BadUSB Exposure

h1. Input/HID devices

h2. Disassembled devices

h3. Truly Ergonomic keyboard 

* Product page: https://trulyergonomic.com/store/products

* Architecture: unknown [TODO, need Windows to run .exe]

* has a dip switch to allow/disallow firmware update

h3. Apple USB Mighty Mouse Model-No. A1152

* Chip: Cypress cy7c63743

* Contains EPROM memory, can only be written once

* => %{color:green}Not vulnerable%

h3. Logitech RX250 optical mouse

* Chip: Cypress cy7c63813

* Product page: http://www.cypress.com/?mpn=CY7C63813-PXC

* Datasheet: http://www.cypress.com/?docID=41007

* M8C core (a simple 8 bit microcontroller core from Cypress)

* 8 KiB flash memory, 256 bytes Ram, 24 MHz clock

* Reprogrammable via USB bootloader

* Documentation is available from Cypress, it should not be too difficult to write a malicious firmware upgrade

* Device should be able to act as a HID keyboard entering a predefined keystroke sequence, e.g. to download a Powershell script from the Internet.

<pre>

Bootloader information:

http://www.cypress.com/?rID=12994

The term "user code" refers to the actual firmware of the device

providing the intended functionality such as a USB HID mouse.

On powerup, the bootloader verifies a 16 bit checksum of the user code.

If it matches, it jumps to the user code.

If it does not match, the device goes to bootloader mode and

communicates with the computer via USB. There are commands for reading

and writing the flash contents.

Unfortunately, the bootloder requires an 8 byte bootloader key. However,

the key verification is done on a byte-by-byte basis (assembly listings

are available in the ZIP file from Cypress) and so it could be

incrementally guessed by counting the number of clock cycles until the

verification fails. It is likely that the bootloader key is equal for a

large number of produced units and so it would probably be enough to

extract it in a lab setup for a few units.

In a lab setup, it is probably possible to make the flash checksum

verification fail e.g. via voltage glitching, clock glitching, extreme

temperatures or UV/X-Ray radiation so that the device boots into

bootloader code. Then the bootloader key can be extracted by guessing

bytes and counting the number of clock cycles the verification takes.

After that, it should be possible to extract the firmware binary for

reverse engineering. Once the firmware is available, it may be possible

to find a hidden command which allows switching the device to bootloader

mode via a special USB command (so that other identical devices can be

reprogrammed via USB).

*Update:* It looks like the controller itself has a proprietary non-USB 

programming protocol. The USB bootloader from http://www.cypress.com/?rID=12994 

is optional and I do not know how many actual devices come shipped with a 

USB bootloader at all.

The integrated programming functionality can be accessed with a programmer,

which is available for 30$ from Cypress:

http://www.cypress.com/?rID=37459

The following document describes the update process:

http://www.cypress.com/?docID=19520

If there is no bootloader, the chips can still be reflashed via the USB contacts

using a custom (non-USB) protocol with a MiniProg programming adapter.

However, the controller does have flash protection fuses. I do not know

whether these fuses are set for typical low cost USB devices.

*Update 20141107:*

I have tried to read out the chip with a Cypress Miniprog adapter. Unfortunately,

the flash protection fuses are set and I could only read one 64-byte block of the

flash memory. Since I cannot dump the firmware, I can't tell whether there is a 

USB bootloader on the chip or not.

</pre>

h3. USB Mouse Tchibo

* ApexOne A2624D, Chip is sold as an ASIC just for USB mouse application

* datasheet does not indicate that there is any microcontroller.

=> %{color:green}Most likely not vulnerable%

h3. USB Laser Mouse Generalkeys

* Chip-on-Board, no label

h3. USB Mouse Logilink

* Chip without label, could be cypress like in the Logitech mouse

h3. Noname USB numeric Keypad

* Chip-on-Board, no label

h3. Hama mini USB mouse

* USB + Sensor integrated into one package, Label: A1198 TSP635B

* http://nutsandboltsandflyingsparks.blogspot.de/2012_07_01_archive.html => Pin Layout, no more info found

* Bus 001 Device 011: ID 062a:0003 Creative Labs

h3. Noname mini optical mouse

* 1bcf:0007 Sunplus Innovation Technology Inc. Optical Mouse => Datasheet for Sunplus devices mention that it contains a µC, but no indication about upgradeability or any persistent storage found

* Chip label: C2165 => Datasheet available 6502 µC [http://en.wikipedia.org/wiki/MOS_Technology_6502]

* No leaked tools

h3. Microsoft Comfort 2000 keyboard v1.0

* Chip-on-Board, no label

h3. Speedlink SL-6535-BK Game pad controller (Mediamarkt 20141106)

* Chip-on-Board without label

* Bus 001 Device 012: ID 0079:0006 DragonRise Inc. Generic USB Joystick

* No further info found

h3. Speedlink ACUTE Presenter (Mediamarkt 20141106)

* Receiver: SL-6198-RRBK  433.92 Mhz

* Chip1: Missing label, 8 pins SOIC, connected to USB

* Chip2: 4608\n1320, connected to antenna

* Bus 001 Device 014: ID 1223:3f07 SKYCABLE ENTERPRISE. CO., LTD. 

* No further info found

h3. Logitech G5 mouse

* Was sold from around 2007-20012 for about 35-60 Euro

* Official firmware upgrade from Logitech available: G5Update12.exe

* Contains large text area with hex format, similar to ihex

* Contains valid USB Descriptors and interesting strings after decoding: "D:\Project\Mecha\FW_Current Version\bin\jw32.abs","ICP"

* Could be MC68HC908JW32 => architecture M68HC05

* %{color:red}=> Most likely vulnerable%

h3. Logitech G502 Proteus Core Gaming Mouse (launched in 2014, current price: 65 Euro)

* Official software (Logitech Gaming Software) contains firmware update utility: G502Update_v16.exe

* Contains area valid USB Descriptors (Device, Configuration and String descriptors)

* http://pclab.pl/art57551-7.html

* PCB Shots show chip label: ARM STM32L100\nR8T6 => STM32L100R8

* ARM Microcontroller with USB, 64 KiB internal flash, 2 KiB eeprom and 8 KiB RAM

=> STM documentation show that the controller does support DFU (Device firmware upgrade) standard

* %{color:red}=> Most likely vulnerable%