Although I haven't seen discussion of disassembling factory BIOS, I would think this has been discussed; why is this not routine, or effective, for these unruly Intel chipsets which can't get DRAM happy?
Not that I would look forward to disassembling legacy BIOS, but if that's what it takes...
Drew
On Thu, 30 Nov 2006, Drew Lundsten wrote:
Although I haven't seen discussion of disassembling factory BIOS, I would think this has been discussed; why is this not routine, or effective, for these unruly Intel chipsets which can't get DRAM happy?
Not that I would look forward to disassembling legacy BIOS, but if that's what it takes...
Unfortunately, bad idea.
First, the DMCA.
Second, how would you defend yourself against alleged copyright violations?
Russ
I will mention an idea I mentioned some time ago.
Now that we have cache-as-ram, it is possible to build a linuxbios which contains the emulator. But what you would do is very different: 1. start up 2. enable CAR 3. start the emulator with the proprietary BIOS as the payload. 4. Run it with max debug, and observe the register activity, etc. 5. you will be able to see enough to probably figure out how DRAM is started up.
This is doable TODAY.
Note also that you can run the proprietary bios under the emulator in the linuxbios tree and observe behaviour to see things, which might help. I.e. from under linux, you can run the BIOS and observe a little bit of what it does, up to the point at which it starts messing with I/Os and gets confused.
Our emulator is a very powerful tool, just waiting for someone to use it in CAR mode to figure out what the BIOS is doing. The question is, is the CAR memory size big enough for the emulator? I think it might be.
I think we have more than enough smart people on this list now to make this work. If you made it work, it would open up a lot of chipsets.
thanks
ron
Although I haven't seen discussion of disassembling factory BIOS, I would think this has been discussed; why is this not routine, or effective, for these unruly Intel chipsets which can't get DRAM happy?
Of course it has been done. It won't tell you _why_ certain things are done, or what they actually do -- disassembling machine code doesn't give you source code.
Not that I would look forward to disassembling legacy BIOS, but if that's what it takes...
If that's what it takes I for one am not al that interested in supporting such chipsets.
Segher
On Thu, 2006-11-30 at 18:26 +0100, Segher Boessenkool wrote:
Although I haven't seen discussion of disassembling factory BIOS, I would think this has been discussed; why is this not routine, or effective, for these unruly Intel chipsets which can't get DRAM happy?
Of course it has been done. It won't tell you _why_ certain things are done, or what they actually do -- disassembling machine code doesn't give you source code.
Another factor you have think about is how do you even know the factory bios is doing it right? If you imitate what a buggy factory BIOS is doing, you are going to reproduce the exact same bug in LB. I guess this is not pretty in the court.
Ollie
-
Drew Lundsten -
Li-Ta Lo -
ron minnich -
Russ Whitaker -
Segher Boessenkool