[Alpine-info] Permissions of user mailboxes in /var/mail

[Mark Crispin]

On Sat, 29 Mar 2008, James Pittman wrote:

>> "If I understand your question correctly, the answer is yes.  It is OK for 

>> mailbox files in /var/mail to be protected 0600 without any specific group 

>> setting.  In fact, this is the normal and expected protection for mailbox 

>> files for the UW c-client library.

>

>> "Nothing in any version of Pine, Alpine, UW imapd, ipop3d, or any other UW 

>> c-client based application has any dependency upon a mailbox file being 

>> accessible by group mail.

>

> This being the case, is there any way to stop Alpine whinging about the fact 

> that my INBOX doesn't have 1777 protection every time I open it?


You have two choices:
  . set the mail spool protection to 1777 (there is NO reason AT ALL not
    to do this on a single user system, and only bad reasons not to do this
    on a multi-user system)
or
  . install the mlock tool which is included as part of the UW IMAP
    distribution.  mlock runs setgid mail, and thus does what c-client does
    not do.

The reason for that "whinging" is that there is a serious problem.  The 
problem is that NOTHING that protects your mailbox from being corrupted if 
a delivery attempt is made while Alpine/Pine/imapd/ipop3d/etc. is updating 
the mailbox.

The message is obnoxious for a reason; it is to induce you to fix the 
problem, rather than wait until your mailbox gets corrupted.

Put another way: you MUST arrange matters so that Alpine can create the 
lock files that it needs to create.

The best solution is to set the spool directory protection to 1777.  This 
whole "group mail" stuff is hokey, doesn't really buy you any true 
security, and opens security weaknesses of its own.  But if you insist 
upon using group mail, then you MUST install mlock since Alpine will not 
do so itself.

Note as well that ALL of the above applies to the spool directory. 
NOTHING should EVER need group mail access to the mailbox files 
themselves; that is a security bug that you can drive a truck through.

-- Mark --

http://staff.washington.edu/mrc
Science does not emerge from voting, party politics, or public debate.
Si vis pacem, para bellum.