Page 1 of 1
[GRUB] Error 7: Loading below 1MB is not supported
Posted: Fri Mar 07, 2008 11:40 am
by Paw
Dear board members,
after four hours of debugging attempts, I gave up and decided to register here in order to ask you for help for my current "pet OS" problem.
My Kernel's bootloader is based on JamesM's bootloader tutorial, using GRUB. The Kernel itself is written in C.
After adding a few minor functions to the kernel code, the Kernel wouldn't be recognized by GRUB anymore.
Due to the increasing code size, the multiboot loader header was shifted out of reach for GRUB. So I modified the boot loader by adding a section called .mboot:
Code: Select all
MBOOT_PAGE_ALIGN equ 1<<0 ; Load kernel and modules on a page boundary
MBOOT_MEM_INFO equ 1<<1 ; Provide your kernel with memory info
MBOOT_HEADER_MAGIC equ 0x1BADB002 ; Multiboot Magic value
; NOTE: We do not use MBOOT_AOUT_KLUDGE. It means that GRUB does not
; pass us a symbol table.
MBOOT_HEADER_FLAGS equ MBOOT_PAGE_ALIGN | MBOOT_MEM_INFO
MBOOT_CHECKSUM equ -(MBOOT_HEADER_MAGIC + MBOOT_HEADER_FLAGS)
[BITS 32] ; All instructions should be 32-bit.
[GLOBAL mboot] ; Make 'mboot' accessible from C.
[EXTERN code] ; Start of the '.text' section.
[EXTERN bss] ; Start of the .bss section.
[EXTERN end] ; End of the last loadable section.
SECTION .mboot ALIGN 4
mboot:
dd MBOOT_HEADER_MAGIC ; GRUB will search for this value on each
; 4-byte boundary in your kernel file
dd MBOOT_HEADER_FLAGS ; How GRUB should load your file / settings
dd MBOOT_CHECKSUM ; To ensure that the above values are correct
dd mboot ; Location of this descriptor
dd code ; Start of kernel '.text' (code) section.
dd bss ; End of kernel '.data' section.
dd end ; End of kernel.
dd start ; Kernel entry point (initial EIP).
[GLOBAL start] ; Kernel entry point.
[EXTERN kmain] ; This is the entry point of our C code
SECTION .text ALIGN 4096
start:
push ebx ; Load multiboot header location
; Execute the kernel:
cli ; Disable interrupts.
call kmain ; call our main() function.
jmp $ ; Enter an infinite loop, to stop the processor
; executing whatever rubbish is in the memory
; after our kernel!
And I modified my linker script link.ld accordingly, so the header information is reachable again, for GRUB.
Code: Select all
ENTRY(start)
phys = 0x00100000;
SECTIONS {
start = .;
.text ALIGN(4096) : AT(phys) {
code = .;
*(.mboot)
*(.text)
*(.rodata)
}
.data ALIGN(4096) : {
data = .;
*(.data)
}
.bss ALIGN(4096) : {
bss = .;
*(.bss)
}
end = .;
}
After these modifications, however, I get following GRUB error message:
Code: Select all
Error 7: Loading below 1MB is not supported
I'm pretty much stuck now, since I cannot see which section is about to be loaded below 1MB.
Here's the objdump output:
Code: Select all
kernel.bin: file format elf32-i386
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00001320 00000000 00100000 00001000 2**4
CONTENTS, ALLOC, LOAD, READONLY, CODE
1 .data 0000000c 00002000 00102000 00003000 2**2
CONTENTS, ALLOC, LOAD, DATA
2 .comment 00000174 00000000 00000000 0000300c 2**0
CONTENTS, READONLY
I'd appreciate a helping hand here, since solutions to similar problems on this board unfortunately don't seem to apply to my situation
Many thanks in advance!
Paw
Posted: Fri Mar 07, 2008 12:36 pm
by jzgriffin
Try this:
Code: Select all
MBOOT_PAGE_ALIGN equ 1<<0 ; Load kernel and modules on a page boundary
MBOOT_MEM_INFO equ 1<<1 ; Provide your kernel with memory info
MBOOT_HEADER_MAGIC equ 0x1BADB002 ; Multiboot Magic value
; NOTE: We do not use MBOOT_AOUT_KLUDGE. It means that GRUB does not
; pass us a symbol table.
MBOOT_HEADER_FLAGS equ MBOOT_PAGE_ALIGN | MBOOT_MEM_INFO
MBOOT_CHECKSUM equ -(MBOOT_HEADER_MAGIC + MBOOT_HEADER_FLAGS)
[BITS 32] ; All instructions should be 32-bit.
[EXTERN code] ; Start of the '.text' section.
[EXTERN bss] ; Start of the .bss section.
[EXTERN end] ; End of the last loadable section.
[GLOBAL start]
start:
jmp start2
align 4
mboot:
dd MBOOT_HEADER_MAGIC ; GRUB will search for this value on each
; 4-byte boundary in your kernel file
dd MBOOT_HEADER_FLAGS ; How GRUB should load your file / settings
dd MBOOT_CHECKSUM ; To ensure that the above values are correct
dd mboot ; Location of this descriptor
dd code ; Start of kernel '.text' (code) section.
dd bss ; End of kernel '.data' section.
dd end ; End of kernel.
dd start ; Kernel entry point (initial EIP).
[EXTERN kmain] ; This is the entry point of our C code
start2:
push ebx ; Load multiboot header location
; Execute the kernel:
cli ; Disable interrupts.
call kmain ; call our main() function.
jmp $ ; Enter an infinite loop, to stop the processor
; executing whatever rubbish is in the memory
; after our kernel!
You shouldn't need a whole section for the multiboot header.
Edit:
Code: Select all
ENTRY(start)
phys = 0x00100000;
SECTIONS {
start = .;
.text 0x00100000 : {
code = .;
*(.text)
. = ALIGN(4096);
}
.data : {
data = .;
*(.data)
. = ALIGN(4096);
}
.bss : {
bss = .;
*(.bss)
. = ALIGN(4096);
}
end = .;
}
Posted: Fri Mar 07, 2008 12:48 pm
by Paw
Thanks Jeremiah, for the suggestion. I tried your code, but unfortunately, the former GRUB problem re-appears:
Code: Select all
Error 13: Invalid or unsupported executable format
And mbcheck returns following message:
I think I'm going to give some more context. Here is my Makefile for building the kernel.bin image:
Code: Select all
CSOURCES := $(shell find . -name "*.c")
ASOURCES := $(shell find . -name "*.s")
SOURCES := $(CSOURCES) $(ASOURCES)
OBJFILES := $(patsubst %.c,%.o,$(CSOURCES)) $(patsubst %.s,%.o,$(ASOURCES))
DEPFILES := $(patsubst %.c,%.d,$(SRCFILES))
CC := gcc
AS := nasm
CFLAGS := -Wall -Werror -nostdlib -nostartfiles -nodefaultlibs -nostdinc -fno-builtin -fno-stack-protector
LDFLAGS := -Tlink.ld
ASFLAGS := -felf
all: kernel
kernel: $(OBJFILES)
@ld $(LDFLAGS) -o kernel.bin $(OBJFILES)
%.o: %.c Makefile
@$(CC) $(CFLAGS) -c $< -o $@
%.o: %.s Makefile
@$(AS) $(ASFLAGS) -o $@ $<
clean:
-@for file in $(OBJFILES) $(DEPFILES) kernel.bin; do if [ -f $$file ]; then rm $$file; fi; \
done
Posted: Fri Mar 07, 2008 12:53 pm
by jzgriffin
Code: Select all
OBJFILES := $(patsubst %.s,%.o,$(ASOURCES)) $(patsubst %.c,%.o,$(CSOURCES))
You need to link the .o file containing start and mboot before your C objects.
Posted: Fri Mar 07, 2008 1:03 pm
by Paw
Jeremiah Griffin wrote:Code: Select all
OBJFILES := $(patsubst %.s,%.o,$(ASOURCES)) $(patsubst %.c,%.o,$(CSOURCES))
You need to link the .o file containing start and mboot before your C objects.
Ouch.... hehe, yeah that sounds reasonable. Unfortunately it still does not work. It's the "below 1MB error" again. Something appears to be pretty screwed here :-/
I'll dive a little deeper and have a look at the GRUB multiboot header specification.
Posted: Fri Mar 07, 2008 1:17 pm
by jzgriffin
What does objdump say, now?
Also, here's my Startup.asm and System.ld files (C++ crap removed):
Code: Select all
OUTPUT_FORMAT("elf32-i386")
ENTRY(Startup)
SECTIONS
{
.text 0x0100000 :
{
code = .;
*(.text)
. = ALIGN(4096);
}
.data :
{
data = .;
*(.data)
. = ALIGN(4096);
}
.bss :
{
bss = .;
*(.bss)
. = ALIGN(4096);
}
end = .;
}
Code: Select all
bits 32
global Startup
Startup:
mov esp, Stack
jmp Startup2
align 4
Multiboot:
MULTIBOOT_PAGE_ALIGN equ 1<<0
MULTIBOOT_MEMORY_INFO equ 1<<1
MULTIBOOT_HEADER_MAGIC equ 0x1BADB002
MULTIBOOT_HEADER_FLAGS equ MULTIBOOT_PAGE_ALIGN | MULTIBOOT_MEMORY_INFO
CHECKSUM equ -(MULTIBOOT_HEADER_MAGIC + MULTIBOOT_HEADER_FLAGS)
extern code, bss, end
dd MULTIBOOT_HEADER_MAGIC
dd MULTIBOOT_HEADER_FLAGS
dd CHECKSUM
dd Multiboot
dd code
dd bss
dd end
dd Startup
Startup2:
extern Setup
cli
call Setup
cli
hlt
section .bss
resb 8192
Stack:
Maybe that will help you.
Posted: Fri Mar 07, 2008 1:34 pm
by JamesM
Hi,
If you look at your objdump output you'll see that the VMA != LMA, as in the virtual address the section is to be loaded starts at zero.
I'm not sure what's causing this, and it's different to my own kernel (which is what caused me to flag up the problem in the first place) - but it's got to be something in the link stage.
The only difference between my link script and yours is I don't use the AT() syntax, mine is pretty much identical to Jeremiah's. You should try that, and see if it resolves the problem.
Cheers,
James
Posted: Fri Mar 07, 2008 1:37 pm
by Paw
Thanks for your patience. I took over your boot loader and linker script and modified it to my needs. The "under 1MB problem" still stays, though.
objdump tells:
Code: Select all
kernel.bin: file format elf32-i386
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00002000 00100000 00100000 00001000 2**4
CONTENTS, ALLOC, LOAD, READONLY, CODE
1 .data 00001000 00102000 00102000 00003000 2**2
CONTENTS, ALLOC, LOAD, DATA
2 .bss 00002000 00103000 00103000 00004000 2**2
ALLOC
3 .comment 00000174 00000000 00000000 00004000 2**0
CONTENTS, READONLY
I suspect that the problem is somewhere else. The strange thing is, that everything worked fine yesterday. I just added a C code stub for a stack datastructure, which doesn't contain anything. Since then the kernel is broken...
EDIT:
@JamesM: Thanks for the hint. Very strange, that my latest objdump gives identical VMA/LMA values... Ah... what a bad start in OS deving
Posted: Fri Mar 07, 2008 1:49 pm
by jzgriffin
Objdump of my kernel for comparison:
Code: Select all
System.sys: file format elf32-i386
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00002000 00100000 00100000 00001000 2**4
CONTENTS, ALLOC, LOAD, READONLY, CODE
1 .rodata 00000126 00102000 00102000 00003000 2**2
CONTENTS, ALLOC, LOAD, READONLY, DATA
2 .data 00000ed8 00102128 00102128 00003128 2**2
CONTENTS, ALLOC, LOAD, DATA
3 .bss 00003000 00103000 00103000 00004000 2**5
ALLOC
4 .comment 0000010c 00000000 00000000 00004000 2**0
CONTENTS, READONLY
Our .text sections are the same, and mine boots fine. Where did you get your GRUB?
Posted: Fri Mar 07, 2008 2:07 pm
by Paw
Jeremiah Griffin wrote:Where did you get your GRUB?
I'm using the "floppy.img" file from James' tutorial, which has GRUB preinstalled. Unfortunately I don't know which version it is.
Posted: Fri Mar 07, 2008 2:20 pm
by jzgriffin
Would it be possible for you to post all of the code? You know, so that we could get a better look at it, run it through our own compilers, etc. It's fine if you don't want to.
Posted: Fri Mar 07, 2008 2:46 pm
by Paw
Jeremiah Griffin wrote:Would it be possible for you to post all of the code? You know, so that we could get a better look at it, run it through our own compilers, etc. It's fine if you don't want to.
This must be one of the most helpful forums on the Net
I added a tar package as an attachement to this post.
Posted: Fri Mar 07, 2008 3:08 pm
by jzgriffin
Okay, cool. I haven't fixed it yet but I'll keep working on it. Spotted a problem with your linker script, though. It should be:
Code: Select all
OUTPUT_FORMAT("elf32-i386")
ENTRY(Startup)
SECTIONS {
.text 0x0100000 : {
code = .;
*(.text)
. = ALIGN(4096);
}
.data : {
data = .;
*(.data)
. = ALIGN(4096);
}
.bss : {
bss = .;
*(.bss)
. = ALIGN(4096);
}
end = .;
}
Posted: Fri Mar 07, 2008 3:19 pm
by Paw
Jeremiah Griffin wrote:Okay, cool. I haven't fixed it yet but I'll keep working on it. Spotted a problem with your linker script, though. It should be:
Code: Select all
OUTPUT_FORMAT("elf32-i386")
ENTRY(Startup)
SECTIONS {
.text 0x0100000 : {
code = .;
*(.text)
. = ALIGN(4096);
}
.data : {
data = .;
*(.data)
. = ALIGN(4096);
}
.bss : {
bss = .;
*(.bss)
. = ALIGN(4096);
}
end = .;
}
It
works again! Pew... what a relieve. Thank you very much for your help
Posted: Fri Mar 07, 2008 3:24 pm
by jzgriffin
You're welcome. It seemed like the right fix, but it wouldn't boot for me. Maybe I forgot to ./update_image.sh.
