`Failed to instantiate` bug when a class with private constructor has value class constructor param

[edwardmp]

I stumbled upon an issue that puzzled me for a bit, but I managed to reproduce the conditions under which it happens.
In #1947 support for Kotlin value classes was added. However, I strongly believe an assumption was made in the code, that doesn't hold true in all cases, leading to bug I encounter.

When a data class (or any class) defines a private primary constructor with parameters that do not have default values, the Kotlin compiler does not generate the synthetic constructor with the (int mask, DefaultConstructorMarker) signature. Instead, it only emits the (DefaultConstructorMarker) variant.

This leads to a runtime exception like the following:

org.springframework.data.mapping.model.MappingInstantiationException: Failed to instantiate nl.edward.backend.domain.order.Order using constructor fun `<init>`(nl.edward.backend.domain.shared.Identifier<nl.edward.backend.domain.order.Order>, nl.edward.backend.domain.shared.Identifier<nl.edward.backend.domain.buyer.Buyer>, kotlin.collections.Set<nl.edward.backend.domain.order.Seller>, nl.edward.backend.domain.asset.Asset, java.math.BigDecimal, kotlin.String?, kotlin.String?, java.math.BigDecimal, kotlin.Boolean, java.time.LocalDate, kotlin.Boolean, kotlin.Long): nl.edward.backend.domain.order.Order with arguments 35a8e2d0-3013-4740-8a2a-3436afb267f2,dd4f5b94-d0e3-46ec-890a-94e5c0632306,[Seller(organizationId=59cd1422-74bc-48fb-a4ad-b85f2a37a5d4, quantity=50, unitPrice=10.00, purchaseInvoicePaidAt=null, purchaseInvoiceIsCreated=false, isDelivered=false)],HBE2025NL,15.00,null,TEST-REF-INTEGRATION,21.00,false,2025-10-27,false,1
	at org.springframework.data.mapping.model.ReflectionEntityInstantiator.createInstance(ReflectionEntityInstantiator.java:98) ~[spring-data-commons-3.5.4.jar:3.5.4]
	at org.springframework.data.mapping.model.ClassGeneratingEntityInstantiator.createInstance(ClassGeneratingEntityInstantiator.java:98) ~[spring-data-commons-3.5.4.jar:3.5.4]
	at org.springframework.data.relational.core.conversion.MappingRelationalConverter.read(MappingRelationalConverter.java:471) ~[spring-data-relational-3.5.4.jar:3.5.4]
	at org.springframework.data.relational.core.conversion.MappingRelationalConverter.readAggregate(MappingRelationalConverter.java:366) ~[spring-data-relational-3.5.4.jar:3.5.4]
	at org.springframework.data.relational.core.conversion.MappingRelationalConverter.readAggregate(MappingRelationalConverter.java:329) ~[spring-data-relational-3.5.4.jar:3.5.4]
	at org.springframework.data.jdbc.core.convert.MappingJdbcConverter.readAndResolve(MappingJdbcConverter.java:304) ~[spring-data-jdbc-3.5.4.jar:3.5.4]
	at org.springframework.data.jdbc.core.convert.EntityRowMapper.mapRow(EntityRowMapper.java:65) ~[spring-data-jdbc-3.5.4.jar:3.5.4]
	at org.springframework.jdbc.core.RowMapperResultSetExtractor.extractData(RowMapperResultSetExtractor.java:94) ~[spring-jdbc-6.2.11.jar:6.2.11]

Caused by: org.springframework.beans.BeanInstantiationException: Failed to instantiate [nl.edward.backend.domain.order.Order]: Illegal arguments for constructor
	at org.springframework.beans.BeanUtils.instantiateClass(BeanUtils.java:220) ~[spring-beans-6.2.11.jar:6.2.11]
	at org.springframework.data.mapping.model.ReflectionEntityInstantiator.createInstance(ReflectionEntityInstantiator.java:96) ~[spring-data-commons-3.5.4.jar:3.5.4]
	... 302 common frames omitted

Caused by: java.lang.IllegalArgumentException: object is not an instance of declaring class
	at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.checkReceiver(DirectMethodHandleAccessor.java:197) ~[na:na]
	at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:99) ~[na:na]
	at java.base/java.lang.reflect.Method.invoke(Method.java:580) ~[na:na]
	at kotlin.reflect.jvm.internal.calls.ValueClassAwareCaller.call(ValueClassAwareCaller.kt:200) ~[kotlin-reflect-2.2.20.jar:2.2.20-release-333]
	at kotlin.reflect.jvm.internal.KCallableImpl.callDefaultMethod$kotlin_reflection(KCallableImpl.kt:250) ~[kotlin-reflect-2.2.20.jar:2.2.20-release-333]
	at kotlin.reflect.jvm.internal.KCallableImpl.callBy(KCallableImpl.kt:155) ~[kotlin-reflect-2.2.20.jar:2.2.20-release-333]
	at org.springframework.beans.BeanUtils$KotlinDelegate.instantiateClass(BeanUtils.java:943) ~[spring-beans-6.2.11.jar:6.2.11]
	at org.springframework.beans.BeanUtils.instantiateClass(BeanUtils.java:191) ~[spring-beans-6.2.11.jar:6.2.11]
	... 303 common frames omitted

I'm not enough of a Kotlin/reflection expert to pin-point the fix, however I believe it has something to do with this piece of code found in KotlinInstantiationDelegate

if (!KotlinInstantiationDelegate.hasDefaultConstructorMarker(detectedConstructorParameters)) {

	// candidates must contain at least two additional parameters (int, DefaultConstructorMarker).
	// Number of defaulting masks derives from the original constructor arg count
	int syntheticParameters = KotlinDefaultMask.getMaskCount(detectedConstructor.getParameterCount())
			+ /* DefaultConstructorMarker */ 1;

	if ((detectedConstructor.getParameterCount() + syntheticParameters) != candidate.getParameterCount()) {
		continue;
	}
}

Notably, the code assumes there are always at least two synthetic parameters in this case, one for the default constructor marker (which is correct), and one for the mask (KotlinDefaultMask.getMaskCount always returns at least 1 which is apparently incorrect in the case of a private default constructor with no default params).
I guess one way or another we might need to detect the latter cases and prevent assuming an extra synthetic parameter that does not exist in reality.
For what it's worth, modifying the value of syntheticParameters at runtime with the debugger to be one less, solves the issue.

Shortly, I will link a pull request (#3390) showcasing two additional test cases that show the issue. One of the test cases displays the issue, the other shows that the issue does not happen under the exact same circumstances other than a default constructor param value being added.

[edwardmp]

@mp911de as the original author you might have insight into this

[mp911de]

Thanks for the test cases. It makes sense to revisit constructor detection and make private constructors work. Part of the problem is that Kotlin applies some magic on the JVM to make things work requiring us to reverse engineer what is actually happening.

Going forward, I suggest to capture the code here in the comments as PR's are actually intended to bring fixes or enhancements to our code base.

[edwardmp]

Thanks @mp911de for the quick reply ❤️
And I totally understand the complexities around the magic value classes unfortunately seems to bring, I literally created a PR for mockk the other day that also had such a bug around value classes specifically.

The idea behind the PR is that eventually, this should be merged anyway to validate any fix.
I did try to create a fix, but assuming that the second to last param is a mask simply when it's an int doesn't feel safe. Any points how one might detect that a param is a mask so I can perhaps try another stab at solving it?

[mp911de]

I don’t know. You could ask on the public Kotlin slack for help. I will hate myself for this, but that is one of the rare occasions I would ask AI for help to see what it comes around with. Am 23. Okt. 2025, 18:02 +0200 schrieb Edward Poot ***@***.***>:

…

edwardmp left a comment (spring-projects/spring-data-commons#3389) Thanks @mp911de for the quick reply ❤️ And I totally understand the complexities around the magic value classes unfortunately seems to bring, I literally created a PR for mockk the other day that also had such a bug around value classes specifically. The idea behind the PR is that eventually, this should be merged anyway to validate any fix. I did try to create a fix, but assuming that the second to last param is a mask simply when it's an int doesn't feel safe. Any points how one might detect that a param is a mask so I can perhaps try another stab at solving it? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>