xref: /src/usr.sbin/rpc.pcnfsd/README

July 1995 -  Ported to NetBSD by Gordon Ross <gwr@netbsd.org>

Note: to build clnt.pcnfsd use the command:
	make -f Makefile.clnt

Text of original README file from Sun follows:
----------------------------------------------------------------------

This is version @(#)README	1.6 1/28/92 of the pcnfsd README file.
----------------------------------------------------------------------
This is the second cut at PCNFSD v2. All printing now uses the same
SVR4 based printing model: lp, lpstat, cancel. This implies SunOS 4.1
or later on a Sun. The printing hasn't been properly tested on SVR4,
since my SVR4 system is a bit quirky, but authentication via shadow
passwords works fine.

The Makefile supplied will build either SunOS 4.1 or SVR4 versions;
however you will need to tweak a couple of definitions. See the
Makefile for details.

For this round, I have frozen the rpcgen'd code and modified it by hand
to get around a few rpcgen deficiencies. If in doubt, you can take the
pcnfsd.x file included here off to a separate directory, rpcgen everything
up and diff against the versions given here.

You should also be sure to read the man page. I haven't tried [nt]roff'ing
this anywhere outside SunOS, so let me know if there are any quirks.
Note especially the /etc/pcnfsd.conf configuration model.

Standard disclaimers. Read the source for more of the same.

Geoff Arnold (Geoff.Arnold@Sun.COM)
May 2, 1991

----------------------------------------------------------------------
This is the first major spin since the version I posted in May.
Changes are described in the SCCS deltas, but anyone analysing
or porting this should watch for the following:

- Lots of SVR4 stuff. Look at the Makefile and common.h; you will
  need to make sure that SVR4 is defined to build a version for SVR4.
- Fixed a major bug whereby aliased printers (a.k.a. virtual printers)
  didn't show up in the printer list.
- Since some commands (especially lpq) can hang up forever if, say,
  the host of a remote printer is down, the su_popen mechanism now
  includes a watchdog timer. I HAVE NOT HAD A CHANCE TO TEST THIS
  LOGIC ON SVR4 YET! The code was derived from the System V compatiility
  section of the AnswerBook, so it _should_ be OK, but I can't wait to
  test it before posting this.
- I've mades lots of lint-suggested improvements. It doesn't lint
  100% cleanly yet, but it's a lot better.
- The client testbed has been tightened up a bit and the reporting
  is somewhat clearer. Please run it.

Feedback is solicited.

Geoff Arnold (Geoff.Arnold@Sun.COM)
July 24, 1991

----------------------------------------------------------------------------

This version includes support for TCP as well as UDP, both in the
server and in the test client.  Some of the calls can have long
responses, in which case TCP is the preferred transport.

Jon Dreyer
January 28, 1992

----------------------------------------------------------------------------

This release incorporates minor bug fixes occasioned by internal testing
against preliminary versions of PC-NFS subsystems which actually use the
V2 protocol features. It also addresses a security hole which was identified:
see the comment in pcnfsd_print.c for details. This is the version
of the daemon which we expect to send out for beta testing.

Geoff Arnold
November 16, 1991

----------------------------------------------------------------------------

Several people pointed out that if I was going to fix the security
hole, I should *REALLY* fix it. So this revision checks every argument
for every shell metacharacter. This means that if your print subsystem
allows such characters in printer names, print job id's, user names,
etc. you're going to have to relax the restrictions or tighten up
elsewhere. The routine "suspicious()" in pcnfsd_print.c is the starting
point for this.

Geoff Arnold
November 21, 1991

----------------------------------------------------------------------------

This version includes one more tweak on the previous security fix, plus
a fix for the "user shell" security hole.

Geoff Arnold
December 17, 1991

----------------------------------------------------------------------------

Fix a stupid bug - the secondary groups array wasn't static, so it
got overwritten, resulting in the corruption of some GIDs.

Geoff Arnold
December 18, 1991