Writing bootloader to FAT32 USB problem

[makuc9]

Hello, everyone.

I've written (found & made a few minor adjustments) a simple "Hello world" bootloader, which works if compiled as a img file and used on a virtual machine, but I want to write it to a FAT32 USB flash drive without corrupting it's file system. For this task I tried using HxD hex editor, under menu "Extras" -> "Open disk..." and selecting my USB pen drive by letter. I after that replaced values after SysID (everything after FAT32 index table...or whatever) up to the final boot signature and modified the first 3 bytes accordingly. After saving the changes I removed my USB, inserted it again and everything worked (no FS corrupted). However, by booting "real hardware" from USB (my PC does support that, since I've installed my operating system - Win7,8 - that way several times) it just simply jumps over it and starts booting from my second drive (HDD with wins) bootstart list.
What am I missing? If you want to see the assembly code to check if anything there is corrupted say so.
(Is it possible for this to be caused because I changed in my BIOS setting the way it "reads" disks from default - IDE - to ASCI? I did not want to try to change it because my WIN won't boot up afterwards.)

Sincerely.

[Combuster]

An USB stick is likely to have a partition table, where you'd need to set the bootable bit and maybe add/replace a MBR instead of dealing with just the filesystem. One other thing to check is if the boot signature is present at the correct location.

[egos]

What am I missing? If you want to see the assembly code to check if anything there is corrupted say so.
(Is it possible for this to be caused because I changed in my BIOS setting the way it "reads" disks from default - IDE - to ASCI? I did not want to try to change it because my WIN won't boot up afterwards.)

This option is called AHCI. It has effect only if you are trying to work with SATA device directly (not through BIOS). I think Combuster is right. Check active flag in partition table. If you don't know how to do it then open your USB flash drive as a physical disk in HxD and show hex dump of its first sector (MBR).

[makuc9]

Well, I sure hope we are talking about the same stuff. Anyway. I have edited the first sector (sector 0) on USB. There are a few specifications about FS (like OEM name, sectors per cluster...). Boot record, actually. The first 3 bytes here are JUMP code + nop, then is FAT32 stuff and after that is the place where your bootloader can actually be written (am I correct?). and than at 511 & 512 (last 2 bytes in 512 bytes large boot record) is boot signature.

Code: Select all

BITS    16
ORG      0x7C00
Start:
   jmp      Skip
   nop
   
;      OEM Parameter block, FAT 12/16/32
bOemName      db      "MSDOS4.0"

wBlockSize      dw      0
bSecPerClus      db      0
wReservedSecs   dw      0
bNumFats      db      0
wRootEntries   dw      0
wSectors16      dw      0
bMediaType      db      0
wFATSz16      dw      0
wSecPerTrack   dw      0
wNumHeads      dw      0
lSecCache      dd      0
lSectors32      dd      0

lFATSz32      dd      0
wExtFlags      dw      0
wFSVer         dw      0
lRootCluster   dd      0
wFSInfo         dw      0
wBkBootSec      dw      0
BytesPerCluster   dd      0
CurrentCluster   dd      0
lDataSector      dd      0
bDriveNum      db      0
bReserved      db      0
bBootSig      db      0
lVolSerial      dd      0xDEADF007
szVolLabel      db      "-----------"
bSysID         db      "COPYHERE"

Skip:
   cli
   xor ax, ax
   mov ds, ax
   mov es, ax
   ; Other stuff for bootloader to do
   hlt

   
TIMES 510-($-$$) db 0
DW 0xAA55

I compile it using NASM and use HxD editor (steps explained in my first post) to write it to Sector 0 of USB.

[egos]

And what about MBR?

[iansjack]

A disk does not require an MBR, does it? Although, conventionally, hard disks are partitioned the MBR is - in essence - just a first stage boot loader. What is to stop you having an unpartitioned disk and replacing the MBR by a custom bootloader (i.e. treating the disk, or USB stick, as a large floppy disk)?

[egos]

I'm not sure that TS doesn't have MBR on the flash drive.

[makuc9]

Hmm, what am I doing wrong? Why doesn't that work on my USB drive?

Code: Select all

BITS 16
ORG		0x7C00
Start:
	jmp	Skip
	nop

;****************************************;
; ----- OEM Parameter block, FAT32 ----- ;
;****************************************;

bOemName		db	"MSDOS4.0"	; 8 bytes

; ----- FAT32 Bios Parameter Block ----- ;
wBlockSize		dw	0
bSecPerClus		db	0
wReservedSecs		dw	0
bNumFats		db	0
wRootEntries		dw	0
wSectors16		dw	0
bMediaType		db	0
wFATSz16		dw	0
wSecPerTrack		dw	0
wNumHeads		dw	0
lSecCache		dd	0	; lHiddenSecs
lSectors32		dd	0

;*******************************************;
; ----- Extended BIOS Parameter Block ----- ;
;*******************************************;

lFATSz32		dd	0
wExtFlags		dw	0	; Bits 0-3 = Active FAT, 7 = !FAT
wFSVer			dw	0
lRootCluster		dd	0
wFSInfo			dw	0
wBkBootSec		dw	0
BytesPerCluster		dd	0	;Reserved 1
CurrentCluster		dd	0	;Reserved 2
lDataSector		dd	0	;Reserved 3
bDriveNum		db	0
bReserved		db	0
bBootSig		db	0
lVolSerial		dd	0xDEADF007
szVolLabel		db	"-----------"
bSysID			db	"COPYHERE"

; ----- Booting process pro ----- ;
Skip:
; ----- Adjust registers ----- ;
	cli
	xor ax, ax
	mov ds, ax
	mov es, ax
	mov fs, ax
	mov gs, ax

; ----- Display load message ----- ;
	mov si, msgLoading
	call print_s

	jmp $

	hlt

; ----- Predefined functions ----- ;
print_s:
	lodsb			; get next character
	or al, al		; check, if null-terminator
	jz end			; yes? -> printing done
	mov ah, 0eh		; no? -> print the character
	int 10h			; call BIOS
	jmp print_s		; repeat until null-terminator found

end:			; Return upon function exe completion
	ret

; ----- Data ----- ;
msgLoading db "Will be loading operating system...", 0
   
TIMES 510-($-$$) db 0
DW 0xAA55

Link to the assembled binary file:
https://mega.co.nz/#!KFdGCAQI!CMB95XMmU ... IH-xo-kUyg

Link to the MBR record of rewritten USB FAT32:
https://mega.co.nz/#!zJkAhK5C!fL6UtnLRq ... gsFX8txW50

Any help appreciated. As I said before, all I can achieve is been successfully skipped by bootloader on another disk.

Edit:
If I disconnect all other disks, my PC only stays there and nothing happens (it remains on "Verifynd DMI Pool Data ......" and there is char indicator (you know, that flashing thingy indicating where you are to write next letter) in a new line.

[kohlrak]

Hmm, what am I doing wrong? Why doesn't that work on my USB drive?

Code: Select all

BITS 16
ORG		0x7C00
Start:
	jmp	Skip
	nop

;****************************************;
; ----- OEM Parameter block, FAT32 ----- ;
;****************************************;

bOemName		db	"MSDOS4.0"	; 8 bytes

; ----- FAT32 Bios Parameter Block ----- ;
wBlockSize		dw	0
bSecPerClus		db	0
wReservedSecs		dw	0
bNumFats		db	0
wRootEntries		dw	0
wSectors16		dw	0
bMediaType		db	0
wFATSz16		dw	0
wSecPerTrack		dw	0
wNumHeads		dw	0
lSecCache		dd	0	; lHiddenSecs
lSectors32		dd	0

;*******************************************;
; ----- Extended BIOS Parameter Block ----- ;
;*******************************************;

lFATSz32		dd	0
wExtFlags		dw	0	; Bits 0-3 = Active FAT, 7 = !FAT
wFSVer			dw	0
lRootCluster		dd	0
wFSInfo			dw	0
wBkBootSec		dw	0
BytesPerCluster		dd	0	;Reserved 1
CurrentCluster		dd	0	;Reserved 2
lDataSector		dd	0	;Reserved 3
bDriveNum		db	0
bReserved		db	0
bBootSig		db	0
lVolSerial		dd	0xDEADF007
szVolLabel		db	"-----------"
bSysID			db	"COPYHERE"

; ----- Booting process pro ----- ;
Skip:
; ----- Adjust registers ----- ;
	cli
	xor ax, ax
	mov ds, ax
	mov es, ax
	mov fs, ax
	mov gs, ax

; ----- Display load message ----- ;
	mov si, msgLoading
	call print_s

	jmp $

	hlt

; ----- Predefined functions ----- ;
print_s:
	lodsb			; get next character
	or al, al		; check, if null-terminator
	jz end			; yes? -> printing done
	mov ah, 0eh		; no? -> print the character
	int 10h			; call BIOS
	jmp print_s		; repeat until null-terminator found

end:			; Return upon function exe completion
	ret

; ----- Data ----- ;
msgLoading db "Will be loading operating system...", 0
   
TIMES 510-($-$$) db 0
DW 0xAA55

Link to the assembled binary file:
https://mega.co.nz/#!KFdGCAQI!CMB95XMmU ... IH-xo-kUyg

Link to the MBR record of rewritten USB FAT32:
https://mega.co.nz/#!zJkAhK5C!fL6UtnLRq ... gsFX8txW50

Any help appreciated. As I said before, all I can achieve is been successfully skipped by bootloader on another disk.

Edit:
If I disconnect all other disks, my PC only stays there and nothing happens (it remains on "Verifynd DMI Pool Data ......" and there is char indicator (you know, that flashing thingy indicating where you are to write next letter) in a new line.

Flash drives are just emulated usb emulated floppy and hdd drives. Consequently, the same rules apply. A well made floppy or HD bootloader written to the MBR will load your code. I'm not sure how to work with partitions, however, but if you look a few posts down at my older post (i have 2 posts on the front page, one dealing with my bootloader issues, the other with v86), you'll see an example of a bootloader that isn't using a file system. I don't know anything about fat (i opted out of it when i realized i didn't need it for my current project), but i think you could learn from my sample code (but read and see how it was made to work).

EDIT: just from glancing at your code, it does work, you're just putting it in the wrong place. Your BIOS isn't finding it.

[makuc9]

Flash drives are just emulated usb emulated floppy and hdd drives. Consequently, the same rules apply. A well made floppy or HD bootloader written to the MBR will load your code. I'm not sure how to work with partitions, however, but if you look a few posts down at my older post (i have 2 posts on the front page, one dealing with my bootloader issues, the other with v86), you'll see an example of a bootloader that isn't using a file system. I don't know anything about fat (i opted out of it when i realized i didn't need it for my current project), but i think you could learn from my sample code (but read and see how it was made to work).

EDIT: just from glancing at your code, it does work, you're just putting it in the wrong place. Your BIOS isn't finding it.

You were right, in a way... I finally managed to solve my problem. Here is cause of the problem, for anyone who is facing the same thing:
http://support.microsoft.com/kb/954457

And I've been trying so hard to discover what I was doing wrong, until you got me thinking to the right way (and into more googling ).
Thanks. (And I only discover this now, after modifying the code numerous times already, modifying for doing the same thing different way. Argh. Stupid me ...

Thank you all for your time and cooperation. Sincerely.