Lustre devel - Jan 2007 - [Bug 11463] lfs quotachown <file system> deletes security relevant setuid bit

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11463

           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #9366 is|0                           |1
           obsolete|                            |


Created an attachment (id=9395)
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
 --> (https://bugzilla.lustre.org/attachment.cgi?id=9395&action=view)
patch for bug11463(following the adilger''s instruction)

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11463



(In reply to comment #11)
> (From update of bug11463">attachment 9366)
> I have no problem with the technical parts of this patch, but I have some
> concern that this will cause Lustre to violate POSIX requirements for the
> behaviour of chmod.
> SUSv2 says for chmod(2):
I am not sure why is this relevant. This is chmod, and we modify chown
behaviour.
And If we do not change group or user, no point in dropping suid/sgid, I wou;d
presume and we violate nothing.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11463



I just want to confirm something - the second patch has obsoleted the first
patch, yet the whiteboard status update says that 2 patches have been
supplied...

The latter would appear to suggest that both patches are necessary?

Fergal.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11463

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Attachment #9395|review?(adilger@clusterfs.co|review+
               Flag|m)                          |


(From update of attachment 9395)
I have no objection to this change, but in hindsight I think we also need to
reset the atime and mtime via utimes after chmod() is done (this was true of
the earlier code also and is not related to the addition of chmod()).