> + /* > + * In most cases the direct IO writes will be done with IOLOCK_SHARED > + * allowing them to be done in parallel with reads and other direct IO > + * writes. However,if the IO is not aligned to filesystem blocks, the > + * direct IO layer needs to do sub-block zeroing and that requires > + * serialisation against other direct IOs to the same block. In this > + * case we need to serialise the submission of the unaligned IOs so > + * that we don't get racing block zeroing in the dio layer. > + * To avoid the problem with aio, we also need to wait for outstanding > + * IOs to complete so that unwritten extent conversion is completed > + * before we try to map the overlapping block. This is currently > + * implemented by hitting it with a big hammer (i.e. xfs_ioend_wait()). > + */ > + This comment is lacking one level of indentation, that's because the old kernel is missing the split of xfs_file_aio_write that reduces one level of indentation. Otherwise the backport looks fine, please send it to stable@xxxxxxxxxxxxxxx, with a [PATCH 0/0] coverletter stating it's for 2.6.35-stable. _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
