org.graalvm.compiler.replacements.amd64

Class AMD64ConvertSnippets

java.lang.Object

org.graalvm.compiler.replacements.amd64.AMD64ConvertSnippets

All Implemented Interfaces:

Snippets

public class AMD64ConvertSnippets
extends Object
implements Snippets

Snippets used for conversion operations on AMD64 where the AMD64 instruction used does not match the semantics of the JVM specification.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	AMD64ConvertSnippets.Templates

Constructor Summary

Constructors

Constructor and Description
AMD64ConvertSnippets()

Method Summary

Modifier and Type	Method and Description
static int	d2i(double input, int result) Converts a double to an int.
static long	d2l(double input, long result) Converts a double to a long.
static int	f2i(float input, int result) Converts a float to an int.
static long	f2l(float input, long result) Converts a float to a long.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AMD64ConvertSnippets

public AMD64ConvertSnippets()

Method Detail

f2i

public static int f2i(float input,
                      int result)

Converts a float to an int.

This snippet accounts for the semantics of the x64 CVTTSS2SI instruction used to do the conversion. If the float value is a NaN, infinity or if the result of the conversion is larger than Integer.MAX_VALUE then CVTTSS2SI returns Integer.MIN_VALUE and extra tests are required on the float value to return the correct int value.

Parameters:

input - the float being converted

result - the result produced by the CVTTSS2SI instruction

f2l

public static long f2l(float input,
                       long result)

Converts a float to a long.

This snippet accounts for the semantics of the x64 CVTTSS2SI instruction used to do the conversion. If the float value is a NaN or infinity then CVTTSS2SI returns Long.MIN_VALUE and extra tests are required on the float value to return the correct long value.

Parameters:

input - the float being converted

result - the result produced by the CVTTSS2SI instruction

d2i

public static int d2i(double input,
                      int result)

Converts a double to an int.

This snippet accounts for the semantics of the x64 CVTTSD2SI instruction used to do the conversion. If the double value is a NaN, infinity or if the result of the conversion is larger than Integer.MAX_VALUE then CVTTSD2SI returns Integer.MIN_VALUE and extra tests are required on the double value to return the correct int value.

Parameters:

input - the double being converted

result - the result produced by the CVTTSS2SI instruction

d2l

public static long d2l(double input,
                       long result)

Converts a double to a long.

This snippet accounts for the semantics of the x64 CVTTSD2SI instruction used to do the conversion. If the double value is a NaN, infinity or if the result of the conversion is larger than Long.MAX_VALUE then CVTTSD2SI returns Long.MIN_VALUE and extra tests are required on the double value to return the correct long value.

Parameters:

input - the double being converted

result - the result produced by the CVTTSS2SI instruction