The writeback path has a lot of code that's been dead for a while. We're never asked to release dirty and thus delayed allocated or unwritten pages from ->releasepage, and since the introduction of ->page_mkwrite we also don't have to deal with dirty but unmapped buffers in ->writepage. But there's a case masquerading as such, which needs to be fixed by a 1.5 years old patch from Eric first. That patch also happens to fix xfstests 194. _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
