WiFiWart Boots Linux, Moves To Next Design Phase

Over the last few months we’ve been keeping an eye on WiFiWart, an ambitious project to develop a Linux single-board computer (SBC) small enough to fit inside a USB wall charger. Developer [Walker] says the goal is to create an easily concealable “drop box” for penetration testing, giving security researchers a valuable foothold inside a target network from which to preform reconnaissance or launch attacks. Of course, we don’t need to tell Hackaday readers that there’s plenty of other things you can do with such a tiny open hardware Linux SBC.

Today we’re happy to report that [Walker] has gotten the first version of the board booted into Linux, though as you might expect given a project of this complexity, there were a few bumps along the way. From the single missing resistor that caused U-Boot to throw up an error to the finer points of compiling the kernel for an embedded board, the latest blog post he’s written up about his progress provides fascinating insight into the little gotchas of bringing up a SBC from scratch.

Once the board was booted into Linux, [Walker] started testing out different aspects of the system. A memory benchmark confirmed the finicky DDR3 RAM was working as expected, and he was able to load the kernel modules for the dual RTL8188 interfaces and connect to a network. While the two WiFi modules are currently hanging off the board’s full-sized USB ports, they will eventually be integrated into the PCB.

Critically, this prototype board is also allowing [Walker] to get an idea of what the energy consumption of the final hardware might be. Even at full tilt, this larger board doesn’t go over 500 mA at 5 VDC; so if he designs the power supply with a maximum output of 1 A, he should have a nice safety margin. As mentioned in the previous post, the plan is currently to put the PSU on its own board, which will allow more effective use of the charger’s internal volume.

With the software and hardware now largely locked in, [Walker] says his attention will be turned towards getting everything small enough to fit into the final form factor. This will certainly be the most challenging aspect of the project, but with a growing community of hackers and engineers lending their expertise to the cause, we’re confident the WiFiWart will soon be a reality.

12 thoughts on “WiFiWart Boots Linux, Moves To Next Design Phase

  1. I hope the data lines on the USB get man in the middle attacks for whoever is dumb enough to plug their phone into an unknown charger. That’s just as important as the rest of the network equipment. Most hacks are due to human failure. Knowing about your target is invaluable.

    1. From what I can see their starting point was the open hardware of the A33-OLinuXino, saved them time and a lot of effort. They had a known working schematic and a known working software stack as a starting point.

      The other thing is that I checked on mouser, digikey, lcsc, ebay, aliexpress and can’t find the SigmaStar SSD210 chip. And it looks like the SDK is currently under a NDA ( https://mangopi.club/post/2820 ). I can’t even find a basic datasheet for the SigmaStar SSD210, so it could be an amazing chip but if the documentation sucks, it is not much good. All I could find was the product brief: http://comake.online/uploadfile/file/20210127/20210127083050_76347.pdf

      1. >The other thing is that I checked on mouser, digikey, lcsc, ebay,
        >aliexpress and can’t find the SigmaStar SSD210 chip.

        You can’t buy allwinner parts from normal distributors in small quantities either.

        Anyhow the SSD210 is available from taobao.

        Taobao: https://item.taobao.com/item.htm?spm=a230r.1.14.21.6fdb7d1eZCpBpB&id=647699987417&ns=1&abbucket=5#detail

        I have 50 on my desk so they aren’t unobtainium.

        >And it looks like the SDK is currently under a NDA ( https://mangopi.club/post/2820 ).

        Most of the hardware has been reverse engineered and is being pushed to the mainline kernel.
        Another chip in the same lineage, the SSD210, has been reverse engineered to the point ethernet, usb, sd etc work fairly well. Enough of the display hardware works to play doom:

        https://twitter.com/linux_chenxing/status/1415308882918797323

        >I can’t even find a basic datasheet for the SigmaStar SSD210, so it could be an amazing
        >chip but if the documentation sucks, it is not much good.

        This is basically the case for all of these asian ARM SoCs.

        http://linux-chenxing.org/pioneer3/#ssd210

        Anyhow I didn’t say the must use these chips but I was suggesting that if you want to make small linux machines that are ulinux garbage you can now… and because these chips are QFN packages you can do it on JLCPCBs 2 layer service or maybe even a flex PCB. Sure you can’t just download someone else’s premade schematic change it a bit, add their software and then claim you’ve done something earth shattering and you might actually have to but some effort in. That’s part of the fun isn’t it?

      1. Sorry that is the expected price for the SoM when they are eventually available. It is an interesting chip, but the lack of chips that can be bought today and the missing documentation would shift it far down my list of chips to use.

      2. Integrated DDR2 memory means that you don’t need extra memory chip ($). Also it would make the PCB smaller and a lot easier to route. One probably get by on a 2 layer PCB and not need to deal with high speed signals i.e. impedance control that kind of stuff.

Leave a Reply to dmitry GrinbergCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.