I have speculated that the XFree86 int10 code that (re)posts the card using I believe vm86 syscalls, would be a suitable test harness for this. So, boot off a second video card, then XFree86 will post the card in question in the DOS box when doing multihead or just running on the card that wasn't initialized at boot. If you were to modify the flags passed to vm86, not to allow IO, you just need a trap handler that logs the access then continues. For MMIO, page table permissions could also be used to generate traps, which log the access then continue, for the PCI IO region.
Jeremy ----- Original Message ----- From: "Adam Sulmicki" adam@cfar.umd.edu To: "Peter Busser" peter@trusteddebian.org Cc: Linuxbios@clustermatic.org Sent: Wednesday, May 21, 2003 9:07 AM Subject: Re: Reverse engineering
If you are only interested in those IO ports, an logic analyzer is probably better and easier to use. Modern LA can be programmed to log/filter various IO scenario.
It's probably a bit expensive to buy an LA or ICE or any other
specialised
hardware just to get a board going. So I thought it might be possible to
use
Bochs to do the job instead. But maybe I'm mistaken.
The problem with bochs is that it is not just "any hardware" but they attempt to emulate Intel 440LX (IIRC). So if you put here an standard bios I would expect it to complain a lot.
Of course it is still better than nothing and I believe there are people on the list which went this track before. Actualy it was other way around with getting LinuxBIOS bios to run under BOCHS IIRC.
While I myself hadn't used BOCHS to this purpose I have used Wine on regular basic, and I was able to *easily* track IO ports for communication between ChipWriter software and the ChipWriter hardware itself.
Just my two bits.
-- Adam Sulmicki http://www.eax.com The Supreme Headquarters of the 32 bit registers
Linuxbios mailing list Linuxbios@clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios