topic DNS$MSG.EXE protection in Operating System - OpenVMS

DNS$MSG.EXE protection

Thomas A. Williams — Mon, 21 Aug 2006 12:49:17 GMT

Someone noticed that SYS$MESSAGE:DNS$MSG.EXE doesn't have W:RE protection. (World has NO access). All the other EXE files in SYS$MESSAGE have W:RE access. Does anyone know why? Is this on purpose?

Re: DNS$MSG.EXE protection

Steven Schweda — Mon, 21 Aug 2006 16:29:47 GMT

I have two:

DNS$MSG.EXE;1 (RWED,RWED,RE,)
DNS$MSG_VMS.EXE;1 (RWED,RWED,RE,)

but there's a similar-looking:

DNS$MSG_V.EXE;1 (RWED,RWED,RE,RE)

Looks to me like an error.

A bunch of mine have G:REWD, too, but most
are G:RE.

Re: DNS$MSG.EXE protection

John Gillings — Mon, 21 Aug 2006 23:09:38 GMT

Thomas,

It might be an oversight. On my V6.2 systems, the protection is W:RE, but on later versions it's W(none).

I guess it depends on what images are intended to use the message file. If they're all installed, privileged images, or they're intended to be used only by privileged users, then the security may be deliberate, or may not be relevant.

A quick look at a V8.2 system disk, it appears DNS$MSG is referenced from NET$EVENT_DISPATCHER, NCLSHR, NET$EVD_RELAY_FORMATTER and DNS$MSG_V. I can't imagine joe-unprivileged-user messing with any of them.

On one hand, it probably doesn't matter if you open W:RE access to the message file, but on the other, what's broke that you think you're fixing?

If your users aren't complaining about failure to translate DNS messages, it might be better to just leave it be. If you have file access failure auditing enabled, you'll now quickly enough if anyone who should have access is being denied.

Re: DNS$MSG.EXE protection

Steven Schweda — Mon, 21 Aug 2006 23:21:07 GMT

I don't know about anyone else, but I have a
procedure which tries to get the message text
for a status code, and when it gets "NOMSG",
it tries to "set message" on each of the
files in sys$message until it gets a good
message (or runs out of files).

Anyone, privileged or not, might wish to
translate such an error code, but it won't
work well if he can't read the file.

The probability of anyone actually caring may
be fairly small, but I estimate the
probability of W:RE on any of these files
being wrong enough to cause trouble to be
_vanishingly_ small.

I'd vote for setting all of them the same,
and then waiting for a disaster. I'd expect
to be waiting a very long time.