On 1/4/21 1:31 PM, Darrick J. Wong wrote: > On Tue, Jan 05, 2021 at 02:57:54AM +0800, Zorro Lang wrote: >> On Mon, Jan 04, 2021 at 10:29:21AM -0600, Eric Sandeen wrote: >>> >>> >>> On 1/4/21 9:28 AM, Eric Sandeen wrote: >>>> On 1/4/21 5:30 AM, Zorro Lang wrote: >>>>> When set inobtcount=1/0, we can't see it from xfs geometry report. >>>>> So make it visible. >>>>> >>>>> Signed-off-by: Zorro Lang <zlang@xxxxxxxxxx> >>>> Hi Zorro - thanks for spotting this. >>>> >>>> I think the libxfs changes need to hit the kernel first, then we can >>>> pull it in and fix up the report_geom function. Nothing calls >>>> xfs_fs_geometry directly in userspace, FWIW. >>> >>> Hah, of course I forgot about libxfs_fs_geometry. o_O >>> >>> In any case, I think this should hit the kernel first, want to send >>> that patch if it's not already on the list? >> >> I can give it a try, if Darrick haven't had one in his developing list :) > > Why do we need to expose INOBTCNT/inobtcount via the geometry ioctl? > It doesn't expose any user-visible functionality. Just to follow up, my thought here is that mkfs output / xfs_info output should reflect what's needed to create a filesystem w/ the same geometry/features, and since this is a mkfs option, it should be reflected. Darrick already RVB'd the kernel patch, so we're now just cruising along like a well-oiled machine. ;) -Eric
