From Bugzilla Helper: User-Agent: Mozilla/5.0 (compatible; Konqueror/3; Linux) Description of problem: after creating a block file (Following the loopback encryption howto), tying to issue the following command: losetup -e des /dev/loop3 /path/tofile gives ioctl: LOOP_SET_STATUS: Invalid Argument using -e xor instead of -e des lets it complete. From what I can gather, this means that the util-linux doesn't contain some patches that make the des encryption work. Version-Release number of selected component (if applicable): 2.11r-3 How reproducible: Always Steps to Reproduce: 1. losetup -e des /dev/loop3 /path/tofile (created by dd) 2. enter a password 3.enter a hex offset Actual Results: error: ioctl: LOOP_SET_STATUS: Invalid Argument Expected Results: the system should have mounted the file via the lookback device encrypted with DES. Additional info:
looks like the actual rpm is named losetup, but that isn't a component available to select.
*** Bug 70002 has been marked as a duplicate of this bug. ***
losetup comes from the util-linux source package. I'm not going to add anything in this area that's not in the upstream package, so you need to talk to the util-linux maintainer upstream (util-linux.no) - there are multiple patchsets to implement loopback encryption support, and none of them had been submitted directly to him last I heard.
According to traffic on the limbo-list, the entire crytoapi has been added to the rawhide kernel 2.4.18-7.85...wouldn't this require changes to the util-linux package (losetup) to put it all together?
Yes, but my earlier comments still stand (the changes will have to come from upstream).
I contacted Andries at the email address you provided, he responded quickly. I'm sure I don't understand the issues involoved, but I think having all of this working in Limbo/8.0 or even 8.1 would be a huge plus for Redhat. Here is the snippet: >> plans to integrate the cryptoapi patches? >Yes. >> when? >Long ago the obstacle was copyrights. People sent me >patches that could not be distributed because there >was copyrighted material. >Now there are at least two sets of crypto patches, >the "international crypto patch" and the patch >maintained by Jari Ruusu. >Don't know whether the former is considered dead. >Some time ago I tried to contact A. Kjeldaas in vain. >Andries