Re: noatime mounts inhibiting atime updates via utime()

[David Holland <dholland-tech%netbsd.org@localhost>]

On Wed, Dec 05, 2018 at 12:16:41PM -0500, Thor Lancelot Simon wrote:
 > I think LFS may still have a local workaround for this.  Perhaps it
 > should be generalized.

LFS nowadays keeps atimes in the ifile, which cuts significantly down
on the write traffic they generate, but doesn't translate readily to
any other fs.

A new fs might well want to use a separate chunk of blocks as an
atime map; but that would create a huge write hotspot too so isn't
necessarily a good plan.

-- 
David A. Holland
dholland%netbsd.org@localhost