[PATCH 0/4] xfs: candidate fixes for 3.12-rc4

[Dave Chinner <david@xxxxxxxxxxxxx>]

Hi folks,

The first two patches are regressions and hence -rc4 candidates -
the first patch fixes a lockdep false positive as a result of adding
the third quota inode, and the second fixes a on-disk format
interpretation problem in the v4 superblock dtype-in-dirent feature.

The third patch fixes the AGF vs AGI lock inversion problem I
reported a short while ago, and the last reduces the overhead of
atomic version change accounting. These two aren't critical fixes,
but would be nice to have out there for 3.12.

Cheers,

Dave.