On Mon, 22 Aug 2022, Jeff Layton wrote: > Add an explicit paragraph codifying that atime updates due to reads > should not be counted against the i_version counter. None of the > existing subsystems that use the i_version want those counted, and > there is an easy workaround for those that do. > > Cc: NeilBrown <neilb@xxxxxxx> > Cc: Trond Myklebust <trondmy@xxxxxxxxxxxxxxx> > Cc: Dave Chinner <david@xxxxxxxxxxxxx> > Link: https://lore.kernel.org/linux-xfs/166086932784.5425.17134712694961326033@xxxxxxxxxxxxxxxxxxxxx/#t > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx> > --- > include/linux/iversion.h | 10 ++++++++-- > 1 file changed, 8 insertions(+), 2 deletions(-) > > diff --git a/include/linux/iversion.h b/include/linux/iversion.h > index 3bfebde5a1a6..da6cc1cc520a 100644 > --- a/include/linux/iversion.h > +++ b/include/linux/iversion.h > @@ -9,8 +9,8 @@ > * --------------------------- > * The change attribute (i_version) is mandated by NFSv4 and is mostly for > * knfsd, but is also used for other purposes (e.g. IMA). The i_version must > - * appear different to observers if there was a change to the inode's data or > - * metadata since it was last queried. > + * appear different to observers if there was an explicit change to the inode's > + * data or metadata since it was last queried. Should rename change the i_version? It does not explicitly change data or metadata, though it seems to implicitly change the ctime. > * > * Observers see the i_version as a 64-bit number that never decreases. If it > * remains the same since it was last checked, then nothing has changed in the > @@ -18,6 +18,12 @@ > * anything about the nature or magnitude of the changes from the value, only > * that the inode has changed in some fashion. > * > + * Note that atime updates due to reads or similar activity do _not_ represent > + * an explicit change to the inode. If the only change is to the atime and it > + * wasn't set via utimes() or a similar mechanism, then i_version should not be > + * incremented. If an observer cares about atime updates, it should plan to > + * fetch and store them in conjunction with the i_version. > + * If an implicit atime update happened to make the atime go backwards (possible, but not common), the updating i_version should be permitted, and possibly should be preferred. NeilBrown > * Not all filesystems properly implement the i_version counter. Subsystems that > * want to use i_version field on an inode should first check whether the > * filesystem sets the SB_I_VERSION flag (usually via the IS_I_VERSION macro). > -- > 2.37.2 > >
