Date: Tue, 16 Dec 2014 14:37:16 -0500 From: John Baldwin <jhb@freebsd.org> To: Warner Losh <imp@bsdimp.com> Cc: arch@freebsd.org Subject: Re: Change default VFS timestamp precision? Message-ID: <201412161437.16767.jhb@freebsd.org> In-Reply-To: <708ECB13-C3A1-46E9-BF29-6F544CC4FDE6@bsdimp.com> References: <201412161348.41219.jhb@freebsd.org> <708ECB13-C3A1-46E9-BF29-6F544CC4FDE6@bsdimp.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday, December 16, 2014 1:51:09 pm Warner Losh wrote:
>=20
> > On Dec 16, 2014, at 11:48 AM, John Baldwin <jhb@freebsd.org> wrote:
> >=20
> > We still ship with vfs.timestamp_precision=3D0 by default meaning that =
VFS
> > timestamps have a granularity of one second. It is not unusual on mode=
rn
> > systems for multiple updates to a file or directory to occur within a s=
ingle
> > second (and thus share the same effective timestamp). This can break t=
hings
> > that depend on timestamps to know when something has changed or is stal=
e (such
> > as make(1) or NFS clients). On hardware that has a cheap timecounter, =
I we
> > should use the most-precise timestamps (vfs.timestamp_precision=3D3). =
However,
> > I'm less sure of what to do for other cases such as i386/amd64 when not=
using
> > TSC, or on other platforms. OTOH, perhaps you aren't doing lots of hea=
vy I/O
> > access on a system with a slow timecounter (or if you are doing heavy I=
/O,
> > slow timecounter access won't be your bottleneck)?
> >=20
> > I can think of a few options:
> >=20
> > 1) Change vfs.timestamp_precision default to 3 for all systems.
> >=20
> > 2) Only change vfs.timestamp_precision default to 3 for amd64/i386 usin=
g an
> > #ifdef.
> >=20
> > 3) Something else?
> >=20
> > What do other folks think?
>=20
> (1). If there=E2=80=99s a specific kernel / platform that=E2=80=99s slow,=
we can make it an option
> for those kernels.
Mm, so something like:
Index: vfs_subr.c
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=2D-- vfs_subr.c (revision 275828)
+++ vfs_subr.c (working copy)
@@ -632,7 +632,11 @@ vfs_getnewfsid(struct mount *mp)
*/
enum { TSP_SEC, TSP_HZ, TSP_USEC, TSP_NSEC };
=20
+#ifdef VFS_CHEAP_TIMESTAMPS
static int timestamp_precision =3D TSP_SEC;
+#else
+static int timestamp_precision =3D TSP_NSEC;
+#endif
SYSCTL_INT(_vfs, OID_AUTO, timestamp_precision, CTLFLAG_RW,
×tamp_precision, 0, "File timestamp precision (0: seconds, "
"1: sec + ns accurate to 1/HZ, 2: sec + ns truncated to ms, "
where VFS_CHEAP_TIMESTAMPS becomes a new kernel config option?
(We should also probably make this a loader tunable)
=2D-=20
John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201412161437.16767.jhb>