similar to: no, I see now, tru64 pty ownership wrong on entry to setup_sia, may need /usr/lbin/chgpt (WAS Re: Tru64 privsep patch testing)

OK, I got a chance to try out the Tru64 patch for privsep. I applied the patch to 3.4p1. Partial success, in that it now works for me for logins to "root". Logins to ordinary accounts fail after authentication, when trying to set tty characteristics. See the excerpt from the debug messages below. This is for Tru64 V4.0F (with enhanced_security turned on, obviously.) I guess it's time

What do we loose by not having post-auth privsep? What code is executed between authorization and actual setting of the effective uid? On Tue, 3 Sep 2002, Chris Adams wrote: > Once upon a time, Toni L. Harbaugh-Blackford <harbaugh at nciaxp.ncifcrf.gov> said: > > It appears that the integration of the sia session setup will either > > have to be rethought or abandoned

Either this never made it to the list or no one cares about Tru64. This is the last time I'll send this patch to the list. If no one steps up and finishes it or provides me with enough information to fix any remaining bugs (one being complaint that 'ssh site cmd' does not work right). If there is no activity on this for a week. I'll post it to bugzilla and will ignore any

Ok.. I need wider testing for this. I'm getting reports back it works mostly. 'ssh site ls' fails, but they can login with Privsep enbled. Can I get those who are using Tru64 or OSF/1 that have SIA enabled to test? This should apple to either -cvs or the current snapshot (I would perfer not to use 3.4p1 due to bugs). I'm going on a trip next week and will be around very spotty

Hi- Under privsep, I experimented with moving the session_setup_sia() out of do_child() and into do_setusercontext(), which is where the uids/gids are set to the final execution user. The call is made with a NULL tty, and this is functional provided that any later pty allocation uses grantpty() to set the device permissions. Logging in with this method shows that a utmp entry does get made for

This privsepifies OSF/1 SIA, but I'm still being told the same error occurs. I'm stumped. Without an OSF/1 box near me I can't do too much more help unless someone can either tell me what is wrong or show me why SIA is failing in their logs. (And tell me if it's different w/ or w/out this patch) - Ben Index: auth-sia.c

Hi- Unfortunately, I just found out about the patch that was available for tru64 privsep. I was entirely unaware that there was a lack of support. Will the patch be considered for approval if it is applied to 3.4p1, or does it have to be done against -current? The reason I'm asking is that I have 3.4p1 working as is, so I know if I have a problem it is likely related to the patch and not

I have tried this with openssh-2.5.2p2 and openssh-SNAP-20010501. I have noticed that with a .hushlogin file in my home directory I still get messages. With the .hushlogin file I get these messages: Last successful login for ricardo: Tue May 1 08:06:00 2001 from blah.mnsu.edu Last unsuccessful login for ricardo: Tue May 1 08:05:21 2001 from blah.mnsu.edu Without the .hushlogin file I get

Hi- Since the tru64 patch was designed for -current, I thought I would try to build it with a recent snapshot before backporting to 3.4p1. So I downloaded openssh-SNAP-20020826.tar.gz frpm the portable snapshots, but it does not contain the 'configure' script. I tried copying the 'configure' from 3.4p1, but that does not create a Makefile from the Makefile.in. Where are the

It still is not right, but thanks to Steve we have gotten this far.. The issue seems to be here: debug3: entering: type 26 debug3: entering debug1: session_new: init debug1: session_new: session 0 debug3: entering: type 26 : sendmsg(12): Invalid argument debug1: Calling cleanup 0x1200365c0(0x14000d9d8) debug1: session_pty_cleanup: session 0 release /dev/ttyp4 debug1: Calling cleanup

Hi- I applied the privsep patch to Tru64 5.1A openssh 3.4p1 and it *almost* works. I get in from the client side and xauth is run, but in the meantime the server side disconnects. Running sshd in debug mode level 3 gives the following output: . . . debug1: session_input_channel_req: session 0 req shell debug1: fd 5 setting TCP_NODELAY debug1: channel 0: rfd 13

Well, I had just finally gotten around to downloading a snapshot to test the latest on Tru64 a couple of days ago but hadn't had a chance to build it yet, and 3.7p1 has now been released. Sigh. The problem is that Tru64 setreuid() and setregid() are broken, so privsep doesn't work. This could also be a security problem for SIA authentication in general (any version of OpenSSH on Tru64,

As I understand it, the idea behind privsep is to prevent malicious data from the client-side of a connection corrupting a server-side process running as root. To achieve that, it is important that post-auth privilege separation happen, ie, that the sshd process change uid to the (authenticated) user. But it is also true that this very same process can perform root-level work without risk of being

A while back, I sent in a patch that added Digital Unix SIA authentication to OpenSSH. Well, I just figured out that it didn't handle everything correctly (locked accounts could still log in). I thought I had checked that, but I guess I missed it. Anyway, here is a patch against OpenSSH 2.2.0p1 that fixes this. -- Chris Adams <cmadams at hiwaay.net> Systems and Network Administrator

We don't currently support SVR4-style fd passing which involves I_SENDFD/I_RECVFD ioctls. I'm not sure if that might help some p platforms with privsep. Some platforms that use STREAMS ptys use a set-uid root helper to handle the privileged portions of pty creation. grantpt(3) has a bit more info. on Solaris 8 it's: /usr/lib/pt_chmod on HP-UX it's: /usr/lbin/chgpt Perhaps this

Tru64 patch will not make it into 3.5 (this is final) due to lack of willing people to test. I have given the Tru64/osf1 community almost a month to test it. And *ONE* person came forward to give me verification. And don't give me shit about "I don't have time." The person who tested it was LEAVING his employer with Tru64. He found time. IT IS YOUR GAWD DAMN PLATFORM. IF

Hi, I'm having issues getting the multiple authentication methods of dovecot 1.0-test65 to work correctly on NetBSD. I am combining a MySQL database of virtual host accounts with the system password database of local machine accounts. Both work correctly when enabled on their own, but when enabled together only the one listed first in the "auth default" section of dovecot.conf

Is anyone maintaining the OSF_SIA support in openssh? This seems to be an obvious bug triggered if you try to connect as a non-existant user. >From auth1.c line 459 #elif defined(HAVE_OSF_SIA) (sia_validate_user(NULL, saved_argc, saved_argv, get_canonical_hostname(), pw->pw_name, NULL, 0, NULL, "") == SIASUCCESS)) { #else /*

A patch for a few more hiccups and trivialities in kinit.c: * The check_path() calls check for "/root" and "/old_root" - I believe that should be "/root" and "/root/old_root". * chdir("/") is recommended after pivot_root() * init_argv[0] isn't set properly to the basename pointed to by char *s - this fix also eliminates six lines of

Well, after digging around and thinking some more, I'm giving up on the idea of preallocating a TTY to get post-auth privsep working on Tru64. I don't think it will work, because just allocating a TTY doesn't fix the problem - there's no valid way to tie that TTY back to the client process (because it hasn't requested a TTY yet and may not ever do so). The problem is that the