To work around this, you can use the <<worker_api.adoc#tasks_parallel_worker, Worker API>> with classloader or process isolation to encapsulate the library code.
The bytecode of the worker's classpath is not modified, so the self-checks should pass.
When process isolation is used, the worker action is executed in a separate worker process that doesn't have the Gradle Java agent installed.

			// to be very rare.
			return false
		}
		if v.Op.SymEffect()&SymAddr != 0 {
			// This case prevents an operation that calculates the
			// address of a local variable from being forced to schedule
			// before its corresponding VarDef.
			// See issue 28445.
			//   v1 = LOAD ...
			//   v2 = VARDEF
			//   v3 = LEAQ
			//   v4 = CMPQ v1 v3

     * has grown but the threshold has not yet been updated, so there are no atomicity requirements
     * for this with respect to reads.
     *
     * As a guide, all critical volatile reads and writes to the count field are marked in code
     * comments.
     */

    @Weak final MapMakerInternalMap<K, V, E, S> map;

    /**
     * The number of live elements in this segment's region. This does not include unset elements

// included by any test program that uses Google Test.
//
// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
// leave some internal implementation details in this header file.
// They are clearly marked by comments like this:
//
//   // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
//
// Such code is NOT meant to be used by a user directly, and is subject

// included by any test program that uses Google Test.
//
// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
// leave some internal implementation details in this header file.
// They are clearly marked by comments like this:
//
//   // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
//
// Such code is NOT meant to be used by a user directly, and is subject

type runCache struct {
	disableCache bool // cache should be disabled for this run

	buf *bytes.Buffer
	id1 cache.ActionID
	id2 cache.ActionID
}

// stdoutMu and lockedStdout provide a locked standard output
// that guarantees never to interlace writes from multiple
// goroutines, so that we can have multiple JSON streams writing
// to a lockedStdout simultaneously and know that events will

	// exactly one worker thread, especially syncClaim() is not reentrant.
	// Two syncClaims could bind two different claims to the same volume or one
	// claim to two volumes. The controller would recover from this (due to
	// version errors in API server and other checks in this controller),
	// however overall speed of multi-worker controller would be lower than if
	// it runs single thread only.

		// Drop all the cgo_import_static declarations.
		// Turns out we won't be needing them.
		for symIdx := range hostObjSyms {
			if l.SymType(symIdx) == sym.SHOSTOBJ {
				// If a symbol was marked both
				// cgo_import_static and cgo_import_dynamic,
				// then we want to make it cgo_import_dynamic
				// now.
				su := l.MakeSymbolUpdater(symIdx)

	// and only exported methods are accessible.
	//
	// For an interface type, the returned Method's Type field gives the
	// method signature, without a receiver, and the Func field is nil.
	//
	// Methods are sorted in lexicographic order.
	Method(int) Method

	// MethodByName returns the method with that name in the type's
	// method set and a boolean indicating if the method was found.
	//