Heh, you beat me to it. I’m testing exactly the same fix now… stay tuned.
Erez. On May 14, 2014, at 6:31 PM, Vaibhav Nagarnaik <[email protected]> wrote: > On Tue, May 13, 2014 at 8:15 PM, Theodore Ts'o <[email protected]> wrote: >> But ext4, as with more advanced file systems, such as xfs and btrfs, >> has a page_mkwrite function which is defined in the >> vm_operations_struct(). It looks like unionfs's vm_operation_struct >> does not dispatch page_mkwrite() or remap_pages(). > > Once I add page_mkwrite() handler for unionfs, I don't see the BUG_ON > anymore. It simply is a pass-through dispatch to the underlying > lower_file just like unionfs_fault(). I checked using the earlier > failing xfstests and Ted's repro case to make sure that BUG_ON is not > hit. > > I am seeing another issue where any O_DIRECT IOs are failing on ext4 > backed layers. It's the same as I saw on tmpfs backed layers. It could > be because unionfs file_operations doesn't define aio_read and > aio_write handlers. I am not sure if there's anything else needed to > support dio on unionfs. > > Vaibhav > _______________________________________________ unionfs mailing list: http://unionfs.filesystems.org/ [email protected] http://www.fsl.cs.sunysb.edu/mailman/listinfo/unionfs
