1. 12
  1. 4

    I believe you should still be able to build the base system and packages for i586, it just isn’t the default. If you have another FreeBSD system, Poudriere can spit out a disk image (including bootable USB images and so on) for both the base system and any packages that you want, and also build package sets for you, with the default CPU type set to i586.

    1. 3

      What a quirky little system! I hadn’t heard about this processor before.

      Though, if I was in the market for very cheap computing, I’d likely go for a Chromebook with a broken screen. For around $25 USD, that gets you a 64-bit Celeron, 2 to 4GB of RAM and 16GB or more of storage. And perhaps a working keyboard and touchpad. The main trick is to install a new BIOS via Mr Chromebox.

      1. 1

        I’m not following. How broken of a screen are we talking here? A working screen seems like a requirement for a Chromebook.

        1. 6

          Laptops with broken screens in general are a good way of very cheaply buying low-power headless systems. Most generally have some form of video output for the initial install (or, if they don’t, can boot from a USB image with a preconfigured network and an SSH key that you can run an installer from). You get low-power components, a built-in UPS, and a lot of them have a decent wired Ethernet connection (if not, GigE over USB-2 is generally able to do 300 Mb/s pretty reliably) and WiFi.

          1. 3

            Most (all?) Chromebooks also have an HDMI output as well. That DMP eBox-3300 also requires an external display, in that case a RGB monitor.

            I expect that in either case (broken Chromebook, eBox-3300) that these would end up running headless as a server for something.

            The downside of a Chromebook is no built-in Ethernet, though they all have WiFi. It is typical that they will have at least one USB-3 port and a SD card slot, so expanding the storage is quite viable.

        2. 2

          it turns out some Vortex86 CPUs lack support for the Conditional Move (CMOV) instruction, and code targetting i686 will fail on those models

          Heh. Years ago I was trying to get Debian running on a VIA machine. Installing on the machine didn’t succeed for whatever reason, so I tried installing on a laptop and moving the disk to the machine… the installer put i686 on there because the laptop was modern, and the kernel didn’t boot on the VIA machine, explicitly complaining about cmov!

          1. 1

            Interesting that 12.1 was still bootable. Last time I tried, which admittedly was 11-CURRENT, it didn’t boot on 486/586 at all. It was doing some FPU reset instruction in early boot that required 686 (possibly SSE, but I don’t remember exactly). Neat little experiment.

            1. 0

              Off topic, but might I suggest to disable or change syntax highlighting for output and where it’s wrong? It can be a bit distracting. Both the “highlighting” of things between ’ and the “errors” for the package repository config.