[PATCH] fix the XIP size and fallback computations for M2V-MX SE

[Rudolf Marek]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

It seems I'm quite tired to formulate the description of the patch accurately I
hope it is understandable somehow.

Following patch fixes the XIP computation issue. I removed the normal image
because it was not working anyway (it was hardcoded) and because it allows me to
fix the XIP base to something sane (and use generic computation and approach)

This board is bit tricky because until now it required the VGA BIOS on the flash
start. XIP will work with 64KB aligned base, therefore the VGA ROM image must be
aligned too to 64KB.

Third reason is that now it makes 384KB for additional ROMs and payload - plenty
of space for CBFS fun ;)

Now it uses generic _ROMBASE etc computation which is nice too.

Signed-off-by: Rudolf Marek

Rudolf
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkn8wQgACgkQ3J9wPJqZRNUftgCfXjuQUkJwksgI4VU81mdJ78Kh
PyMAoMMWXjUKmq5YVQhsWZ6iFMEdgTSu
=mtSc
-----END PGP SIGNATURE-----

[Rudolf Marek]

Hi,

Can someone take a look here?

Rudolf

[PATCH] fix the XIP size and fallback computations for M2V-MX SE

[ron minnich]

Acked-by: Ronald G. Minnich

Looks ok. Long term, we do need to figure out what to do about
failover/fallback/normal.

I had an idea that we could use failover to select CBFS rom code
entries for fallback and normal, but it's not obvious that is the
right way to go. Plus, the still-unresolved question of fixed-location
CBFS entries needs to be resolved.

I only mention this in the hope that someone can take a look at the idea.

ron

[PATCH] fix the XIP size and fallback computations for M2V-MX SE

[Rudolf Marek]

Hi Ron,

Thanks its in as: Committed revision 4265.

Here are some random comments (while watching icehockey - so not so clear ;)

Maybe the trick with different CS would work for normal / fallback? Maybe better 
than linker stuff. Not sure how is the stuff for dataseg maybe it is just copied 
to ram I guess (next stage)

Or maybe we can just have some small panic loader through serial/USB instead of 
second full image?

U-boot is doing the relocation for its RAM image (just FYI) there is some code 
for that.

Thanks,
Rudolf

[PATCH] fix the XIP size and fallback computations for M2V-MX SE

[ron minnich]

On Sun, May 10, 2009 at 1:44 PM, Rudolf Marek  wrote:
> U-boot is doing the relocation for its RAM image (just FYI) there is some
> code for that.

we might want to see if what they are doing could work for us. The CS
approach does seem to make lots of sense, since we use CS zero-based
anyway. Just build the ROM parts  with 0-based code and adjust the CS
when the ROm code is started.

ron

[PATCH] fix the XIP size and fallback computations for M2V-MX SE

[Peter Stuge]

Rudolf Marek wrote:
> Or maybe we can just have some small panic loader through
> serial/USB instead of second full image?

Yes, we should have a panic room.

ron minnich wrote:
> The CS approach does seem to make lots of sense,

I like it too! Make some use of the paging.

//Peter

[Carl-Daniel Hailfinger]

It seems I forgot to send this mail although it sat finished in my
drafts folder. Sorry.

On 02.05.2009 23:54, Rudolf Marek wrote:
> It seems I'm quite tired to formulate the description of the patch
> accurately I
> hope it is understandable somehow.
>
> Following patch fixes the XIP computation issue. I removed the normal
> image
> because it was not working anyway (it was hardcoded) and because it
> allows me to
> fix the XIP base to something sane (and use generic computation and
> approach)
>
> This board is bit tricky because until now it required the VGA BIOS on
> the flash
> start. XIP will work with 64KB aligned base, therefore the VGA ROM
> image must be
> aligned too to 64KB.
>
> Third reason is that now it makes 384KB for additional ROMs and
> payload - plenty
> of space for CBFS fun ;)
>
> Now it uses generic _ROMBASE etc computation which is nice too.
>
> Signed-off-by: Rudolf Marek

The Config.lb part is awesome and I really appreciate it because I
didn't dare to convert that file to the ROM size calculation include file.
The rest looks good as well, but I'm not an expert for that code.

Acked-by: Carl-Daniel Hailfinger

Regards,
Carl-Daniel