Re: [Bug-grub] Re: grub-0.93-win32-iso9660.diff

[Timothy Baldwin]

On Friday 05 Sep 2003 4:44 pm, Yoshinori Okuji wrote:

> The reason why I may compromise without finishing Multiboot issues is just
> that the missing features are not necessary to  boot Multiboot-compliant
> OSes in most cases (such as GNU/Hurd).

With respect to track_int13, tracking I/O ports isn't a reliable means of 
binding BIOS devices to OS devices on modern hardware/firmware beacuse:

1. It doesn't identify the order of devices on an interface, and even if it 
did adding/removing devices may invalidate that order.
2. If port access is tracked during interrupts, it may track other use of I/O 
ports of an interface to which disks are attached.
3. If port access is not tracked during interrupts, it's possible port usage 
may not be detected

(I'm not sure about points 2 & 3, but have seen 1 happen)

int 0x13 with ah=0x48 should be used to deterimine the mapping, with I/O port 
tracking used only if is not supported.

-- 
Member AFFS, WYLUG, SWP (UK), ANL, Leeds SA, Leeds Anti-war coalition
No to software patents!    No to DRM/EUCD - hands off our computers!

pgp4s1xtoMqM_.pgp
Description: signature