Call for testing: OpenSSH 7.5p1
jjelen at redhat.com
Wed Mar 15 20:36:40 AEDT 2017
On 03/15/2017 01:47 AM, Damien Miller wrote:
> On Tue, 14 Mar 2017, Jakub Jelen wrote:
>> I don't see this option deprecated in current portable master. Still in place.
>> How are we going to be able to switch between Sandbox and just privilege
> We might consider some way of disabling sandboxing (apart from editing
> the source) if there is user demand, but I think developers/packagers
> can figure out sandbox violations pretty easily from dmesg, etc.
That would be very desirable.
The sandbox (at least rlimit if I remember well) had a problems with
running under valgrind.
>> As far as I remember, various people still use this use case to test
>> other SSH implementation (privsep requires running a root, isn't it?).
> No, only setuid to an unprivileged user needs that; the rest of privsep
> including sandboxing is still active if sshd is run as a regular user.
>>> * Fix various fallout and sharp edges caused by removing SSH protocol
>>> 1 support from the server, including the server banner string being
>>> incorrectly terminated with only \n (instead of \r\n), and
>>> confusing error messages from ssh-keyscan bz#2583.
>> I would vouch for this bug get fixed too in the relation to the SSH1 removal:
>> Also this one is a bit confusing:
> Will take a look.
Thanks. One more thing that I noticed:
Build against OpenSSL 1.1.0 fails, because there is no
EVP_R_BN_DECODE_ERROR defined. The last usage was removed >10 years ago
from OpenSSL sources .
I understand that OpenSSL 1.1.0 is not a priority, but I believe it is
not used in other forks either so compatibility patch (attached) would
probably make sense.
The other tests passed so I guess we are good to go from our side.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 441 bytes
Desc: not available
More information about the openssh-unix-dev