Using LVM snapshot integration with ext3 I've noted that even if my ext3 filesystem which was snapshotted was in a clean state at snapshot time (no needs_recovery flag set) when that filesystem is mounted read-only (via mounting the LVM snapshot) it appears to still try to modify the filesystem during the mount procedure as I get back some "Bad lvm_map in ll_rw_blk" messages from LVM. In all, I gather nothing is truely changed, but it does generate some debugging noise from the kernel/LVM. Looking at the code it looks like this could be related to an iput() of the journal or such possibly, as the code shows that the journal is the first thing addressed and has many checks regarding it.
Is this still an open problem?
The current release works without the mentioned problem. I'm using 0.0.6b now. I'm closing the bug.