Navigating Linux on (jailbroken) PS4 Remotely

14 Nov 2026

I recently attempted and got NixOS working on my PS4 on FW 9.00 through the browser-based (PSFree + Lapse) jailbreak. My PS4 Slim is based on the Baikal southbridge and LLMs helped me integrate its NixOS configuration into my ecosystem. So right now, the configuration is tightly coupled and exists in my dotfiles repository here:

https://github.com/ritiek/dotfiles/tree/main/machines/nixstation

From what I understand, installing Linux into the PS4's internal hard drive has historically been unsupported on the Baikal southbridge, so I run Linux off an external drive connected to the PS4. Newer kernels reportedly support Baikal much better now though, and I'm already on the Strawberry 7.0.8 Baikal kernel but haven't tried an internal install with it yet.

I also plan to get a minimal NixOS configuration for it that'd be decoupled from the rest of my flake, but it'll be a few months before I get physical access to my PS4 again and take an attempt at this.

The more recent jailbreak on 9.00 (the PSFree WebKit exploit chained with the Lapse kernel exploit, which I run through WebKitty) is completely software-based and unlike the previous pOOBs4 kernel exploit, it doesn't require physically plugging the USB flash storage in and out of the PS4. That said, I've noticed my PS4 getting stuck in a frozen state and requiring a hard reboot to recover from, in like 10-15% of all attempts. This requires physical access to the PS4 to hard reboot. I guess I could work around this through a relay switch controllable over Wi-Fi (like the Tasmota relay or the OpenBeken switch I've flashed before) by forcing a hard reboot remotely. I initially wondered if my PS4 being based on the Baikal southbridge could be relevant to this freezing issue, but it seems the Lapse kernel exploit itself isn't fully reliable and failed attempts are known to kernel panic or freeze the console.

I also found out that my PS4 consumes 41-42W power on idle, and jumps to 49W when playing the game I tried on the PlayStation's own OS (heavier games reportedly pull closer to 75-80W on the Slim) or when watching videos on YouTube in Zen browser on NixOS. I'm mainly concerned about the high power use on idle and so I'd prefer it to be off when not in use. The PlayStation's own rest mode would draw only a couple of watts, but that's not an option once the PS4 has booted into Linux.

Powering the PS4 from the wall doesn't automatically boot it up. It also requires a physical tap on the console's power button to boot it or alternatively, pressing the PS button on a DualShock 4 controller previously paired to the PS4 over Bluetooth also boots it up. My PS4 consumes 0.3W when it's connected to the wall adapter while it's awaiting boot. I figured I could simulate this PS button tap from my RPi400, which lives in very close proximity to my PS4 since I connect my RPi400 with PS4 to get keyboard and mouse working with it whenever I have physical access to my PS4. I was also able to cheaply bring this idea to life with the help of LLMs. It seems one needs to spoof the DualShock 4's Bluetooth MAC address and send a connection attempt through this spoofed MAC address to boot the PS4 up, as it wouldn't for me with RPi400 sending out the connection attempt through its hardware MAC address. The LLM came up with this:

# The PS4 wakes from full power-off when its paired DS4 pages it over
# Bluetooth, so borrow the DS4's address briefly and page the PS4.
(writeShellApplication {
  name = "ps4-wake";
  runtimeInputs = [ bluez iputils ];
  text = ''
    # hci0 is re-registered after an address change and btmgmt exits 0
    # even on failure, so retry power on until it reports powered.
    use_addr() {
      btmgmt power off >/dev/null
      btmgmt public-addr "$1" >/dev/null
      for _ in $(seq 20); do
        btmgmt power on >/dev/null 2>&1
        [[ $(btmgmt info 2>/dev/null) == *"current settings: powered"* ]] && return
        sleep 0.25
      done
      echo "hci0 did not come back up as $1" >&2
      return 1
    }
    trap 'use_addr E4:5F:01:15:E9:A9' EXIT
    use_addr 84:30:95:02:28:35  # DS4
    for _ in 1 2 3 4 5; do
      l2ping -c 1 -t 5 28:3A:4D:B1:82:B2 >/dev/null 2>&1 && break  # PS4
    done
    use_addr E4:5F:01:15:E9:A9
    trap - EXIT
    echo "Paged PS4, waiting for it to boot..."
    for _ in $(seq 180); do
      ping -c 1 -W 1 10.13.37.2 >/dev/null 2>&1 && { echo "PS4 is up"; exit 0; }
    done
    echo "PS4 did not come up" >&2
    exit 1
  '';
})

With this, running ps4-wake on my RPi400 works well to boot it up.

I also have the initial Linux boot .bin payload stored on the external drive which I'm able to execute through the menu in GoldHEN v2.4b18.10 (or newer) under Utilities > PayLoader > LaunchPad. I also have my PS4 and RPi400 connected directly to each other through an Ethernet cable. The RPi400 serves its own custom hosted DNS server to block firmware updates and lets me access the PS4 remotely from the outside through a Tailscale subnet route (on my Headscale tailnet) where the PS4 has a static IP. I'm then able to access my PS4 using chiaki-ng and run the jailbreak and have it execute the relevant .bin payload and boot into NixOS. The NixOS then runs its own Tailscale node which lets me cut off the RPi400 hop.

I tried some old GPU-dependent games through Lutris but ran into display and rendering glitches after a couple of minutes of play. Still needs debugging.