Call for testing: OpenSSH 6.6

mancha mancha1 at hush.com
Sat Mar 1 20:54:00 EST 2014


On Fri, 28 Feb 2014 22:41:37 +0000 "Damien Miller" wrote:
>Hi,
>
>OpenSSH 6.6 is almost ready for release, so we would appreciate 
>testing on as many platforms and systems as possible. This is a
>small release mostly to fix some minor but annoying bugs in
>openssh-6.5.

>Running the regression tests supplied with Portable OpenSSH does 
>not require installation and is a simply:
>
>$ ./configure && make tests

Hi.

After configure && make tests, sshd defaults to looking for the
system moduli file at /usr/local/etc/moduli. If it doesn't find
it there, the fallback is using dh group-14 (2048-bit modulus).

This is causing the the dhgex.sh test errors reported by Hisashi
& Loganaden (i.e. 3072 != 2048).

To resolve this on my system (where moduli file is at /etc/ssh
/moduli), I use:

$ ./configure && make tests sysconfdir=/etc/ssh

Also, on the system I tested (Slackware Linux), the client logfile
has CRLF line terminators so $gotbits contains a trailing ^M and
the comparison fails. Patch below is one way to fix this: 

--- a/dhgex.sh
+++ b/dhgex.sh
@@ -29,7 +29,7 @@ ssh_test_dhgex()
                fail "$tid unexpected GEX sizes, expected $groupsz, 
got $got"
        fi
        # check what we got (depends on contents of system moduli 
file)
-       gotbits="`awk '/bits set:/{print $4}' ${LOG} | head -1 | 
cut -f2 -d/`"
+       gotbits="`awk '/bits set:/{print $4}' ${LOG} | head -1 | 
cut -f2 -d/ | sed -e 's/\r$//'`"
        if [ "$gotbits" -lt "$bits" ]; then
                fatal "$tid expected $bits bit group, got $gotbits"
        fi

With these two changes, all tests pass on Slackware Linux 14.1.

--mancha



More information about the openssh-unix-dev mailing list