The machines that have *not* been patched look like this: -r-------- 1 root root 14050 Aug 24 15:13 /etc/shadow -rwsr-xr-x 1 root root 19632 Jul 22 17:35 /sbin/unix_chkpwd
The ones that have look like this: -r-------- 1 root root 14050 Aug 24 15:13 /etc/shadow -rwxr-sr-x 1 root shadow 26372 Aug 20 12:24 /sbin/unix_chkpwd and I see that, as of pam 1.0.1-1: * The password-changing helper functionality for SELinux systems has been split out into a separate unix_update binary, so at long last we can change unix_chkpwd to be sgid shadow instead of suid root. Closes: #155583. So, I'm guessing that we need: -r--r----- 1 root shadow 14050 Aug 24 15:13 /etc/shadow ? Thanks, -- Steve Lane System, Network and Security Administrator Doudna Lab Biomolecular Structure and Mechanism Group UC Berkeley On Sun, Aug 24, 2008 at 03:04:28PM -0700, Steve Lane wrote: > On Sun, Aug 24, 2008 at 02:41:37PM -0700, Steve Langasek wrote: > > severity 496457 normal > > tags 496457 unreproducible moreinfo > > thanks > > > > On Sun, Aug 24, 2008 at 02:13:53PM -0700, Steve Lane wrote: > > > Justification: breaks unrelated software > > > > False. Software that invokes PAM is not "unrelated". > > Sorry, I guess I wasn't clear. The software that breaks is the entire > machine, since once the screen is locked the (non-admin) user has no > recourse except to (A) find an admin to kill kdesktop_lock, or (B) > powercycle the machine. > > > > Aug 24 13:22:23 aspen unix_chkpwd[3472]: check pass; user unknown > > > Aug 24 13:22:23 aspen unix_chkpwd[3472]: password check failed for user > > > (iamsteve) > > > > Please report the permissions from /etc/shadow and /sbin/unix_chkpwd. I > > suspect you have wrong perms on /etc/shadow. > > [EMAIL PROTECTED]> ls -l /etc/shadow /sbin/unix_chkpwd > 16 -r-------- 1 root root 14020 Aug 24 14:53 /etc/shadow > 28 -rwxr-sr-x 1 root shadow 26372 Aug 20 12:24 /sbin/unix_chkpwd* > > Note that we have 13 workstations with this identical config, all but > two of which have been patched in the last two days. All of the ones > which have been patched are broken in this way; the two that have not > been patched are not. I am thus wondering if something changed with > the last libpam patch..? > > Please be aware that this bug has been cross-posted to bugs #496455 and > #496456 as well as this one (#496457). > > Thanks much, > > -- > Steve Lane > System, Network and Security Administrator > Doudna Lab > Biomolecular Structure and Mechanism Group > UC Berkeley -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]