[linux-cifs-client] Re: chmod 01777 has different SETFILEBITS EA from chmod 01444

[Martin Koeppe]

On Thu, 17 Nov 2005, Steven French wrote:

> OK - got your results must have been user error before.  SETFILEBITS value
> is double the value of the 4th octal digit   Still I have no idea what bit
> 0 means (since I never see it set).

Yes, SETFILEBITS value is 32bit, and only 3 we currently know about, 
so there remain 29 unknown bits, from that point of view.
Maybe it's simply because of this:

full_mode = lower_9_bit_mode | SETFILEBITS;


> Created the files locally with the following snippet of code, then did
> "getfattr -d -e hex /mnt/test-dir/*"
> getfattr worked fine for me (I have cifs compiled with xattr support
> enabled - perhaps you have that Kconfig flag turned off?)

Oh, yes, thank you. I used a Debian supplied kernel, in which xattr 
support is diabled. While compiling my own kernel, I noticed that 
xattr is still marked as EXPERIMENTAL (2.6.14).

Do you think it's stable / useful enough to be generally enabled?

If yes, I could write to the Debian kernel team and ask to enable 
this. The EXPERIMENTAL flag itself should not be a problem, as there 
are other such Kconfigs already enabled in Debian kernels.


Martin