The switch to 64-bit mode allows us to use 48-bit addressing and to
relocate the kernel to virtual address 0xffff800000000000 + 1M. The
actual kernel is loaded by the bootloader at physical address 1M.
The rt0 code has been split in two parts. The 32-bit part provides the
entrypoint that the bootloader jumps to after loading the kernel. Its
purpose is to make sure that:
- the kernel was booted by a multiboot-compliant bootloader
- the multiboot info structures are copied to a reserved memory block
where they can be accessed after enabling paging
- the CPU meets the minimum requirements for the kernel (CPUID, SSE,
support for long-mode)
Since paging is not enabled when the 32-bit code runs, it needs to
translate all memory addresses it accesses to physical memory addresses
by subtracting PAGE_OFFSET. The 32-bit rt0 code will set up a page table
that identity-maps region: 0 to 8M and region: PAGE_OFFSET to
PAGE_OFFSET+8M. This ensures that when paging gets enabled, we will still
be able to access the kernel using both physical and virtual memory
addresses. After enabling paging, the 32-bit rt0 will jump to a small
64-bit trampoline function that updates the stack pointer to use the
proper virtual address and jumps to the virtual address of the 64-bit
entry point.
The 64-bit entrypoint sets up the minimal g0 structure required by the
go function prologue for stack checks and sets up the FS register to
point to it. The principle is the same as with 32-bit code (a segment
register has the address of a pointer to the active g) with the
difference that in 64-bit mode, the FS register is used instead of GS
and that in order to set its value we need to write to a MSR.
The go compiler uses SSE instructions to optimize some of the generated code. We need to explicitly enable SSE support by manipulating the appropriate CR flags; otherwise the kernel will triple-fault
We still keep the required main func in stub.go to prevent the compiler
from optimizing the code out. We also force the compiler not to inline
the call to kernel.Kmain so we can find the symbol in the generated .o
file.
The makefile provides the following targets:
- kernel
- iso
- run
- gdb
It sniffs the OS type and when running on non-linux hosts it uses
vagrant ssh and runs make with the above targets inside the vagrant box.
The kernel build process consists of the following steps:
1) Compile arch-specific (only x86 for now) assembly files.
2) Run go build -n to obtain the build commands for our kernel. The
makefile sets the build target to 386/linux so that our current rt0
implementation does not need to switch to long-mode.
3) The build commands are then patched to:
- use build/ as the output directory
- force the go linker to use external link mode and to place its output files
(--tmpdir) to the build folder. By forcing external link mode, the go
linker will emit a single go.o file which can be used by ld.
4) We run our own link step and use ld to link the rt0 .o files with the
go.o file and provide a custom linker script to ensure that our
multiboot record is located at the top of the kernel image so that grub
can find it.
The ISO build process sets up a minimal folder structure for building a
bootable ISO (basically the kernel image plus a grub configuration) and
runs grub-mkrescue to produce the ISO file.
Both the run and the gdb targets assume that qemu is installed. The gdb
target starts qemu, attaches the debugger and sets a breakpoint to the
rt0 entrypoint.
