Re: LYNX-DEV Re: ...vulnerability in Lynx...

[Bela Lubkin]

Matthew Kelly wrote:

> I don't think there's any way around a /tmp directory that doesn't have
> the sticky bit set.  There isn't any monkeying that we can do to prevent
> the file from being moved/deleted unless the admin sets sticky.  Perhaps
> the only thing we should do is present a warning message if the temp dir
> does not have appropriate permissions set (ie sticky if world writeable)
> that tells them either
> 
>  (1) contact their admin to set sticky on the temp
> dir and provide a URL to give the admin for more info on the problem
> or (2) run lynx using their homedir as the temp space.
> 
> I don't this Lynx is the problem here -- a non-sticky /tmp causes problems
> down to the OS (eg ps_data race condition on Solaris -- fixed by Sun by
> making /tmp sticky).  Lets not change our default behaviour, but advise
> the user when running in an insecure situation.

Warnings about whether /tmp is "sticky" could badly mislead the user.
Not every Unix system understands the sticky bit on a directory to mean
"only owner can delete a file".  Lynx would be announcing a false
solution.

>Bela<
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;