[Bug 3489] New: "MaxStartups 0" passes "sshd -t", but configuration reload dies the server
bugzilla-daemon at mindrot.org
bugzilla-daemon at mindrot.org
Mon Oct 24 06:56:52 AEDT 2022
https://bugzilla.mindrot.org/show_bug.cgi?id=3489
Bug ID: 3489
Summary: "MaxStartups 0" passes "sshd -t", but configuration
reload dies the server
Product: Portable OpenSSH
Version: 9.1p1
Hardware: Other
OS: Linux
Status: NEW
Severity: enhancement
Priority: P5
Component: sshd
Assignee: unassigned-bugs at mindrot.org
Reporter: petr.pisar at atlas.cz
I tried disabling MaxStartups feature (disabling is not documented in a
manual) in openssh-9.1_p1 with setting:
MaxStartups 0
"sshd -t" command did not report any errors. So I sent SIGHUP to the
deamon to reload the configuration, and then I tried to ssh-login to
the machine and I found I cannot connect it.
I assumed 0 was interpreted as no connections allowed, thus I reverted
the change in sshd configuration file, sent SIGHUP signal again, and my
init (OpenRC) system reported the daemon is not running. It was obvious
the sshd daemon "crashed".
A system log contains:
Oct 23 21:18:48 router start-stop-daemon[23658]: Will stop PID 3640
Oct 23 21:18:48 router start-stop-daemon[23658]: Sending signal 1 to
PID 3640
Oct 23 21:18:48 router sshd[3640]: Received SIGHUP; restarting.
Oct 23 21:18:49 router sshd[3640]: Server listening on 0.0.0.0 port 22.
Oct 23 21:18:49 router sshd[3640]: Server listening on :: port 22.
Oct 23 21:18:49 router sshd[3640]: fatal: xcalloc: zero size
Oct 23 21:19:14 router start-stop-daemon[23691]: Will stop PID 3640
Oct 23 21:19:14 router /etc/init.d/sshd[23691]: start-stop-daemon: no
matching processes found
The "fatal: xcalloc: zero size" message explains the "crash". The
server attempted to allocate a memory for 0 MaxStartups connections, 0
was rejected as in invalid size by an allocator and the daemon
terminated.
Could you please patch "sshd -t" code path to reject "MaxStartups 0"
configuration line?
--
You are receiving this mail because:
You are watching the assignee of the bug.
More information about the openssh-bugs
mailing list