With more reports showing up that the VFS writeback code is not able to write back our size updates dirtied from the I/O completion handler in reasonable time I think it's time to move to logging all file size updates ASAP, that is for 3.2 and maybe after a reasonable testing period even -stable. This series has been sent out a few times, and I've been doing QA on it for weeks. Note that I haven't implemented the log space reservations from ->writepage - see the actual patch for the rationale. Also throw in the fork offset fix as it's an user-triggerable oops. _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
