Bukva & Khaelen
Khaelen Khaelen
Found a 1974 ARPANET packet header that still validates on a 2024 router. The checksum algorithm was a simple XOR, but it was used to bypass security checks in early systems. It's a textbook example of how weak design propagates. You catalog stories like these? Let me know if there's an obscure hack from the same era you want to add to your shelf.
Bukva Bukva
That’s a neat one—ARPANET still talking to a 2024 router, and a checksum that’s just a XOR. I’ve got a few others up my sleeve, like the BBN worm that exploited the "SND" command to spread itself via the early network mail. Or the odd little trick where a user on a DEC PDP‑10 would drop a null character into a command prompt to bypass a security check on a NCP server. If you want the full backstory, just say the word.
Khaelen Khaelen
The BBN worm was a small, self‑replicating payload that lived in the BBN mainframe’s mail system. When a user typed “SND <user> <msg>” the system would send a message to that user’s mailbox. The worm hijacked that command, appending a malicious stub to the <msg> field. Every time the stub was read, it executed, sending a copy of itself to every user listed in the mailbox’s recipient list. Because the early ARPANET mail had no authentication or attachment filtering, the worm spread unchecked across the network until a patch was issued that stripped the “SND” command’s payload. The PDP‑10 trick was a classic buffer overflow exploit. In the NCP (Network Control Protocol) server on the DEC system, the command line parser did not check the length of the user input. An attacker would type a null character (ASCII 0) into the command line, effectively truncating the string at that point. The NCP server would then read past the null, treating the following bytes as part of the command buffer. This allowed the attacker to overwrite return addresses on the stack and redirect execution to shellcode that bypassed authentication checks. The exploit worked because the server compiled with no stack canaries or bounds checking, typical of early DEC software.
Bukva Bukva
Nice rundown. There’s also that little “circuit‑breaker” hack from the mid‑70s: someone on a PDP‑11 punched a sequence of control‑C characters into the telnet session to flood the terminal driver with interrupts. The kernel would spin on the interrupt handler, effectively DoS’ing the machine without any network traffic. That one made the early system designers add a simple counter‑measures flag. If you want the exact command sequence, I’ve got it in my notes.
Khaelen Khaelen
The control‑C flood was a classic buffer‑overflow on the terminal driver. Each Ctrl‑C (0x03) caused the kernel to trigger an interrupt, and the driver was written to immediately re‑enter the interrupt routine without decrementing a counter. So a burst of 255 Ctrl‑C characters would exhaust the interrupt queue, forcing the kernel to spend all its cycles servicing interrupts and never get back to user mode. The designers responded by adding a watchdog flag that counted pending interrupts and throttled the handler once a threshold was reached. I already have the exact byte sequence in my log, but you’re welcome to let me know if you want to review it.
Bukva Bukva
Sounds like the classic interrupt‑throttling patch. Got the same sequence in my archive, just a note in the “terminal driver hacks” box. If you want to cross‑reference it with the early NCP overflow, let me know—two old tricks on one shelf is always more fun.
Khaelen Khaelen
Cross‑reference queued. Pulling up the NCP overflow log and aligning it with the Ctrl‑C flood bytes. I’ll log the overlap in the archive.
Bukva Bukva
Sounds good, just ping me if you hit any snags while aligning the two.