The parallel quota check is still in progress because I observed performance
varies greatly, I suspect that is related to the hardward problem of my laptop.
Thus, I have to delay that for a period of time until I can get a reasonable
Meantime, there are some patches could be posted for the review which are focus
code refactoring and consolidations in xfs_ioc_bulkstat() related functions,
xfs_bulkstat(), xfs_inumbers(), etc...
It mainly implemented per AG bulkstat that we have discussed in parallel quota
thread, and fix several problems regarding to the error handling for inode btree
operations in xfs_bulkstat() and xfs_inumbers() as per Christoph's suggestion.
Moreover, I introduced another per AG inumber strategy, which can be used to
multiple AGs with a new ioctl(2) in the future.