Skip to content

Latest commit

 

History

History
32 lines (21 loc) · 2.05 KB

boot.md

File metadata and controls

32 lines (21 loc) · 2.05 KB
Raw

DOCS

Bootloaders

The boot process of Prickly Pear Kernel could be divided into 2 parts: bootloader(prekernel) and kernel. These parts are tight together and share the same ABI to prepare the required environment for the kernel.

Kernel boot requirements

  • VM is on.
  • Filled boot args.
  • All kernel data required for a kernel bootup (e.g. the kernel itself, boot args and vm translation tables) should be a continuous region of memory (both physical and virtual). This region is described with boot_args->vaddr, boot_args->paddr and boot_args->kernel_data_size.
  • After kernel data 4mb should be mapped for a successful kernel boot process. This part is used for kernel structs (e.g physical allocation table) which are required before the kernel can set up its own vm translation tables.
  • Initial debug devices should be mapped (UART and/or framebuffer).
  • For 64-bit kernels all RAM should be mapped to itself.

Prekernel

The kernel is an ELF file but there are some use cases when it is needed a raw binary image or an UEFI-capable image. In order to prepare the required environment for a kernel bootup there is a part called prekernel. It is a position independent raw binary which contains code to boot the kernel.

Raw image consists of prekernel, kernel and devtree.

Bootloaders

An alternative to prekernel part is a standalone bootloader which could prepare the required environment for the kernel.

Dirs