract void execute(Object operandValue); + * // ... example here ... + * } + *+ * + *
+ * @Specialization + * void doCached(int operand, @Local("operand") int cachedOperand) { + * CompilerAsserts.compilationConstant(cachedOperand); + * ... + * } + * + * Example executions: + * execute(1) => doCached(1, 1) // new instantiation, localOperand is bound to 1 + * execute(0) => doCached(0, 1) + * execute(2) => doCached(2, 1) + * + *+ * + *
int
values and for each
+ * individual int
value a new specialization would get instantiated. The
+ * {@link Specialization#limit()} property defines a limit for the number of specializations that
+ * can get instantiated. If the specialization instantiation limit is reached then no further
+ * specializations are instantiated. Like for other specializations if there are no more
+ * specializations defined an {@link UnsupportedSpecializationException} is thrown. The default
+ * specialization instantiation limit is 3
.
+ *
+ * + * @Specialization(guards = "==(operand, cachedOperand)") + * void doCached(int operand, @Cached("operand") int cachedOperand) { + * CompilerAsserts.compilationConstant(cachedOperand); + * ... + * } + * + * Example executions: + * execute(0) => doCached(0, 0) // new instantiation, cachedOperand is bound to 0 + * execute(1) => doCached(1, 1) // new instantiation, cachedOperand is bound to 1 + * execute(1) => doCached(1, 1) + * execute(2) => doCached(2, 2) // new instantiation, cachedOperand is bound to 2 + * execute(3) => throws UnsupportedSpecializationException // instantiation limit overflows + * + *+ * + *
doNormal
. This specialization has the same type restrictions but does not have local
+ * state nor the operand identity guard. It is also declared after doCached
therefore
+ * it is only instantiated if the limit of the doCached
specialization has been
+ * reached. In other words doNormal
is more generic than doCached
. The
+ * doNormal
specialization uses contains="doCached"
to specify
+ * that all instantiations of doCached
get removed if doNormal
is
+ * instantiated. Alternatively if the contains
relation is omitted then all
+ * doCached
instances remain but no new instances are created.
+ *
+ * + * @Specialization(guards = "==(operand, cachedOperand)") + * void doCached(int operand, @Cached("operand") int cachedOperand) { + * CompilerAsserts.compilationConstant(cachedOperand); + * ... + * } + * + * @Specialization(contains = "doCached") + * void doNormal(int operand) {...} + * + * Example executions with contains = "doCached": + * execute(0) => doCached(0, 0) // new instantiation, cachedOperand is bound to 0 + * execute(1) => doCached(1, 1) // new instantiation, cachedOperand is bound to 1 + * execute(1) => doCached(1, 1) + * execute(2) => doCached(2, 2) // new instantiation, cachedOperand is bound to 2 + * execute(3) => doNormal(3) // new instantiation of doNormal due to limit overflow; doCached gets removed. + * execute(1) => doNormal(1) + * + * Example executions without contains = "doCached": + * execute(0) => doCached(0, 0) // new instantiation, cachedOperand is bound to 0 + * execute(1) => doCached(1, 1) // new instantiation, cachedOperand is bound to 1 + * execute(1) => doCached(1, 1) + * execute(2) => doCached(2, 2) // new instantiation, cachedOperand is bound to 2 + * execute(3) => doNormal(3) // new instantiation of doNormal due to limit overflow + * execute(1) => doCached(1, 1) + * + *+ * + *
private
.
+ *
+ * + * @Specialization + * void s(int operand, @Cached("transformLocal(operand)") int cachedOperand) { + * } + * + * int transformLocal(int operand) { + * return operand & 0x42; + * } + * + *
new
keyword can be used to initialize a cached parameter using a constructor of
+ * the parameter type.
+ *
+ * + * @Specialization + * void s(Object operand, @Cached("new()") OtherNode someNode) { + * someNode.execute(operand); + * } + * + * static class OtherNode extends Node { + * + * public String execute(Object value) { + * throw new UnsupportedOperationException(); + * } + * } + * + *+ * + *
+ * @Specialization + * void s(int operand, @Local("create()") BranchProfile profile) { + * } + *+ * + *
+ * Defines a method of a node subclass to represent one specialization of an operation. Multiple + * specializations can be defined in a node representing an operation. A specialization defines + * which kind of input is expected using the method signature and the annotation attributes. The + * specialized semantics of the operation are defined using the body of the annotated Java method. A + * specialization method must be declared in a class that is derived from {@link Node} that + * references a {@link TypeSystem}. At least one specialization must be defined per operation. If no + * specialization is valid for the given set of input values then an + * {@link UnsupportedSpecializationException} is thrown instead of invoking any specialization + * method. + *
+ *+ * A specialization must have at least as many parameters as there are {@link NodeChild} annotations + * declared for the enclosing operation node. These parameters are declared in the same order as the + * {@link NodeChild} annotations (linear execution order). We call such parameters dynamic input + * parameters. Every specialization that is declared within an operation must have an equal number + * of dynamic input parameters. + *
+ *+ * The supported kind of input values for a specialization are declared using guards. A + * specialization may provide declarative specifications for four kinds of guards: + *
true
. Expression guards are modeled using Java expressions that return a
+ * boolean
value. If the guard expression returns false
, the
+ * specialization is no longer applicable and the operation is re-specialized. Guard expressions are
+ * declared using the {@link #guards()} attribute.true
. Assumptions can be assigned to specializations using the
+ * {@link #assumptions()} attribute.
+ * The enclosing {@link Node} of a specialization method must have at least one public
+ * and non-final
execute method. An execute method is a method that starts with
+ * 'execute'. If all execute methods declare the first parameter type as {@link Frame},
+ * {@link VirtualFrame} or {@link MaterializedFrame} then the same frame type can be used as
+ * optional first parameter of the specialization. This parameter does not count to the number of
+ * dynamic parameters.
+ *
+ * A specialization method may declare multiple parameters annotated with {@link Cached}. Cached + * parameters are initialized and stored once per specialization instantiation. For consistency + * between specialization declarations cached parameters must be declared last in a specialization + * method. + *
+ *+ * If the operation is re-specialized or if it is executed for the first time then all declared + * specializations of the operation are tried in declaration order until the guards of the first + * specialization accepts the current input values. The new specialization is then added to the + * chain of current specialization instances which might consist of one (monomorph) or multiple + * instances (polymorph). If an assumption of an instantiated specialization is violated then + * re-specialization is triggered again. + *
+ *+ * With guards in combination with cached parameters it is possible that multiple instances of the + * same specialization are created. The {@link #limit()} attribute can be used to limit the number + * of instantiations per specialization. + *
+ * + * @see NodeChild + * @see ShortCircuit + * @see Fallback + * @see Cached + * @see TypeSystem + * @see TypeSystemReference + * @see UnsupportedSpecializationException + */ @Retention(RetentionPolicy.RUNTIME) @Target({ElementType.METHOD}) public @interface Specialization { @@ -36,39 +121,242 @@ @Deprecated int DEFAULT_ORDER = -1; /** - * The order has no effect anymore. The declaration order specialization methods is used - * instead. - * * @deprecated use declaration order instead. Will get removed in the next release. */ @Deprecated int order() default DEFAULT_ORDER; /** - * Inserts this and all specializations that are declared after this specialization before a - * specialization in the superclass. By default all specializations of the subclass are appended - * to the specializations of the superclass. + * References a specialization of a super class by its method name where this specialization is + * inserted before. The declaration order of a specialization is not usable for nodes where + * specializations are partly declared in the super class and partly declared in a derived + * class. By default all specializations declared in the derived class are appended to those in + * the super class. This attribute can be used to override the default behavior. */ String insertBefore() default ""; + /** + *+ * Declares an event guards that trigger re-specialization in case an exception is thrown in the + * specialization body. This attribute can be used to declare a list of such exceptions. Guards + * of this kind are useful to avoid calculating a value twice when it is used in the guard and + * its specialization. + *
+ * + *+ * If an event guard exception is triggered then all instantiations of this specialization are + * removed. If one of theses exceptions is thrown once then no further instantiations of this + * specialization are going to be created for this node. A specialization that rewrites on an + * exception must ensure that no non-repeatable side-effect is caused until the rewrite is + * triggered. + *
+ * + * Example usage: + * + *+ * @Specialization(rewriteOn = ArithmeticException.class) + * int doAddNoOverflow(int a, int b) { + * return ExactMath.addExact(a, b); + * } + * + * @Specialization + * long doAddWithOverflow(int a, int b) { + * return a + b; + * } + * ... + * + * Example executions: + * execute(Integer.MAX_VALUE - 1, 1) => doAddNoOverflow(Integer.MAX_VALUE - 1, 1) + * execute(Integer.MAX_VALUE, 1) => doAddNoOverflow(Integer.MAX_VALUE, 1) + * => throws ArithmeticException + * => doAddWithOverflow(Integer.MAX_VALUE, 1) + * execute(Integer.MAX_VALUE - 1, 1) => doAddWithOverflow(Integer.MAX_VALUE - 1, 1) + *+ * + * + * + * @see ExactMath#addExact(int, int) + */ Class extends Throwable>[] rewriteOn() default {}; /** - * The contains attribute declares all specializations that are contained by this - * specialization. A containing specialization must be strictly generic as the contained - * specialization. + *
+ * Declares other specializations of the same node to be contained by this specialization. Other + * specializations are references using their unique method name. If this specialization is + * instantiated then all contained specialization instances are removed and never instantiated + * again for this node instance. Therefore this specialization should handle strictly more + * inputs than which were handled by the contained specialization, otherwise the removal of the + * contained specialization will lead to unspecialized types of input values. The contains + * declaration is transitive for multiple involved specializations. + *
+ * Example usage: + * + *+ * @Specialization(guards = "b == 2") + * void doDivPowerTwo(int a, int b) { + * return a >> 1; + * } + * + * @Specialization(contains ="doDivPowerTwo", guards = "b > 0") + * void doDivPositive(int a, int b) { + * return a / b; + * } + * ... + * + * Example executions with contains="doDivPowerTwo": + * execute(4, 2) => doDivPowerTwo(4, 2) + * execute(9, 3) => doDivPositive(9, 3) // doDivPowerTwo instances get removed + * execute(4, 2) => doDivPositive(4, 2) + * + * Same executions without contains="doDivPowerTwo" + * execute(4, 2) => doDivPowerTwo(4, 2) + * execute(9, 3) => doDivPositive(9, 3) + * execute(4, 2) => doDivPowerTwo(4, 2) + *+ * + * + * + * @see #guards() */ String[] contains() default {}; + /** + *
+ * Declares boolean
expressions that define whether or not input values are
+ * applicable to this specialization instance. Guard expressions must always return the same
+ * result for each combination of the enclosing node instance and the bound input values.
+ *
+ * If a guard expression does not bind any dynamic input parameters then the DSL assumes that + * the result will not change for this node after specialization instantiation. The DSL asserts + * this assumption if assertions are enabled (-ea). + *
+ *
+ * Guard expressions are defined using a subset of Java. This subset includes field/parameter
+ * accesses, function calls, type exact infix comparisons (==, !=, <, <=, >, >=), logical
+ * negation (!), logical disjunction (||) and integer literals. The return type of guard
+ * expressions must be boolean
. Bound elements without receivers are resolved using
+ * the following order:
+ *
+ * Example usage: + * + *
+ * static boolean acceptOperand(int operand) { + * assert operand <= 42; + * return operand & 1 == 1; + * } + * + * @Specialization(guards = {"operand <= 42", "acceptOperand(operand)"}) + * void doSpecialization(int operand) {...} + *+ * + * + * + * @see Cached + * @see ImportStatic + */ String[] guards() default {}; /** - * Defines the assumptions to check for this specialization. When the specialization method is - * invoked it is guaranteed that these assumptions still hold. It is not guaranteed that they - * are checked before the {@link #guards()} methods. They may be checked before after or in - * between {@link #guards()}. To declare assumptions use the {@link NodeAssumptions} annotation - * at class level. + *
+ * Declares assumption guards that optimistically assume that the state of an {@link Assumption}
+ * remains valid. Assumption expressions are cached once per specialization instantiation. If
+ * one of the returned assumptions gets invalidated then the specialization instance is removed.
+ * An assumption expression may return different assumptions per specialization instance. The
+ * returned assumption instance must not be null
.
+ *
+ * Assumption expressions are defined using a subset of Java. This subset includes + * field/parameter accesses, function calls, type exact infix comparisons (==, !=, <, <=, >, + * >=), logical negation (!), logical disjunction (||) and integer literals. The return type of + * the expression must be {@link Assumption}. Assumption expressions are not allowed to bind to + * dynamic parameter values of the specialization. Bound elements without receivers are resolved + * using the following order: + *
+ * Example usage: + * + *
+ * static abstract class DynamicObject() { + * abstract Shape getShape(); + * ... + * } + * static abstract class Shape() { + * abstract Assumption getUnmodifiedAssuption(); + * ... + * } + * + * @Specialization(guards = "operand.getShape() == cachedShape", assumptions = "cachedShape.getUnmodifiedAssumption()") + * void doAssumeUnmodifiedShape(DynamicObject operand, @Cached("operand.getShape()") Shape cachedShape) {...} + *+ * + * + * + * @see Cached + * @see ImportStatic */ String[] assumptions() default {}; + /** + *
+ * Declares the expression that limits the number of specialization instantiations. The default
+ * limit for specialization instantiations is defined as "3"
. If the limit is
+ * exceeded no more instantiations of the enclosing specialization method are created. Please
+ * note that the existing specialization instantiations are not removed from the
+ * specialization chain. You can use {@link #contains()} to remove unnecessary specializations
+ * instances.
+ *
+ * The limit expression is defined using a subset of Java. This subset includes field/parameter
+ * accesses, function calls, type exact infix comparisons (==, !=, <, <=, >, >=), logical
+ * negation (!), logical disjunction (||) and integer literals. The return type of the limit
+ * expression must be int
. Limit expressions are not allowed to bind to dynamic
+ * parameter values of the specialization. Bound elements without receivers are resolved using
+ * the following order:
+ *
+ * Example usage: + * + *
+ * static int getCacheLimit() { + * return Integer.parseInt(System.getProperty("language.cacheLimit")); + * } + * + * @Specialization(guards = "operand == cachedOperand", limit = "getCacheLimit()") + * void doCached(Object operand, @Cached("operand") Object cachedOperand) {...} + *+ * + * + * + * @see #guards() + * @see #contains() + * @see Cached + * @see ImportStatic + */ + String limit() default ""; + }