In the blk driver, the process goes to sleep while waiting for the blk
driver to finish. However, it is possible for the kernel to attempt to
wake up the process before it actually goes to sleep, leading the the
process getting stuck in the sleep state. When there are multiple
processors running, this bug regularly manifests here

  while(owned_buf->disk == 1) {
    sleep((void *)owned_buf, &blk_lock);
  }

To fix this, the kernel will now acquire the blk_lock when handling
a blk interrupt.