Re: utimensat fails to update ctime

[Christoph Hellwig <hch@infradead.org>]

On Tue, Dec 22, 2009 at 05:42:49AM -0700, Eric Blake wrote:
> According to Dave Chinner on 12/22/2009 5:34 AM:
> > Yeah, it looks like the change to utimesat() back in 2.6.26 for
> > posix conformance made ATTR_CTIME appear outside inode truncation
> > and XFS wasn't updated for this change in behaviour at the VFS level.
> > Looks simple to fix, but I'm worried about introducing other
> > unintended ctime modifications - is there a test suite that checks
> > posix compliant atime/mtime/ctime behaviour around anywhere?
> 
> Yes - the gnulib unit test, consisting of:
> 
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/nap.h
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/test-utimens-common.h
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/test-utimens.h
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/test-lutimens.h
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/test-futimens.h
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/test-futimens.c
> http://git.savannah.gnu.org/cgit/gnulib.git/tree/tests/test-utimensat.c
> 
> Taken together, these files produce the executables test-futimens and
> test-utimensat which can demonstrate compliance with POSIX.  They are also
> bundled as part of GNU coreutils (the version bundled with coreutils 8.2
> didn't test for ctime compliance, and coreutils 8.3 hasn't been released
> yet) if you use 'make -C gnulib-tests check'.

Ok, I'll see if I can add the last GPLv2 licensed version into xfstests.