Re: Please review: lookup changes

[Andrew Doran <ad%netbsd.org@localhost>]

On Mon, Mar 09, 2020 at 07:59:03PM +0000, Andrew Doran wrote:

> > Why don't we check our watch immediately when we set IN_CHANGE &c.,
> > and defer the disk write to update the inode like UFS_WAPBL_UPDATE
> > does -- and just skip UFS_ITIMES in VOP_GETATTR?
> 
> This looks to go back least as far as Unix 32V (& possibly further) in one
> form or another, from what I can see.  Whatever the original intent I think
> it might have morphed into an attempt to avoid querying the system clock at
> some point, which is about the only point I can see for it now.

Considering this originates in a different era..  I suppose it might also
yield a better chance of observing distinct timestamps, if the system had
poor clock resolution.

Andrew