[LinuxBIOS] northbridge docs

Hello,
I am working on the code for my northbridge (Intel 82830).  
Particularly the raminit.c file. By looking at the sources from the  
other northbridges and my datasheets, I am getting a little lost. My  
datasheet (Intel) lists all the registers and their defaults but  
doesn't really explain the memory initialization process. Could  
someone direct me to some good docs that explain how memory  
initialization happens at boot time? The more detail the better. I am  
using SDRAM so SDRAM specific would be great. I was looking at the  
code from the via vt8601, but it seems like alot of the registers are  
hardcoded into the code (no disrespect Ron, I saw you converted it to  
C). I would like to write one that is a lot more dynamic (more  
definitions to uniqe code) so it can also be used to as sort of a  
template for other SDRAM based northbridges.

Thanks - Joe

[Uwe Hermann]

On Fri, Mar 16, 2007 at 08:29:23AM -0400, [e-mail] wrote:
> I am working on the code for my northbridge (Intel 82830).  
> Particularly the raminit.c file. By looking at the sources from the  
> other northbridges and my datasheets, I am getting a little lost. My  
> datasheet (Intel) lists all the registers and their defaults but  
> doesn't really explain the memory initialization process.

According to Ron the 440LX datasheets were quite good. See
http://www.intel.com/design/chipsets/datashts/290564.htm

(RAM init starts on page 84 or so)

Of course this is chipset-specific to some extent, but it should give
you a generel idea about the process.

Uwe.

[Yuning Feng]

2007/3/17, Uwe Hermann :
> According to Ron the 440LX datasheets were quite good. See
> http://www.intel.com/design/chipsets/datashts/290564.htm

(RAM init starts on page 84 or so)
> Of course this is chipset-specific to some extent, but it should give
> you a generel idea about the process.

This kind of knowledge is important for a newbie to understand LinuxBIOS,
but takes a long time to find.
Would it be a good idea to collect these beginning informations and make a
reading list or an FAQ ?

[Stefan Reinauer]

* Yuning Feng  [070324 08:16]:
> This kind of knowledge is important for a newbie to understand LinuxBIOS, but
> takes a long time to find.
> Would it be a good idea to collect these beginning informations and make a
> reading list or an FAQ ?

We do have a pretty big FAQ. If you suggest a wording, I'll add it.

BTW: Should we split the FAQ in several pages/FAQs? It is getting a
little hard to search.

[Yuning Feng]

> We do have a pretty big FAQ. If you suggest a wording, I'll add it.

Maybe a subtitle like "basic concepts" ?

While I understand that there should be a process of ram init, I have no
idea what should that look like. Reading an actual doc explaining that is a
good way to start, I think.

Well, it seems most people involved here are quite skillful, I do not know
how necessary would that subtitle be.

So this is only  a suggestion. Maybe just a short description of ram init
and a link to the doc would be good enough.

BTW: Should we split the FAQ in several pages/FAQs? It is getting a
> little hard to search.

Seems titles and subtitles are using the same font and size. That makes it
difficult to search, too.

Quoting Yuning Feng :
> > We do have a pretty big FAQ. If you suggest a wording, I'll add it.
>
> Maybe a subtitle like "basic concepts" ?
>
> While I understand that there should be a process of ram init, I have no
> idea what should that look like. Reading an actual doc explaining that is a
> good way to start, I think.
>
> Well, it seems most people involved here are quite skillful, I do not know
> how necessary would that subtitle be.
>
> So this is only  a suggestion. Maybe just a short description of ram init
> and a link to the doc would be good enough.
>
> BTW: Should we split the FAQ in several pages/FAQs? It is getting a
> > little hard to search.
>
> Seems titles and subtitles are using the same font and size. That makes it
> difficult to search, too.

If your going to go that far you should just have a forum. Forums are  
always a great reference point. That's my two cents worth.

Thanks - Joe

[Stefan Reinauer]

* [e-mail]  [070325 00:52]:
> > Seems titles and subtitles are using the same font and size. That makes it
> > difficult to search, too.
>
> If your going to go that far you should just have a forum. Forums are  
> always a great reference point. That's my two cents worth.

Forums share the disadvantages of mailing lists and wikis. What would a
forum add that we are still lacking?

[Peter Stuge]

On Sat, Mar 24, 2007 at 07:52:50PM -0400, [e-mail] wrote:
> If your going to go that far you should just have a forum. Forums
> are always a great reference point. That's my two cents worth.

Please, no forums. I can't stand them. The mailing list is so much
better IMHO.

//Peter

[Uwe Hermann]

On Sun, Mar 25, 2007 at 05:48:08AM +0200, Peter Stuge wrote:
> On Sat, Mar 24, 2007 at 07:52:50PM -0400, [e-mail] wrote:
> > If your going to go that far you should just have a forum. Forums
> > are always a great reference point. That's my two cents worth.
>
> Please, no forums. I can't stand them. The mailing list is so much
> better IMHO.

Full ack.

Uwe.

[Corey Osgood]

Uwe Hermann wrote:
> On Sun, Mar 25, 2007 at 05:48:08AM +0200, Peter Stuge wrote:
> > On Sat, Mar 24, 2007 at 07:52:50PM -0400, [e-mail] wrote:
> > > If your going to go that far you should just have a forum. Forums
> > > are always a great reference point. That's my two cents worth.
> >
> > Please, no forums. I can't stand them. The mailing list is so much
> > better IMHO.
>
> Full ack.
>
> Uwe.

And another. I was a moderator on a forum for several years, and I
really don't want to go back to that. Mailing lists tend to have so much
fewer trolls.

-Corey

Quoting [e-mail]:
> Hello,
> I am working on the code for my northbridge (Intel 82830).
> Particularly the raminit.c file. By looking at the sources from the
> other northbridges and my datasheets, I am getting a little lost. My
> datasheet (Intel) lists all the registers and their defaults but
> doesn't really explain the memory initialization process. Could
> someone direct me to some good docs that explain how memory
> initialization happens at boot time? The more detail the better. I am
> using SDRAM so SDRAM specific would be great. I was looking at the
> code from the via vt8601, but it seems like alot of the registers are
> hardcoded into the code (no disrespect Ron, I saw you converted it to
> C). I would like to write one that is a lot more dynamic (more
> definitions to uniqe code) so it can also be used to as sort of a
> template for other SDRAM based northbridges.
>
> Thanks - Joe

I just have to say, after hours of reading datasheets and examining  
code that who ever wrote the code for the e7520 and e7525 nothbridges  
is brilliant!! The way the code is setup is going to work perfect for  
me and maybe other people looking for Intel northbridge code starting  
points. I think if any of you working on the 440bx should just scrap  
the existing and start over with e7520 and e7525 as a template, maybe  
not, but you might be able to get it working this way. Anyways, I just  
wanted to Acknowledge some great code writing.

Thanks  - Joe

[Corey Osgood]

[e-mail] wrote:
> Quoting [e-mail]:
> > Hello,
> > I am working on the code for my northbridge (Intel 82830).
> > Particularly the raminit.c file. By looking at the sources from the
> > other northbridges and my datasheets, I am getting a little lost. My
> > datasheet (Intel) lists all the registers and their defaults but
> > doesn't really explain the memory initialization process. Could
> > someone direct me to some good docs that explain how memory
> > initialization happens at boot time? The more detail the better. I am
> > using SDRAM so SDRAM specific would be great. I was looking at the
> > code from the via vt8601, but it seems like alot of the registers are
> > hardcoded into the code (no disrespect Ron, I saw you converted it to
> > C). I would like to write one that is a lot more dynamic (more
> > definitions to uniqe code) so it can also be used to as sort of a
> > template for other SDRAM based northbridges.
> >
> > Thanks - Joe
>
> I just have to say, after hours of reading datasheets and examining  
> code that who ever wrote the code for the e7520 and e7525 nothbridges  
> is brilliant!! The way the code is setup is going to work perfect for  
> me and maybe other people looking for Intel northbridge code starting  
> points. I think if any of you working on the 440bx should just scrap  
> the existing and start over with e7520 and e7525 as a template, maybe  
> not, but you might be able to get it working this way. Anyways, I just  
> wanted to Acknowledge some great code writing.
>
> Thanks  - Joe

Joe, Uwe already did, and I really appreciate it ;) Although neither of
us has gotten it running yet, it definitely is a better design, I've
also been using something similar for via vt82c694 (again, no success
yet). I only wish there was some way to haul apart or examine the
factory bios to figure out what's going on with both of these chips, in
principle this should be so simple. Why does what worked with v1 not
work with v2???

-Corey

Quoting Corey Osgood :
> [e-mail] wrote:
> > Quoting [e-mail]:
> > > Hello,
> > > I am working on the code for my northbridge (Intel 82830).
> > > Particularly the raminit.c file. By looking at the sources from the
> > > other northbridges and my datasheets, I am getting a little lost. My
> > > datasheet (Intel) lists all the registers and their defaults but
> > > doesn't really explain the memory initialization process. Could
> > > someone direct me to some good docs that explain how memory
> > > initialization happens at boot time? The more detail the better. I am
> > > using SDRAM so SDRAM specific would be great. I was looking at the
> > > code from the via vt8601, but it seems like alot of the registers are
> > > hardcoded into the code (no disrespect Ron, I saw you converted it to
> > > C). I would like to write one that is a lot more dynamic (more
> > > definitions to uniqe code) so it can also be used to as sort of a
> > > template for other SDRAM based northbridges.
> > >
> > > Thanks - Joe
> >
> > I just have to say, after hours of reading datasheets and examining
> > code that who ever wrote the code for the e7520 and e7525 nothbridges
> > is brilliant!! The way the code is setup is going to work perfect for
> > me and maybe other people looking for Intel northbridge code starting
> > points. I think if any of you working on the 440bx should just scrap
> > the existing and start over with e7520 and e7525 as a template, maybe
> > not, but you might be able to get it working this way. Anyways, I just
> > wanted to Acknowledge some great code writing.
> >
> > Thanks  - Joe
>
> Joe, Uwe already did, and I really appreciate it ;) Although neither of
> us has gotten it running yet, it definitely is a better design, I've
> also been using something similar for via vt82c694 (again, no success
> yet). I only wish there was some way to haul apart or examine the
> factory bios to figure out what's going on with both of these chips, in
> principle this should be so simple. Why does what worked with v1 not
> work with v2???
>
> -Corey

Could eithor of you (Uwe or Corey) send me the 440bx files described  
above. I would like to take a look :)

Thanks - Joe