trust-dns and trust-dns-server are vulnerable to remotely triggered denial-of-service attacks, consuming both network and CPU resources.
DNS messages with the QR=1 bit set are responded to with a FormErr
response.
This allows creating a traffic loop, in which these FormErr
responses are sent nonstop between vulnerable servers.
There are two scenarios how this can be exploited: 1) Create a loop between two instances of trust-dns, consuming network resources, or 2) consuming the CPU of a single instance.
With two instances A and B an attacker sends a DNS query with a spoofed source IP address to A.
A replies with a FormErr
to B.
Now both servers with ping-pong the message back and forth until by chance the packet is dropped in the network.
Multiple spoofed packets can be sent by the attacker, increasing resource consumption.
A single server can get locked up replying to itself. Same setup as above, but now A sends the reply to itself. The packet is sent out as fast as the CPU and network stack manage. This locks up a CPU core. Multiple packets from the attacker consume multiple CPU cores.
{ "nvd_published_at": null, "cwe_ids": [], "severity": "MODERATE", "github_reviewed": true, "github_reviewed_at": "2023-06-06T02:06:23Z" }