Nanoseconds times on EXT3 always seem to be zero

Nanoseconds times on EXT3 always seem to be zero

[John David Anglin]

On an EXT3 file system, I see the following:

dave@mx3210:~$ touch xyzzy
dave@mx3210:~$ ls --full-time xyzzy
-rw-r--r-- 1 dave dave 0 2008-01-06 15:07:01.000000000 -0500 xyzzy

However, on tmpfs, proc, etc., I see the sub-seconds time.  This is
2.6.22.14 and 2.6.22.15.

The behavior of 32 and 64-bit kernels seems to be the same.

With an old 2.6 x86 kernel (suse), I see sub-second times on a EXT3
file system.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: Nanoseconds times on EXT3 always seem to be zero

[Carlos O'Donell]

On Jan 6, 2008 3:13 PM, John David Anglin <dave@hiauly1.hia.nrc.ca> wrote:
> On an EXT3 file system, I see the following:
>
> dave@mx3210:~$ touch xyzzy
> dave@mx3210:~$ ls --full-time xyzzy
> -rw-r--r-- 1 dave dave 0 2008-01-06 15:07:01.000000000 -0500 xyzzy
>
> However, on tmpfs, proc, etc., I see the sub-seconds time.  This is
> 2.6.22.14 and 2.6.22.15.
>
> The behavior of 32 and 64-bit kernels seems to be the same.
>
> With an old 2.6 x86 kernel (suse), I see sub-second times on a EXT3
> file system.

Can you distill this into a testcase that uses the kernel fstat* syscalls?

That would definitely rule out glibc getting in the way.

c.

Re: Nanoseconds times on EXT3 always seem to be zero

[Matthew Wilcox]

On Mon, Jan 07, 2008 at 12:42:56PM -0500, Carlos O'Donell wrote:
> Can you distill this into a testcase that uses the kernel fstat* syscalls?
> 
> That would definitely rule out glibc getting in the way.

ext3 doesn't store nanoseconds on disc ... unless <blah blah blah>

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."

Re: Nanoseconds times on EXT3 always seem to be zero

[John David Anglin]

> On Mon, Jan 07, 2008 at 12:42:56PM -0500, Carlos O'Donell wrote:
> > Can you distill this into a testcase that uses the kernel fstat* syscalls?
> > 
> > That would definitely rule out glibc getting in the way.
> 
> ext3 doesn't store nanoseconds on disc ... unless <blah blah blah>

I was told today that 2.6.24 may provide nanoseconds with ext3, else ext4.
The documentation for <blah blah blah> seems lacking.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)