Karsten Loebbicke wrote: > maybe your algorithm (twofish) produce this failure? try to create an > aes-encryptet partition... The files that keep giving different checksums were originally created on a twofish dm-crypted ext3 partition. No problems whatsoever there. I also tested them on my twofish dm-crypted laptop homedir, no problems either. Both these systems run the same OS and the same kernel as the faulty system. Despite the fact that /root works fine and lsat does not, there could still be some really weird hardware-related thing behind this and I want to rule that out 100%. Normally I'd just rip out the hard disk and put it in another machine. Right now I don't have another machine where I am, so I need to find a computer repair shop that will let me do the test. If it misbehaves there too, the thing to look at would be the difference between the systems. The systems that work have smaller partitions, one is 30G and the other 70G. The misbehaving partition is 270G, so things like block size and the number of blocks and inodes could theoretically play a role in this. Sadly I know nothing about dm-crypt internals - or ext3 internals for that matter - so I don't even know where to start speculating. Does anyone around here have experience with huge encrypted partitions? Does the partition size affect dm-crypt's memory usage more than marginally? Z --------------------------------------------------------------------- dm-crypt mailing list - http://www.saout.de/misc/dm-crypt/ To unsubscribe, e-mail: dm-crypt-unsubscribe@xxxxxxxx For additional commands, e-mail: dm-crypt-help@xxxxxxxx
