How is linux loaded into a closed platform that has been hacked?

Question

Today I have a question which may be a little uncertain. How would you "port" the Linux OS to a new platform? For example; The nintendo 3ds or the PS4, are consoles which have been recently "hacked", this means you can run your own unsigned code. So basically, hackers usually port Linux to demonstrate they have full control over the software (usually this is a "Live" version of linux since they still can't sign it so even though they can write it to the NAND the bootloader won't load it, since it only runs signed code.

I guess, I may be wrong, they just get a very stripped, linux kernel source code, modify things like the video output (since PC output is completely different from those consoles obviously) and compile it with a cross-compiler for that platform (well the same one they use to compile their programs), and they create some kind of "loader" which will load the OS into memory. Actually I'm just making assumptions from what I've seen, so obviously I'm blind in lots of things.

Your help is greatly appreciated

Cheers, Pedro.

Answer 1

Your question could be split into two:

1. How do people load a new OS on a closed platform that has been "hacked"?
2. How is Linux ported to new systems? (your actual title)

Answers below:

1. In order to run custom OS on a closed device, you need to somehow execute a loader code on a highest possible privilege level. What is a loader? Loader is a small program that performs initial CPU and peripherals configuration, copies some from an external memory device into memory and passes control to it. In case of Linux on game consoles, loader must copy a Linux image to memory, set boot parameters (boot device, available memory ranges — varies with the platform) and run the kernel. Usually loaders operate in privileged mode. On most CPUs with memory protection it's only enabled by OS kernel after the loader has already run.

2. The process of porting Linux can be different depending on whether or not the processor is already supported by the kernel. If the processor is not supported, you'll need to start from the basics — implement base functions such as memory setup, context save/restore, hardware timer support and so on.

   If the processor is supported and there is a Linux kernel for a similar device, the process is a bit easier since it's likely you'll only need to implement a specific platform support — the device you plan to run Linux on likely has a power management controller, various low-speed buses (I2C, SPI) connected to sensors and controllers. Modern Linux kernels use device trees to describe the configuration of the specific platform. Device trees provide a way to instantiate existing device drivers and connect them to a specific hardware configuration.