The vulnerability stems from incorrect buffer size calculation in name.encodingLength, which was patched in commits 25f15dd and 0d0d593. The original implementation used Buffer.byteLength(n) without sanitizing input domain names, causing buffers allocated with allocUnsafe to have uninitialized regions. The patches explicitly trim domain names before size calculation, confirming this function's role in the vulnerability. The function's output directly influences buffer allocation, making it the root cause of memory exposure.