On Fri, May 22, 2009 at 04:21:14PM -0400, John Baldwin wrote: > On Friday 22 May 2009 4:08:31 pm Dimitry Andric wrote: > > On 2009-05-22 21:07, John Baldwin wrote: > > >> Log: > > >> some ports erroneously use the existence of AT_FDCWD to check for > > >> the availability of the *at system calls so > > >> s/AT_FDCWD/AT_FDCWD_notyet/g > > >> > > >> Reported by: Dimitry Andric > > > > > > I wonder if it wouldn't be appropriate to just remove the dirfd bits > > > entirely > > > from 7.x for now. I.e. remove NDINIT_AT(), AT_FDCWD_notyet, the 'dirfd' > > > member from the structure, the 'dirfd' parameter from NDINIT_ALL(), etc. > > > ZFS > > > only really needed NDINIT_ATVP(), yes? > > > > Are there any plans to MFC the *at() calls? (I'd guess not, since it > > looks like they change the ABI...) > > I don't think there are due to the ABI change, and if they were ever MFC'd, > the NDINIT, etc. bits could be restored as part of that merge. No, the MFC is posponed not due to KBI change. As I said, I already have a permission from re@ to change nameidata.
Robert (Cc:ed) hold the MFC because the new fd argument of the at syscalls is not audited.
pgpqmCRKDHZav.pgp
Description: PGP signature
