[LinuxBIOS] DFI P2XBL Motherboard Target (440BX)

[roger]

I just built Uwe's s1846 target and saw no output on /dev/ttyS* (serial
port).

What differences should I be concerned about using the s1846 target for
this motherboard 440BX based motherboard in order to get debug output?
(Box acts like no bios is plugged into the motherboard.)

(When I get time, I'll yank parts out of this Tyan 1832DL to see if the
s1832 built flash boots on it.)

Thu Apr 19 04:06:55 PDT 2007

[Uwe Hermann]

On Thu, Apr 19, 2007 at 04:09:15AM -0700, roger wrote:
> I just built Uwe's s1846 target and saw no output on /dev/ttyS* (serial
> port).
>
> What differences should I be concerned about using the s1846 target for
> this motherboard 440BX based motherboard in order to get debug output?
> (Box acts like no bios is plugged into the motherboard.)

Which Super I/O is on the board? If it's not a PC87309 it won't work.
You'll need to change the Super I/O in auto.c then...

Uwe.

[roger]

On Thu, 2007-04-19 at 13:46 +0200, Uwe Hermann wrote:
> On Thu, Apr 19, 2007 at 04:09:15AM -0700, roger wrote:
> > I just built Uwe's s1846 target and saw no output on /dev/ttyS* (serial
> > port).
> >
> > What differences should I be concerned about using the s1846 target for
> > this motherboard 440BX based motherboard in order to get debug output?
> > (Box acts like no bios is plugged into the motherboard.)
>
> Which Super I/O is on the board? If it's not a PC87309 it won't work.
> You'll need to change the Super I/O in auto.c then...

Just for reference, looks like a Winbond  W83977EF-AW

In the cvs, I see LinuxBIOSv2/src/superio/winbond/w83977tf

I'm going to give it a shot.  Also just downloaded the chip spec pdf
files for comparison.

If things just work, you'll here.

Also for reference, my Tyan 1832DL (440bx dual cpu) has also a PC87309
(IBW* are the first three suffix letters I *think*).

I probably should have ordered a 2nd Bios Savior for my 1832, but for
now, the P2XBL 440BX motherboard is easily torn apart versus this
server.

Thu Apr 19 12:18:25 PDT 2007

[roger]

On Thu, 2007-04-19 at 12:18 -0700, roger wrote:
> On Thu, 2007-04-19 at 13:46 +0200, Uwe Hermann wrote:
> > On Thu, Apr 19, 2007 at 04:09:15AM -0700, roger wrote:
> > > I just built Uwe's s1846 target and saw no output on /dev/ttyS* (serial
> > > port).
> > >
> > > What differences should I be concerned about using the s1846 target for
> > > this motherboard 440BX based motherboard in order to get debug output?
> > > (Box acts like no bios is plugged into the motherboard.)
> >
> > Which Super I/O is on the board? If it's not a PC87309 it won't work.
> > You'll need to change the Super I/O in auto.c then...

src/superio/winbond/w83977tf/w83977tf.h appears to me it's in decimal
format compared to src/superio/nsc/pc87309/pc87309.h which is in
hexadecimal format.  Could this be an issue?

Also, still not getting output after just editing the auto.c and
replacing pc97309 with w83977tf.  One thing I left out was, the ASUS P2B
auto.c impliments w83977tf with the the following (0x3f0) value:
#define SERIAL_DEV PNP_DEV(0x3f0, W83977TF_SP1)

Tyan S1846 uses (0x2e)
/*#define SERIAL_DEV PNP_DEV(0x2e, PC87309_SP1)*/

Basically, I should be getting some linuxbios specific output on ttyS*.
And filo is last with it's output right?

Btw, the w83977tf seems to have been published a year earlier then this
w83977ef/ctf chip.  Just scanning over the pdf files, serial pins look
identical identical. 
--
Roger
http://www.eskimo.com/~roger/index.html
Key fingerprint = 8977 A252 2623 F567 70CD 1261 640F C963 1005 1D61

Thu Apr 19 14:24:40 PDT 2007

[Corey Osgood]

roger wrote:
> src/superio/winbond/w83977tf/w83977tf.h appears to me it's in decimal
> format compared to src/superio/nsc/pc87309/pc87309.h which is in
> hexadecimal format.  Could this be an issue?
>
> Also, still not getting output after just editing the auto.c and
> replacing pc97309 with w83977tf.  One thing I left out was, the ASUS P2B
> auto.c impliments w83977tf with the the following (0x3f0) value:
> #define SERIAL_DEV PNP_DEV(0x3f0, W83977TF_SP1)

The w83977tf worked fine for me with whatever code is currently in
linuxbios for it, and the above line. Others have had problems with
getting the tf's code to work on other versions, you might be better off
to try porting from another super io, and see what happens.
> Tyan S1846 uses (0x2e)
> /*#define SERIAL_DEV PNP_DEV(0x2e, PC87309_SP1)*/
>
> Basically, I should be getting some linuxbios specific output on ttyS*.
> And filo is last with it's output right?

Well, filo or grub or the kernel, depending on your configuration.

-Corey

[roger]

On Thu, 2007-04-19 at 18:07 -0400, Corey Osgood wrote:
> roger wrote:
> > src/superio/winbond/w83977tf/w83977tf.h appears to me it's in decimal
> > format compared to src/superio/nsc/pc87309/pc87309.h which is in
> > hexadecimal format.  Could this be an issue?
> >
> > Also, still not getting output after just editing the auto.c and
> > replacing pc97309 with w83977tf.  One thing I left out was, the ASUS P2B
> > auto.c impliments w83977tf with the the following (0x3f0) value:
> > #define SERIAL_DEV PNP_DEV(0x3f0, W83977TF_SP1)
>
> The w83977tf worked fine for me with whatever code is currently in
> linuxbios for it, and the above line. Others have had problems with
> getting the tf's code to work on other versions, you might be better off
> to try porting from another super io, and see what happens.

Out of curiosity, did a search on google for this superio and noticed a
similar discussion. Some recommendations of also using a Bios Post Card.
If you think it'd be wise, I'll grab one for $30 or so.

w83977ef-ctf.pdf
w83977tf.pdf

Both chips seem similar along with using 0x3f0 and 0x370 twice to put
the chip into extended mode. (I think this is why 0x3f0 is being passed
for?)

Anyways, I've seen no output on serial yet with using the
0x3f0,0x3f0,0x370 values so far.

If these values are what I'm thinking is there a need to use other
values for this function?  (..forgive my C lingo. :-/)

Thu Apr 19 18:49:50 PDT 2007

[Uwe Hermann]

On Thu, Apr 19, 2007 at 02:25:34PM -0700, roger wrote:
> src/superio/winbond/w83977tf/w83977tf.h appears to me it's in decimal
> format compared to src/superio/nsc/pc87309/pc87309.h which is in
> hexadecimal format.  Could this be an issue?

No.
> Also, still not getting output after just editing the auto.c and
> replacing pc97309 with w83977tf.  One thing I left out was, the ASUS P2B
> auto.c impliments w83977tf with the the following (0x3f0) value:
> #define SERIAL_DEV PNP_DEV(0x3f0, W83977TF_SP1)

Looks correct, but if it doesn't work try 0x370, 0x2e, or 0x4e, too.
> Basically, I should be getting some linuxbios specific output on ttyS*.

Yes. You can put some

print_debug("Hello world!\r\n");

after the

console_init();

line to test serial output.

Please post a patch with your changes, maybe we can spot an error...

Uwe.

[roger]

On Fri, 2007-04-20 at 01:20 +0200, Uwe Hermann wrote:
> > Also, still not getting output after just editing the auto.c and
> > replacing pc97309 with w83977tf.  One thing I left out was, the ASUS P2B
> > auto.c impliments w83977tf with the the following (0x3f0) value:
> > #define SERIAL_DEV PNP_DEV(0x3f0, W83977TF_SP1)
>
> Looks correct, but if it doesn't work try 0x370, 0x2e, or 0x4e, too.
> From what I'm guessing, the code is suppose to be doing this:

w83977ef-ctf.pdf (w83977tf.pdf is similar)
---clip---
8.1.1 Extended Function Registers
In Compatible PnP, there are two ways to enter Extended Function and
read or write the configuration registers. HEFRAS (CR26 bit 6) can be
used to select one out of these two methods of entering the
Extended Function mode as follows:

HEFRAS address and value
0 write 87h to the location 3F0h twice
1 write 87h to the location 370h twice
---clip---
> > Basically, I should be getting some linuxbios specific output on ttyS*.
>
> Yes. You can put some
>
> print_debug("Hello world!\r\n");

Ok... loaded with "Hello world!". ;-)

Shouldn't the DEBUG options in Options.lb & Config.lb cover this?

I'll be flashing with these changes until tomorrow.

Thu Apr 19 17:44:28 PDT 2007

[roger]

On Thu, 2007-04-19 at 17:46 -0700, roger wrote:
> On Fri, 2007-04-20 at 01:20 +0200, Uwe Hermann wrote:

I have ordered a BIOS Post card from siliconkit.com (along with the
recommended extension cable).

Should be getting it this Monday.  Got the info from the linuxbios faq
about post codes being primarily used for debugging loading of linuxbios
flash code.

(Also ordered a Bios Savior.)

I'll post my results around Monday or Tuesday on a new thread with "Post
Card Results" in the SUBJECT. ;-)

Then I can start having fun debugging with my (Intel 82815/82801) Dell
Inspiron 8100 laptop with a mini-pci post card, albeit, probably have to
send the board out for soldering a socketed motherboard or something.

Fri Apr 20 15:22:20 PDT 2007