intel 440bx unshadow for doc2001

intel 440bx unshadow for doc2001

[roger]

Figured out from scanning through the openbios/devbios code on howto
unshadow the intel 440bx to be able to successfully load the MTD doc2001
module and have it successfully detect the DOC. -- I am not sure if this
is what you call "memory mapping" (ie
/usr/src/linux-2.4.20/drivers/mtd/maps/l440gx.c)

The devbios/openbios project can be found here:
http://www.openbios.info/download/index.html
And the url is at the bottom of the page for
http://www.openbios.info/bin/devbios-0.3.2.tar.gz

Does anybody know if any of this is already in the MTD drivers?  This
code unshadows and gives access to many bios devices.

Went into the openbios/devbios/pcisets.c and commented out the two
statements within the "static void intel4x0_deactivate(void)" function
(this prevents the unshadowing after the bios.o module is loaded -- and
while i was at it, uncommented the a printk debug statement).  Loading
the bios.o module shows the isa bridge info:

Jul  9 02:40:27  BIOS: host bridge is 8086, 7190, 0
Jul  9 02:40:27  BIOS: isa bridge is 8086, 7110, 38
Jul  9 02:40:27  BIOS: isa bridge cfg is 0x64
Jul  9 02:40:27  BIOS: isa bridge cfg is 0x64
Jul  9 02:40:27  BIOS: isa bridge cfg is 0x64
(last line repeated many times)
Jul  9 02:40:27  BIOS: 0k flashchip (ID 0x5f1f) found at physical
address 0xfff01000 (va=0xd8854000+0x101000).
Jul  9 02:40:27  BIOS: Probing PCI device with 256k rom
Jul  9 02:40:27  BIOS: Probing PCI device with 64k rom



Once I unshadowed the intel 440bx, the following script loads doc with
ease:

#!/bin/bash
modprobe doc2001
modprobe docprobe doc_config_location=0xfff00000

cat /proc/mtd

modprobe nft


If i'm correct, the unshadow code just needs to be written into MTD for
everything to work easily? 

-- 

Roger
http://www.eskimo.com/~roger/index.html

Re: intel 440bx unshadow for doc2001

[David Woodhouse]

On Sun, 2003-07-13 at 12:08, roger wrote:
> If i'm correct, the unshadow code just needs to be written into MTD for
> everything to work easily? 

Seems perfectly reasonable.

In theory, we ought to already have the appropriate unshadowing code in
drivers/mtd/maps/l440gx.c; it's just that that driver will only then
look for NOR flash chips rather than a DiskOnChip. Does it work if you
load l440gx.c, watch it fail to find any flash chips, then unload it and
load docprobe.c?

I'm working on a rewrite of the DiskOnChip drivers; I'll probably make
it so that a separate driver can call a 'probe for DiskOnChip at address
XXXX' function, then we can make that an option for l440gx and other
such drivers rather than having hard-coded tables in the docprobe code
itself.

-- 
dwmw2

Re: intel 440bx unshadow for doc2001

[roger]

On Mon, 2003-07-14 at 04:09, David Woodhouse wrote:
> On Sun, 2003-07-13 at 12:08, roger wrote:
> > If i'm correct, the unshadow code just needs to be written into MTD for
> > everything to work easily? 
> 
> Seems perfectly reasonable.
> 
> In theory, we ought to already have the appropriate unshadowing code in
> drivers/mtd/maps/l440gx.c; it's just that that driver will only then
> look for NOR flash chips rather than a DiskOnChip. Does it work if you
> load l440gx.c, watch it fail to find any flash chips, then unload it and
> load docprobe.c?

no. on 440BX.B683 motherboard, the l440gx.o module fails to even load. 
believe me, in my prior posts, you'll probably notice that i did
everything including pulling a rabbit out of my computer's rear port.

on my 440BX.tyan motherboard, I think the l440gx.o module will load but
it still fails to find the doc(?) (or vice versa with the motherboards)

> I'm working on a rewrite of the DiskOnChip drivers; I'll probably make
> it so that a separate driver can call a 'probe for DiskOnChip at address
> XXXX' function, then we can make that an option for l440gx and other
> such drivers rather than having hard-coded tables in the docprobe code
> itself.

I was scanning over the openbios/devbios/pcisets.c and looks like
there's only a couple lines of code that are required to unshadow the
bus so that the DOC (or motherboard's original bios chip) can become
visible to the mtd modules. The other motherboards (other chipsets)
appear to require more code to unshadow.

If i'm not mistaken, I won't be able to use the DOC with linux bios
until mtd can recognize the chip on bootup. (?)

well, code is there. it just needs to be integrated correct?

-- 

Roger
http://www.eskimo.com/~roger/index.html