cvs commit: src/release Makefile
martin at gneto.com
Wed Feb 1 05:43:29 PST 2006
Sergey Babkin skrev:
>> From: Peter Jeremy <peterjeremy at optushome.com.au>
>> Some time ago, I was surprised to find that my son's oldish (mid 2003)
>> system was happy to try and boot off a USB memory stick.
> The BIOSes have been able to do this feat for a while now. There are two caveats though:
> 1. It's slow, since BIOSes tend to use USB 1.x speeds,
> and even that not in a really fast way. So the big
> images are best to be avoided. Though probably it's
> still faster than floppies.
> 2. Accessing the root filesystem - the USB drivers
> have to be either statically compiled into the
> kernel or be loaded by the boot loader through BIOS
> (maybe through an analog of initrd ram disk - I'm
> not sure what advances have been done to the FreeBSD
> boot loader nowadays).
There is another major caveat for some BIOSes:
3. The BTX loader is a 32-bit program that calls the BIOS functions via
v86 calls, these calls fails on some BIOSes when a disk/cd is attached
with USB resulting in the infamous BTX halted error.
I've been told that this is due to the fact that the BIOS tries to enter
32-bit mode again not realizing that the CPU already is in this mode.
I recently found out that there is something called BIOS32, a 32-bit
interface to the BIOS functions. More info here:
I haven't had time to see if calling the function that blows up when
booting from USB-CD (check which drives are available) thru BIOS32 is
going to work or not.
I have lots of boards that have this problem (Supermicro P8SCi for one),
if someone wants to have any changes tested. I thought using a USB thumb
drive instead of a CD-RW would be easier but I haven't had time to do
anything with yet.
More information about the cvs-src