1. sys/fs/tmpfs/tmpfs_subr.c:1691]: (style) Same expression on both sides of '&&'. if (vap->va_birthtime.tv_nsec != VNOVAL && vap->va_birthtime.tv_nsec != VNOVAL) Looking at nearby code, maybe if (vap->va_birthtime.tv_sec != VNOVAL && vap->va_birthtime.tv_nsec != VNOVAL) 2. sys/fs/tmpfs/tmpfs_subr.c:1697]: (style) Same expression on both sides of '&&'. Same thing a few lines further down.
A commit references this bug: Author: brueffer Date: Sun Jul 26 08:33:47 UTC 2015 New revision: 285885 URL: https://svnweb.freebsd.org/changeset/base/285885 Log: In tmpfs_chtimes(), remove checks on the nanosecond level when determining whether a node changed. Other filesystems, e.g., UFS, only check on seconds, when determining whether something changed. This also corrects the birthtime case, where we checked tv_nsec twice, instead of tv_sec and tv_nsec (PR). PR: 201284 Submitted by: David Binderman Patch suggested by: kib Reviewed by: kib MFC after: 2 weeks Committed from: Essen FreeBSD Hackathon Changes: head/sys/fs/tmpfs/tmpfs_subr.c
Fixed in HEAD, I'll merge this in a little bit. Thanks!
(In reply to Christian Brueffer from comment #2) FWIW, playing with coccinelle, I noticed this issue is still on 10-stable. Gentle reminder to MFC ;).
A commit references this bug: Author: emaste Date: Fri Mar 9 17:59:22 UTC 2018 New revision: 330700 URL: https://svnweb.freebsd.org/changeset/base/330700 Log: MFC r285885 by brueffer: In tmpfs_chtimes(), remove checks on the nanosecond level when determining whether a node changed. Other filesystems, e.g., UFS, only check on seconds, when determining whether something changed. This also corrects the birthtime case, where we checked tv_nsec twice, instead of tv_sec and tv_nsec (PR). PR: 201284 Changes: _U stable/10/ stable/10/sys/fs/tmpfs/tmpfs_subr.c