On Mon, Dec 3, 2018 at 11:22 AM Amir Goldstein <amir73il@xxxxxxxxx> wrote: > > On Mon, Dec 3, 2018 at 10:17 AM Dave Chinner <david@xxxxxxxxxxxxx> wrote: > > > > On Mon, Dec 03, 2018 at 09:25:19AM +0200, Amir Goldstein wrote: > > > On Mon, Dec 3, 2018 at 8:43 AM Dave Chinner <david@xxxxxxxxxxxxx> wrote: > > > > > > > > From: Dave Chinner <dchinner@xxxxxxxxxx> > > > > > > > > Test that copy_file_range will return the correct errors for various > > > > error conditions and boundary constraints. > > .... > > > > > > All the test cases above check for bugs, which I presume your kernel patch > > > series is aimed at fixing(?) > > > > Yes. Document the API (I have a man page patch) write the tests to > > exercise correct API behaviour (these patches), fix the API > > implementation until the tests start passing (still to be posted as > > I wait for these to hit mailing list archives so I can point at > > them). > > > > > This one last test case tests for new functionality that is not > > > currently available > > > for any filesystem in upstream kernel. > > > > Yup. > > > > > Does your kernel patch set also add this functionality to xfs? to generic? > > > > Yes and yes. overlay works, too, but I gave up caring about it > > because it doesn't support the ioctls xfs_io uses in this test to > > change open file state.... > > Oh, not a problem. I can add FS_IOC_FS[SG]ETXATTR > to the overlayfs white list of ioctls. > > That raises the issue that the test is missing: > > _require_xfs_io_command "chattr" "-i" > > and the chattr -i in cleanup(). > > > > > > IMO, it would be better to split this test case for new functionality to a new > > > test, so that this one can pass on stable kernels once all the bug > > > fixes have been > > > applied. > > > > Whatever. I'm tired, I've already put in 13 hours on this today and > > I'm on the back of four 100+ hour weeks working on nothing but this > > broken heap of crap. > > > > Take it or leave it, because I'm just about burnt out on this > > right now... > > > > I hear you. > > I personally have no problem with declaring the cross fs copy_range an > interface bug fix that can also go to stable. It's probably going to be harder > to get your interface fixes without it. > > I will leave the call about insisting on separate test to Eryu. > If you want me to submit that separate test, I can do that. > Getting back to this (how time flies...) Dave, if you don't mind I am going to re-post your tests splitting the cross device copy test, since Darrick also echoed this request. Cheers, Amir.
