NationalSecurityAgency/ghidra · error · MemoryAccessException

Couldn't get requested bytes for CodeUnit

Error message

Couldn't get requested bytes for CodeUnit

What it means

Thrown by DBTraceCodeUnitAdapter.getBytesInCodeUnit() when the underlying getBytes() call returns fewer bytes than the requested length. This indicates the memory backing the code unit's address range does not have enough initialized/available bytes to fill the buffer. It wraps a partial read as a MemoryAccessException so callers know the code unit's byte content is incomplete or unreadable at that snapshot.

Source

Thrown at Ghidra/Debug/Framework-TraceModeling/src/main/java/ghidra/trace/database/listing/DBTraceCodeUnitAdapter.java:403

	default Memory getMemory() {
		return getProgram().getMemory();
	}

	@Override
	default boolean isBigEndian() {
		return getLanguage().isBigEndian();
	}

	@Override
	default byte[] getBytes() throws MemoryAccessException {
		return getBytesInFull(0, getLength()).array();
	}

	@Override
	default void getBytesInCodeUnit(byte[] buffer, int bufferOffset) throws MemoryAccessException {
		int len = Math.min(buffer.length - bufferOffset, getLength());
		if (getBytes(ByteBuffer.wrap(buffer, bufferOffset, len), 0) != len) {
			throw new MemoryAccessException("Couldn't get requested bytes for CodeUnit");
		}
	}
}

View on GitHub (pinned to d5f144c24d)

Solutions

1. Ensure memory bytes are recorded for the address range and snap before creating/reading the code unit — use trace.getMemoryManager().getBytes().
2. Check codeUnit existence and memory coverage with memSpace.contains() or getBytes() return value before calling getBytesInCodeUnit.
3. Catch MemoryAccessException and handle gracefully (treat as uninitialized) if your workflow tolerates missing bytes.
4. Switch to a snapshot where the memory is known to be fully recorded.

Example fix

// before
codeUnit.getBytesInCodeUnit(buf, 0); // may throw

// after — verify bytes exist first
int readable = memSpace.getBytes(startSnap, codeUnit.getAddress(), buf);
if (readable < codeUnit.getLength()) {
    // handle missing bytes
} else {
    codeUnit.getBytesInCodeUnit(buf, 0);
}

// Check memory coverage before reading code unit bytes
ByteBuffer buf = ByteBuffer.allocate(codeUnit.getLength());
int readable = memSpace.getBytes(startSnap, codeUnit.getAddress(), buf);
if (readable < codeUnit.getLength()) {
    // Memory not fully recorded — skip or handle
    return;
}
codeUnit.getBytesInCodeUnit(buf.array(), 0);

try {
    codeUnit.getBytesInCodeUnit(buffer, offset);
} catch (MemoryAccessException e) {
    // Bytes not available at this snap — handle gracefully
    Msg.warn(MyClass.class, "Could not read bytes for " + codeUnit.getAddress());
    Arrays.fill(buffer, offset, offset + codeUnit.getLength(), (byte) 0);
}

When it happens

Trigger: Calling codeUnit.getBytesInCodeUnit(buffer, offset) when the trace memory at the code unit's address range does not contain enough bytes — e.g., the memory was never written/recorded for that snap, or only partially recorded. The getBytes(ByteBuffer, 0) return value is less than len.

Common situations: Inspecting a code unit at a snapshot where memory was not recorded (e.g., before the trace started recording, or a gap in recorded snaps); memory mapped to an overlay that has no backing bytes; truncated or corrupted trace files; attempting to read bytes for a guest-platform code unit where the host memory is uninitialized.

Related errors

• Code units cannot overlap
• Code units cannot overlap
• Length override of {newLength} conflicts with code unit at {
• Byte arrays cannot exceed length of variable
• Byte array values must be hex enclosed in {}

AI-assisted analysis of NationalSecurityAgency/ghidra@d5f144c24d (2026-08-14). Data as JSON: /api/errors/2833b52b042fa9e5. Report an issue: GitHub.