so many errors.

[aayanbaig]

my projecet has had 9 resets because i havent been able to make the damn bootloader.

i've decided to rain hell upon myself and make a uefi bootloader in assembly because c is slightly. VERY slightly annoying. dont question it please ill switch to c later, still kinda learning it.


however,
i have one singular file running my uefi x86-64 bootloader:

Code: Select all

section .text
    global _start

_start:
    mov rdi, rdx  ; Pass SystemTable pointer to efi_main
    call efi_main
    hlt           ; Halt the CPU if efi_main returns

section .data
    align 8
    msg db "U",0,"E",0,"F",0,"I",0," ",0,"I",0,"N",0,"I",0,"T",0,"!",0,0,0  ; UTF-16 string
    newline db 13,10,0  ; CRLF (Carriage Return + Line Feed)

section .text
    global efi_main

efi_main:
    push rbx                  
    push rdi
    push rsi

    mov rbx, rdi              ; RBX = SystemTable
    test rbx, rbx             
    jz halt                   ; Ensure SystemTable is valid

    mov rdi, [rbx + 8]        ; Load SystemTable->ConOut
    test rdi, rdi
    jz halt                   ; Ensure ConOut exists

    ; Print ConOut pointer
    mov rsi, conout_msg
    call debug_print
    mov rsi, rdi
    call print_hex

    ; Debug: Print function pointers dynamically
    mov rcx, 0
find_output_string:
    cmp rcx, 40               ; Limit check to avoid bad memory access
    ja halt
    mov rsi, [rdi + rcx]      ; Read function pointer at offset rcx
    call print_hex
    cmp rsi, 0                ; Skip if NULL
    je next_offset

    ; Check if function pointer is valid
    test rsi, 7               ; Must be 8-byte aligned
    jnz next_offset
    mov rdx, 0x100000000      ; Check if address is too high
    cmp rsi, rdx
    ja next_offset

    ; Found a valid function pointer, assume it's OutputString
    mov rax, rsi
    jmp output_string_found

next_offset:
    add rcx, 8
    jmp find_output_string

output_string_found:
    test rax, rax
    jz halt                   ; Ensure OutputString is valid

    ; Print OutputString pointer before calling it
    mov rsi, output_string_msg
    call debug_print
    mov rsi, rax
    call print_hex

    ; Call OutputString safely
    mov rcx, rdi  ; First argument: ConOut pointer
    mov rdx, msg  ; Second argument: Message pointer
    xor r8, r8    ; Third argument must be NULL
    xor r9, r9    ; Fourth argument must be NULL (UEFI convention)
    call rax

    jmp continue

halt:
    hlt                       ; Stop execution on failure

continue:
    pop rsi
    pop rdi
    pop rbx

    ret

debug_print:
    ; Prints a message to the UEFI console
    mov rax, [rdi + 32]  ; Load OutputString function pointer dynamically
    mov rcx, rdi  ; First argument: ConOut
    mov rdx, rsi  ; Second argument: Message
    xor r8, r8    ; Ensure third argument is NULL (UEFI requirement)
    xor r9, r9    ; Fourth argument also NULL
    call rax
    ret

print_hex:
    ; Converts RSI into a hex string and prints it (stub for now)
    ret

validate_function_pointer:
    ; 🚨 Prevents calling garbage memory 🚨
    mov rsi, validating_func_msg
    call debug_print
    mov rsi, rax
    call print_hex

    ; If function pointer is NULL, halt
    test rax, rax
    jz halt

    ; If function pointer is NOT 8-byte aligned, halt
    test rax, 7
    jnz halt

    ; If function pointer is way too high in memory (bad pointer), halt
    mov rdx, 0x100000000
    cmp rax, rdx
    ja halt

    ret

section .data
    system_table_msg db "SystemTable: ", 0
    conout_msg db "ConOut: ", 0
    output_string_msg db "OutputString: ", 0
    validating_func_msg db "Validating function pointer: ", 0

im getting the following error when testing on qemu:
BdsDxe: failed to load Boot0001 "UEFI QEMU DVD-ROM QM00003 " from PciRoot(0x0)/Pci(0x1,0x1)/Ata(Secondary,Master,0x0): Not Found
BdsDxe: loading Boot0002 "UEFI QEMU HARDDISK QM00001 " from PciRoot(0x0)/Pci(0x1,0x1)/Ata(Primary,Master,0x0)
BdsDxe: starting Boot0002 "UEFI QEMU HARDDISK QM00001 " from PciRoot(0x0)/Pci(0x1,0x1)/Ata(Primary,Master,0x0)
!!!! X64 Exception Type - 0E(#PF - Page-Fault) CPU Apic ID - 00000000 !!!!
ExceptionData - 0000000000000000 I:0 R:0 U:0 W:0 P:0 PK:0 SS:0 SGX:0
RIP - 000000001E3D20B1, CS - 0000000000000038, RFLAGS - 0000000000000202
RAX - 000000001E4CF798, RCX - 000000001E4CFF98, RDX - 000000001F9EE018
RBX - 000000001F9EE018, RSP - 000000001FF107F0, RBP - 0000000000000000
RSI - 000000001E3D3027, RDI - 0000007800020046
R8 - 00000000000000AF, R9 - 000000001FF26310, R10 - 000000001FF28EF8
R11 - 00000000000000F0, R12 - 0000000000000000, R13 - 000000001EE79BE0
R14 - 000000001EE787BC, R15 - 000000001E4CFF98
DS - 0000000000000030, ES - 0000000000000030, FS - 0000000000000030
GS - 0000000000000030, SS - 0000000000000030
CR0 - 0000000080010033, CR2 - 0000007800020066, CR3 - 000000001FC01000
CR4 - 0000000000000668, CR8 - 0000000000000000
DR0 - 0000000000000000, DR1 - 0000000000000000, DR2 - 0000000000000000
DR3 - 0000000000000000, DR6 - 00000000FFFF0FF0, DR7 - 0000000000000400
GDTR - 000000001F9DE000 0000000000000047, LDTR - 0000000000000000
IDTR - 000000001F471018 0000000000000FFF, TR - 0000000000000000
FXSAVE_STATE - 000000001FF10450
!!!! Find image based on IP(0x1E3D20B1) (No PDB) (ImageBase=000000001E3D1000, EntryPoint=000000001E3D2000) !!!!



can someone help? been trying loads

[techdude17]

Your image base is 0x1E3D1000 and the fault occurred at 0x1E3D20B1, subtract them to get your function offset of 0x10B1.
Try looking this up in your executable - disassemble at that address and follow the chain

[aayanbaig]

thank ya. ill work on it and lyk

[zaval]

Entry point (and .text section) is at RVA 0x1000 (of course, edited), one page further from the image base, because of the headers, so the faulting instuction offset is 0xb1 (177) bytes relative the entry point.

PS. I didn't even try to check this marvel, only noticed this:

Code: Select all

mov rdi, [rbx + 8]        ; Load SystemTable->ConOut

WHAT? made you think, that at the offset 8 from the system table is ConOut?

[MichaelPetch]

Have to agree with Zaval. The offset of ConOut in the system table is 64 and not 8. See: https://uefi.org/specs/UEFI/2.10/04_EFI ... e.html#id6 . Other things I see are that your offset for OutputString in the ConOut structure is 32 when it should be 8 (see https://uefi.org/specs/UEFI/2.10/12_Pro ... t-protocol ); the stack (RSP) isn't always properly aligned on 16 byte boundary before an EFI function call; the 32 bytes of shadow space required by 64-bit MS ABI is not allocated for each EFI function call; and some of the strings you are using are not UTF16. It looks like you attempted to use two system calling conventions.

With NASM you can define UTF16 Unicode strings from UTF8 with something like:

Code: Select all

conout_msg dw __utf16__("ConOut: "), 0

The code really needs a lot of work, which suggests you may be better off writing it in C. I know you told us not say that, but it is true - you'll go through a whole lot less suffering by not doing it in assembly.

[aayanbaig]

Got it, thank you so much.