From: Jim F. <jf...@an...> - 2006-03-14 00:28:23
|
On 3/7/06, Wesley Craig <we...@um...> wrote: > Oh well. Terry Lambert is smarter than POSIX, I guess. Any thoughts > on the supposed security hole on Linux? This hole presumably applies I looked for an lchown() in the v3 Single Unix Spec, which is supposed to be harmonized with POSIX and found it: http://www.opengroup.org/onlinepubs/009695399/functions/lchown.html I didn't discover any conceptual conversation on symlink permissions, but I didn't look too hard. In any case, the existence of lchown() must mean that symlink permissions have some viable status. So while Apple may be being inconvenient and stubborn as usual, they may also have some standards to back them up. I'm really curious about what security hole this supposedly fixes though. All the systems types I've run this past haven't been able to think of one either. It certainly does nothing against the "standard" symlink creation/traversal attacks. Now if lchown() only existed on Tiger. I suppose we should just count our blessings instead that they finally bothered to document getattrlist/setattrlist. -Jim |