Email List: Xaustin-review-lX
[All Lists]

RE: Defect in XCU rm

To: Mark Funkenhauser <yyyyyyyyyyyyy@xxxxxxxxxx>
Subject: RE: Defect in XCU rm
From: yyyyyyy@xxxxxxxx
Date: Wed, 12 Mar 2003 22:24:47 -0500 (EST)
Cc: "'Geoff Clare'" <yyy@xxxxxxxxxxxxx>, <yyyyyyyyyyyyyyy@xxxxxxxxxxxxx>
On Wed, 12 Mar 2003, Mark Funkenhauser wrote:

> I don't believe your suggestion would be acceptable.
>
> The wording in 3.4 and 4.4 was explicitly added to the POSIX.1 standard
> to allow implemenations of addition *security* mechanisms and security
> policies.
> (Like enhanced Descretionary Access Controls)
> These are low level mechanisms, integral to the overall security policy
> of the system and are applicable to all files and api's on the system.
>
> The issue being discussed here has to do with a particular unpleasant
> side effect
> of a specific scenario with the rm utility.
> It has nothing to do with file access control or security mechanisms.
>
> The wording in 3.4 and 4.4 cannot be used to justify arbitrary
> implementation specific behaviours of individual utilities.

I agree here. An "Access Control" can prevent the deletion of slash itself
given the current wording, but implementations already may (shall?) fail
with EBUSY (EACCES?) on rmdir("/"). Without additional wording changes
this does not allow rm to avoid recursively removing the *contents* of
slash.

<Prev in Thread] Current Thread [Next in Thread>