My backup system is built entirely around the cryptoloop module, and I have no real interest in re-engineering it. In my application, the weakness referred to in CVE-2004-2135 is largely irrelevant. I'm not sure what "cryptoloop doesn't work with journaling filesystems" is supposed to mean; it has seemed to work fine for me for years and years.
Please put cryptoloop back. I don't have a week to mess with dmcrypt right now, so I'm just going to have to compile my own kernel (or at least my own cryptoloop module) if you don't. I would be really, really annoyed by this. I can't believe I'm the only one.
