summaryrefslogtreecommitdiff
path: root/README.md
diff options
context:
space:
mode:
Diffstat (limited to 'README.md')
-rw-r--r--README.md82
1 files changed, 82 insertions, 0 deletions
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..5af3dcf
--- /dev/null
+++ b/README.md
@@ -0,0 +1,82 @@
+# SADDNS2.0: DNS Cache Poisoning Attack: Resurrections with Side Channels
+
+## Introduction
+**SADDNS2.0** is a tool for launching the **DNS cache poisoning attack**. It infers the ephemeral port number and brute forces the TxID by exploiting ***F**orwarding Information Base(FIB) **N**ext **H**op **E**xception(FNHE)* cache as a side channel.
+
+This is a different side channel cache poisoning attack derived from [SADDNS](https://github.com/seclab-ucr/SADDNS). Most code usage may remain the same.
+
+## How it works
+1. Scan ephemeral ports opened by the resolver.
+2. Brute force TxID.
+
+The side channel leverages the hash table storing fnhe entry as a shared resource (between the **spoofed** and non-spoofed IPs), which controls whether an IP packet should be fragmented or not. This gives the off-path attacker the ability to identify whether previous **spoofed** ICMP fragment needed packets were accepted or not, which further indicates whether the guessed port is correct or not.
+
+The following figure shows the detail of inferring ephemeral ports.
+
+![Off-path port scanning](https://www.saddns.net/attack2.svg)
+
+### Why spoofed IP is still necessary?
+- Compared with [SADDNS](https://github.com/seclab-ucr/SADDNS), SADDNS2.0 uses embedded UDP packet to scan open port and therefore no IP spoofing is needed during the scanning phase.
+- IP spoofing is still required for injecting rogue responses.
+
+## Additional resources
+
+### Publication
+
+[**DNS Cache Poisoning Attack: Resurrections with Side Channels**](https://doi.org/10.1145/3460120.3486219)
+
+Keyu Man, Xin'an Zhou, Zhiyun Qian
+
+*In Proceedings of ACM Conference on Computer and Communications Security (CCS`21), November 15-19, 2021, Virtual Event, Republic of Korea.*
+
+### Website
+
+[**SADDNS**](https://www.saddns.net)
+
+## How to run
+
+### Requirements
+
+- An IP-spoofing-capable host (preferably Linux. Windows is ok but suffers from low performance.).
+- A domain (attacker-controlled name server)
+- Other things needed to make clear:
+ - The resolver to poison (victim resolver)
+ - The domain to poison (victim domain)
+ - *The **victim domain**'s record will be poisoned on the **victim resolver**.*
+
+### Overview
+
+- Determine the attack type (e.g., public or private port, fragment needed or redirect packet as the payload).
+- Guess the seed/key of FNHE hsah table if private port is used.
+- Flood query traffic to mute the name server of the victim domain (see [SADDNS](https://github.com/seclab-ucr/SADDNS) repo for flooding scripts).
+- Run attack program to guess the port number and TxID automatically.
+
+### Steps
+
+1. Compile
+
+ ```go build ucr.edu/SADDNS2.0```(requires ```gopacket``` and ```libpcap```)
+
+2. Seed guessing (only required when probing private ports)
+
+ See the paper for details. ```GuessSeed.go``` provides methods to send out seed guessing packets. ```guessSeed4.c``` implements hash guessing functions to guess the seed.
+
+4. Start flooding
+
+ ```./dns_query.sh &```(requires ```hping3```)
+
+ Please see the comment in the file for usage.
+
+5. Start attacking (flooding is still in progress)
+
+ ```sudo ./saddns [args]```
+
+ Run ```./saddns -h``` for usage.
+
+```attack.sh``` is a sample script for finish the whole PoC (both Step 2 & 3) including the verification of the poisoned result. It's a demonstrative script and please modify the code accordingly (it **won't** run by default).
+
+
+## Questions and issues
+
+Please submit them by opening a new issue.
+