On 9/30/15 4:56 PM, Dave Chinner wrote: > On Wed, Sep 30, 2015 at 10:57:45PM +0300, Roman Lebedev wrote: >> As per overlayfs documentation, any activity on a merged directory >> for a application that is doing such activity should work exactly >> as if that would be a normal, non overlayfs-merged directory. >> >> That is, e.g. simple fopen-fwrite-fsync-fclose sequence should >> work just fine. > > We have plenty of tests that do things like that. > >> But apparently it does not. Add a simple generic test to check that. >> As of right now (linux-4.2.1) this test fails at least on btrfs. >> >> PS: An alternative (and probably better approach) would be to run >> fstests test suite with TEST_DIR set to overlayfs work directory. > > Much better is to run xfstests directly on overlayfs. THere have > been some patches to do that posted in the past, but those patches > and discussions kinda ended up going nowhere: > > http://www.mail-archive.com/fstests@xxxxxxxxxxxxxxx/msg00474.html > > Perhaps you'd like to pick this up, and then overlay will by much > easier to test and hence likely not to have bugs like this... Yeah, that could still be used for fun, but Zach's POV was that we should just have a specific overlayfs config (dictating paths to over/under/merge/around/through/whatever directories), a special mount_overlayfs helper, etc, ala NFS & CIFS. It may actually be easier than what I proposed. If you want to take a stab at it I'm happy to help, answer questions, etc - I'm not sure when I'll get back to it... -Eric -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
