Given the bug we recently had pop up on IRC about lost i_size updates when writing out inodes during umount I think it's time to stop relying on the writeback code to get these things right for us, and always log the inode size updates normally. Patches 1 and 2 are preparations that should go into the tree anyway, and patch 3 switches us to directly log the size updates from the I/O completion handler. _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
