I have an x86_64 kernel with i386 userspace. e4defrag fails on the EXT4_IOC_MOVE_EXT ioctl because it is not wired up for the compat case. It seems that struct move_extent is compat save, only types with fixed widths are used: { __u32 reserved; /* should be zero */ __u32 donor_fd; /* donor file descriptor */ __u64 orig_start; /* logical start offset in block for orig */ __u64 donor_start; /* logical start offset in block for donor */ ...

>>>>> "Eric" == Eric Sandeen <sandeen@redhat.com> writes: Eric> It is expected, as lvm is not yet passing discard requests. I'm working on this. Missed the 2.6.34 merge window so I'm aiming at 2.6.35. -- Martin K. Petersen Oracle Linux Engineering --

Check and see if it's still a problem with what I ultimately ended up pushing to Linus. Yes, the way I chose to fix things means that we'll never do more than 2**32-1 blocks at a time with fiemap, but in practice there's a limit to the number of extents that can be returned in a single fiemap call anyway, so in practice this shouldn't be a problem. - Ted --

I think it would make for a lot of code churn without any rel value. We have lots of lower-case wrappers around functions for things like this already (e.g. ext4_journal_start(), etc. Cheers, Andreas -- Andreas Dilger Sr. Staff Engineer, Lustre Group Sun Microsystems of Canada, Inc. --