A close-up of a circuit board with glowing blue and gold traces leading to a glowing padlock icon.
Flip over a desktop motherboard and you may find a row of pins for a serial port, a parallel port, or a PS/2 connector. These headers date from the 1990s. Is it laziness, or is something more deliberate going on? XDA Developers argues that licensing is to blame. The primary sources support a narrower story: licensing helped, but it isn't the whole explanation. That story is still worth telling, and it ends somewhere relevant to Windows users: BitLocker and the TPM.

A close-up of a circuit board with glowing blue and gold traces leading to a glowing padlock icon. Where it started: Intel's Low Pin Count bus​

Intel announced the Low Pin Count (LPC) interface specification at its Intel Developer Forum in San Francisco on September 29, 1997. Intel's announcement described a new interface between the chipset and the legacy I/O parts of a motherboard.

Intel's own release lays out the design goals:

  • The legacy parts, typically integrated into a Super I/O chip, would move off the ISA/X-bus onto LPC. Full software compatibility would be kept.
  • LPC would cut pin counts, which would make boards cheaper and easier to design. Intel also said it would use the PCI 33 MHz clock, while ISA ran at 8 MHz.
  • Intel said the interface was software transparent for I/O functions. Existing peripherals and applications would keep working.
  • The LPC 1.0 specification came with an associated reciprocal, royalty-free patent license, available from Intel's website.

That last point is the licensing hook behind XDA's headline. Intel's marketing manager Jan Camps said the company had opened the specification so the industry could adopt it quickly. National Semiconductor and Standard Microsystems said they were planning Super I/O devices around it.

The migration plan was never "keep these forever"​

The headline implies that licensing agreements are the reason old headers survive. Intel's 1997 plan says something different. It described a three-phase migration:

  1. On-board legacy I/O peripherals, such as the floppy controller, parallel port and keyboard controller, move to LPC.
  2. ISA slots are removed, with expansion moving to PCI inside the box and to USB and 1394 outside it.
  3. Legacy serial and parallel ports are removed once native USB and 1394 peripherals become ubiquitous.

So LPC was meant to get the ISA bus out of the way first. The legacy ports were a later phase. The royalty-free license helped LPC spread, and it's plausible that it helped the Super I/O ecosystem settle around it. But I found no source that shows licensing terms are why COM, LPT or PS/2 pins still appear on boards today. That causal claim is XDA's inference. It isn't established fact.

What a Super I/O chip actually does​

The Super I/O chip is the plausible reason these functions linger. It bundles old peripheral interfaces with housekeeping jobs that boards still need.

Nuvoton's current NCT6796D-E product page shows the pattern. The chip monitors power-supply voltages, fan speeds and temperatures, and it supports Nuvoton's SMART FAN algorithms for fan control. For peripherals, it lists a printer port, a keyboard controller, two UARTs and GPIO. It talks to the chipset over either LPC or eSPI.

That makes the engineering case plausible. If the chip you need for fan control and sensors already carries legacy logic, then exposing a few pins costs little. The caveat is that this is one product family. It doesn't show that every modern PC uses the same chip, or that every board exposes the same headers. Check your own motherboard manual for which headers exist and what they support.

The TPM chapter: a real Windows security story​

XDA also says the Trusted Computing Group used LPC for its TPM 1.2 chip. That is broadly right as history: discrete TPMs have commonly hung off LPC, though other buses are also used. Where the article goes wrong is in saying TPM 1.2 "fell out of favour" because it sent everything in plain text. The problem is narrower, and it touches modern Windows setups.

Security researchers have shown that a discrete TPM on the LPC bus can be sniffed:

  • Pulse Security's write-up says that by default, BitLocker-protected OS drives can be unlocked by sniffing the LPC bus. The attacker catches the volume master key when the TPM releases it, then uses it to decrypt the drive.
  • The same write-up demonstrates it against an HP laptop with a TPM 1.2 chip and a Surface Pro 3 with a TPM 2.0 chip. So this isn't only a TPM 1.2 problem.
  • A GitHub list of public BitLocker attacks says Windows Boot Manager talks to the TPM in the clear. It notes that firmware TPMs weren't affected, because there's no separate LPC-connected chip to tap.
  • An academic paper on AMD fTPMs notes that dTPMs are tamper-resistant, but the LPC bus linking them to the CPU is not. It says Microsoft advises a TPM-plus-PIN protector to defend against this class of attack.

Microsoft's own TPM documentation gives useful context. It explains that a TPM communicates with the system over a hardware bus. It also says non-migratable keys never leave the TPM in private form. Yet a sealed key must be released for use at some point, and that release is what the sniffing attacks target. The weakness lies in the exchange and the system design. It isn't that the TPM is a leaky chip.

What this means for Windows users and admins​

The sniffing attacks need physical access and some hardware work. They matter most for lost or stolen laptops, not for remote attackers. Here is the practical guidance the sources support:

  • Use a pre-boot factor. SCRT's write-up says the easiest defense is configuring BitLocker with an extra pre-boot factor such as a PIN. The academic paper cites Microsoft's advice along the same lines.
  • Know which TPM you have. A firmware TPM isn't exposed to this attack, per the BitLocker attack list above. On Windows, tpm.msc shows TPM details, though I haven't verified that it identifies discrete versus firmware TPMs.
  • Don't treat headers as a security feature or a hazard. The legacy pins on your board aren't what exposes the TPM. The exposed element is the bus between a discrete TPM and the chipset.

Where XDA overreaches​

Several claims in the source go beyond what the evidence supports:

  • "Licensing agreements are to blame." The royalty-free license is documented for 1997 and plausibly aided adoption. Nothing I found shows it is why headers remain today.
  • "Every modern-day PC still relies on the Super I/O chip." Nuvoton's page shows these chips still exist and are used. It doesn't show that all PCs depend on one. Some designs use other controllers.
  • "Zero additional cost." Nobody has demonstrated this. Cost is a reasonable guess, but it is a guess.
  • TPM 1.2 "sent everything in unencrypted text." The attacks are real, but they target particular exchanges, such as the key release at boot.

Bottom line​

Intel's LPC bus, backed by a royalty-free patent license, gave the industry a cheap way to move legacy I/O off ISA. Super I/O chips that mix old peripheral interfaces with fan and sensor duties kept those functions around. That is a plausible explanation for the old pins, though not a proven single cause. The most useful takeaway for Windows users is the TPM angle. If your laptop uses a discrete TPM and BitLocker in TPM-only mode, adding a PIN is the documented way to close the LPC sniffing gap.

 

References

  1. Your motherboard still has “useless” headers from the 1990s, and licensing agreements are to blame XDA 2026-10-24T20:30:20+00:00
  2. GitHub - SySS-Research/icestick-lpc-tpm-sniffer: FPGA-based LPC bus sniffing tool for Lattice iCEstick Evaluation Kit · GitHub github.com
  3. Intel Corporation Announces New Low Pin Count (LPC) Interface Specification To Facilitate Industry Legacy Migration Toward Future ISA-Less Systems intel.com