Date: Sat, 3 Nov 2012 13:56:04 -0700 From: Garrett Cooper <yanegomi@gmail.com> To: Peter Jeremy <peter@rulingia.com> Cc: FreeBSD FS <freebsd-fs@freebsd.org> Subject: Re: Inconsistent/potentially incorrect behavior with relative lookups via chdir(2) on UFS/ZFS Message-ID: <CAGH67wQ5z=8o%2Bu00zqFdeCbhE=CCPOgms94D943v9uyP5BHBCg@mail.gmail.com> In-Reply-To: <20121103103849.GD12996@server.rulingia.com> References: <CAGH67wSa7y-Nz=Pm%2BpxL3TRyKvBdDq2s3paSCu3LB-AFwmO_2g@mail.gmail.com> <20121103103849.GD12996@server.rulingia.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Nov 3, 2012 at 3:38 AM, Peter Jeremy <peter@rulingia.com> wrote:
> On 2012-Nov-01 15:25:05 -0700, Garrett Cooper <yanegomi@gmail.com> wrote:
> > Just doing some interop testing on UFS/ZFS to develop a baseline for
> >filesystem behavior, and I noticed some inconsistencies with the ENOENT
> >requirement in chdir(2) when dealing with relative ".." paths (dot-dot
> >lookups). In particular...
> > 1. I would have expected chdir('.') to have failed in UFS/ZFS with
> >ENOENT if '.' wasn't present, but it didn't.
> > 2. I would have expected chdir('..') to have failed in ZFS with ENOENT
> >if '..' wasn't present, but it didn't.
> > Sidenote: python doesn't do any special handling with os.chdir, per
> >Modules/posixmodule.c (I checked).
> > The full test I ran is included below.
>
> Whilst playing with the above, I've found some wierd timing issues
> with UFS+SU:
>
> $ mkdir -p ~/p/q/r;sync
> $ cd p/q/r
> $ rm -r ~/p; while date; do sleep 3 ; ls -al;done
> Sat 3 Nov 2012 21:29:19 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:22 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:25 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:28 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:31 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:34 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:37 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:40 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:43 EST
> total 2
> drwxr-xr-x 0 peter jeremy 512 3 Nov 21:28 .
> drwxr-xr-x 0 peter jeremy 0 3 Nov 21:29 ..
> Sat 3 Nov 2012 21:29:46 EST
> total 0
> Sat 3 Nov 2012 21:29:49 EST
>
Interesting; with SU-J or SUJ (my testing was with the default: SUJ)?
Thanks!
-Garrett
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAGH67wQ5z=8o%2Bu00zqFdeCbhE=CCPOgms94D943v9uyP5BHBCg>