How do I compile two different bootloader source code (gcc + nasm) together?

[sanzenyou]

Hello everyone, this topic is the first one I posted on our forum.

I'm writing a simple operating system, that includes a x64 bootloader and a kernel.

Now I have completed the assembly part of the bootloader(.asm), but I need to compile this asm source code with another .c source code together for call c function, I don't know how to do it...

Who can help me, give me some advice? thank youuuu!!!!!!

PS: My platform is Windows 11, NASM + GCC

[nullplan]

Do you even have a 16 bit GCC? Otherwise, even attempting this is going to be miserable. I mean more than it is going to be under all circumstances. I'd suggest to write the entire bootloader in assembler and reserve the C for protected mode or long mode, depending on which way you want to go. Or else, just using GRUB.

If you must attempt this, you can assemble the whole thing into an object format. Write the assembler part with origin 0. Assemble with -f coff. Compile with -c. Then link the whole thing together with -Ttext=0x7800 and ensure that the assembler file is the first on the command line.

[sanzenyou]

Do you even have a 16 bit GCC? Otherwise, even attempting this is going to be miserable. I mean more than it is going to be under all circumstances. I'd suggest to write the entire bootloader in assembler and reserve the C for protected mode or long mode, depending on which way you want to go. Or else, just using GRUB.

If you must attempt this, you can assemble the whole thing into an object format. Write the assembler part with origin 0. Assemble with -f coff. Compile with -c. Then link the whole thing together with -Ttext=0x7800 and ensure that the assembler file is the first on the command line.

thank you, but I see that the xv6 source code can compile two bootloader files together, why?
in bootasm.s, "call bootmain" -- function in bootmain.c

I'm so curious...

[nullplan]

I'm so curious...

But apparently not curious enough to just look it up. From the XV6 Makefile:

Code: Select all

bootblock: bootasm.S bootmain.c
	$(CC) $(CFLAGS) -fno-pic -O -nostdinc -I. -c bootmain.c
	$(CC) $(CFLAGS) -fno-pic -nostdinc -I. -c bootasm.S
	$(LD) $(LDFLAGS) -N -e start -Ttext 0x7C00 -o bootblock.o bootasm.o bootmain.o
	$(OBJDUMP) -S bootblock.o > bootblock.asm
	$(OBJCOPY) -S -O binary -j .text bootblock.o bootblock
	./sign.pl bootblock

OK, so they compile both the assembler and the C source into ELF object files. This works because the assembler code creates a 32-bit environment before calling the C code. They also link it all with -Ttext 0x7c00 (sorry, I had the address wrong earlier), and then they dump the .text section without headers, and that is the boot block.

This is pretty horrible, of course. It does mean that the C they can use is pretty limited. It can't use strings or static data, and if you look into the source files, it hardcodes an expectation to find the kernel file at the start of the first hard disk. I am not convinced that this works; isn't that where the MBR was supposed to be?