[PATCH] Fix bug #11249 - Mangled names do not work with acl_xattr

[Jeremy Allison]

On Mon, May 04, 2015 at 05:08:41PM -0700, Richard Sharpe wrote:
> On Mon, May 4, 2015 at 4:47 PM, Jeremy Allison <jra at samba.org> wrote:
> > On Mon, May 04, 2015 at 04:29:53PM -0700, Jeremy Allison wrote:
> >> On Mon, May 04, 2015 at 10:29:02PM +0200, Andreas Schneider wrote:
> >> > On Monday 04 May 2015 12:39:30 Jeremy Allison wrote:
> >> > > Please review and/or push !
> >> >
> >> > Could you please add 'BUG: ' infront of the bugzilla urls? Thanks :)
> >> >
> >> > Then you have my RB+
> >>
> >> Thanks - looks like there's an issue with this patch and
> >> the streams tests. I'm getting:
> >>
> >> (../source4/torture/raw/streams.c:287) creating a stream on a non-existent file
> >> (../source4/torture/raw/streams.c:306) check that open of base file is allowed
> >> (../source4/torture/raw/streams.c:313) writing to stream
> >> (../source4/torture/raw/streams.c:319) Failed to open stream '\teststreams\stream.txt:Stream One' - NT_STATUS_OBJECT_NAME_NOT_FOUND
> >> UNEXPECTED(failure): samba3.raw.streams.io(nt4_dc)
> >> REASON: Exception: Exception: ../source4/torture/raw/streams.c:324: status was NT_STATUS_OBJECT_NAME_NOT_FOUND, expected NT_STATUS_OK: CHECK_STATUS
> >>
> >> So I'll debug and resubmit. Thanks for the help Andreas !
> >
> > Oh great :-(. Looks like stream names can have EA's attached
> > to them. Well *that* sucks.
> 
> Really? Since streams and EAs are both attributes (there is a term for
> that in NTFS-speak but my book is at work) on the main file, that
> seems unlikely ...

Yeah, I might have been wrong on that... Still trying to
figure out the layering in the test cases. It's hellishly
complex with a VFS stack of:

vfs objects = acl_xattr fake_acls xattr_tdb streams_depot

so debugging this is fun. I'll figure it out... Just might
take me a little while to get something that passes all
the old tests as well as the new foo:bar regression test...