OSDev.org

Here is my first MBR, I wrote it to enable access to >2TB drives. Any questions or comments are welcome, and especially bug/error reports.

Assembled with nasm

Fist of all, this is the wrong sub-forum for such things. Next time, try "Announcements, Test Requests, & Job openings"; it really makes looking for topics a pain if we don't respect this.

Bugs:

• movb rep... I assume you meant rep movsb, when moving the sector.
• Why do you insist on playing with DL? You don't need to specify any drive, BIOS does that for you.
• You only check to see if the first partition is active, the rest remain unchecked.
• The CPU won't halt if an IRQ comes along. CLI first.
• You don't properly check to see if you really support EDD, only do it superficially.
• You jump to the boot sector even if there was an error on the CHS read.
• The boot sector isn't checked for a signature.
• Why do "mov ax", "add ax", "mov si", when you can simply "lea si". Note that "add si" also works, in case you didn't know about LEA
• The disk address buffer should be hardcoded. Why on earth would you make the code give it values if they are fixed? It makes no sense...
• If there's a problem on LBA reads, you may end up reading forever.
• Why are the LBA-64 fields named "partition_x_hi"?
• I can't see the reason for adding the .data and .bss sections.

Also, I'm not sure about the few LBA-64 lines as I am unfamiliar with the thing. AFAIK, it's still a draft so it's never implemented anyway.

Although there are still bugs, I have to say congratulations. It's better than doing nothing.

Love4Boobies wrote:Fist of all, this is the wrong sub-forum for such things. Next time, try "Announcements, Test Requests, & Job openings"; it really makes looking for topics a pain if we don't respect this.

Bugs:

• movb rep... I assume you meant rep movsb, when moving the sector.
• Why do you insist on playing with DL? You don't need to specify any drive, BIOS does that for you.
• You only check to see if the first partition is active, the rest remain unchecked.
• The CPU won't halt if an IRQ comes along. CLI first.
• You don't properly check to see if you really support EDD, only do it superficially.
• You jump to the boot sector even if there was an error on the CHS read.
• The boot sector isn't checked for a signature.
• Why do "mov ax", "add ax", "mov si", when you can simply "lea si". Note that "add si" also works, in case you didn't know about LEA
• The disk address buffer should be hardcoded. Why on earth would you make the code give it values if they are fixed? It makes no sense...
• If there's a problem on LBA reads, you may end up reading forever.
• Why are the LBA-64 fields named "partition_x_hi"?
• I can't see the reason for adding the .data and .bss sections.

Also, I'm not sure about the few LBA-64 lines as I am unfamiliar with the thing. AFAIK, it's still a draft so it's never implemented anyway.

Can u also FIX the bugs? I've searched for a good MBR a lot of weeks, so it would be nice if u can make this mbr without bugs

Sure, here's mine. It has support for both the legacy and EDD interfaces and very good error detection. It's commented fairly well because I also used it for a tutorial I was writing. If you're wondering why I used NASM structures, it was because I also presented some of the NASM functionaility. Let me know how it works out for you.