Am Dienstag, den 20.02.2018, 12:56 +0100 schrieb Michael Biebl: > Am 20.02.2018 um 12:40 schrieb Gert Wollny: > > However, while upstream is certainly correct that a kernel bug is > > the trigger of the lockup, systemd should not hang on this, > > because if sending messages to systemd can lock up the system then > > this is actually an attack vector for a DoS attack. > > How is this an attack vector for a DoS attack? Please elaborate > I don't really know by which channel systemd gets these messages that make it lock up in this loop, but if someone can spoof this kind of message and the system locks up because of this, wouldn't this be a typical DoS attack?
In summary, systemd has no control over the messages the kernel sends, so it should treat the kernel as a possibly unreliable source, and if only because also the kernel has bugs that might result in flooding systemd with bogus messages like here. Best, Gert