On Wed, Dec 01, 2010 at 07:45:40AM -0500, Christoph Hellwig wrote: > Once l_tail_lsn and l_last_sync_lsn are converted to atomic64_ts later > in the series passing them in separately becomes pointless again. Maybe > scale the patch back a bit and require the caller to hold l_grant_lock > for now, until the atomic64_t conversion happens. Same reasoning as for the change to xlog_space_left - the values should only be sampled once for correctness, which is why I pass them into the function. more "const" annotations, I guess. Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
