[Buildroot] Confusion with building at91bootstrap

[Buildroot] Confusion with building at91bootstrap

[Cameron Hutchison]

I am trying to build an image for an AT91SAM9260EK board and I'm getting
a little confused with the build ot at91bootstrap. I'm using BR 2009.11

When at91bootstrap is configured, the buildroot makefile
(at91bootstrap.mk) runs "make at91sam9260ek_defconfig" which sets up
at91bootstrap for building for my specific board.

That config file defines IMG_SIZE to be 0x40000 and passes
-DIMG_SIZE=0x40000 to the compiler.

The buildroot makefile passes -DIMG_SIZE=0x32000. This value comes from
the buildroot config.

This causes a build warning that IMG_SIZE is redefined.

Apart from that I don't yet know what the correct image size should be,
it looks like the at91bootstrap integration into buildroot is broken.
The option to set the image size in the BR config cannot work, since the
at91bootstrap (2.10) source will attempt to override it anyway. Am I
missing something here?

Also, the at91bootstrap defconfig for this board adds the suffix of
"card" to the output file, making that filename contain "dataflashcard"
when I have configured for "dataflash" not "dataflashcard". I dont yet
know what the difference is between these (my board does not appear to
have any sort of card or slot for one), so I'm not sure what should be
done here, but it appears inconsistent.

Finally, is this the best place to discuss at91bootstrap and other
issues related to this board and buildroot, or is there a better forum?

Thanks

[Buildroot] Confusion with building at91bootstrap

[Ulf Samuelsson]

Cameron Hutchison skrev:
> I am trying to build an image for an AT91SAM9260EK board and I'm getting
> a little confused with the build ot at91bootstrap. I'm using BR 2009.11
> 
> When at91bootstrap is configured, the buildroot makefile
> (at91bootstrap.mk) runs "make at91sam9260ek_defconfig" which sets up
> at91bootstrap for building for my specific board.
> 
> That config file defines IMG_SIZE to be 0x40000 and passes
> -DIMG_SIZE=0x40000 to the compiler.
> 
> The buildroot makefile passes -DIMG_SIZE=0x32000. This value comes from
> the buildroot config.
> 
> This causes a build warning that IMG_SIZE is redefined.

Ignore it.

> 
> Apart from that I don't yet know what the correct image size should be,
> it looks like the at91bootstrap integration into buildroot is broken.
> The option to set the image size in the BR config cannot work, since the
> at91bootstrap (2.10) source will attempt to override it anyway. Am I
> missing something here?
> 

This is the size of the image which is loaded into SDRAM.
Typically u-boot.
I think the idea is that buildroot can override the
config file.


> Also, the at91bootstrap defconfig for this board adds the suffix of
> "card" to the output file, making that filename contain "dataflashcard"
> when I have configured for "dataflash" not "dataflashcard". I dont yet
> know what the difference is between these (my board does not appear to
> have any sort of card or slot for one), so I'm not sure what should be
> done here, but it appears inconsistent.

dataflashcard = boot from NPCS0
dataflash = boot from NPCS1

Easy to understand, if you use the SAM9260EK.

You want to use the latest = 2.13.
2.13-rc# has some issues.
Need a modification of the recipe in buildroot.

BR
Ulf Samuelsson



> 
> Finally, is this the best place to discuss at91bootstrap and other
> issues related to this board and buildroot, or is there a better forum?
> 
> Thanks
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot

[Buildroot] Confusion with building at91bootstrap

[Cameron Hutchison]

Ulf Samuelsson <ulf.samuelsson@atmel.com> writes:

>Cameron Hutchison skrev:
>> 
>> This causes a build warning that IMG_SIZE is redefined.

>Ignore it.

OK. I assumed the second definition of IMG_SIZE was ignored, but my
little test program tells me otherwise. Since the BR config of IMG_SIZE
comes last on the command line, that would be the definitive size.

>This is the size of the image which is loaded into SDRAM.
>Typically u-boot.
>I think the idea is that buildroot can override the
>config file.

I spent the rest of yesterday getting my head around the different
storage on the 9260EK board and memory layout options, so I think I'll
be ok there, but ...

>> Also, the at91bootstrap defconfig for this board adds the suffix of
>> "card" to the output file, making that filename contain "dataflashcard"
>> when I have configured for "dataflash" not "dataflashcard". I dont yet
>> know what the difference is between these (my board does not appear to
>> have any sort of card or slot for one), so I'm not sure what should be
>> done here, but it appears inconsistent.

>dataflashcard = boot from NPCS0
>dataflash = boot from NPCS1

>Easy to understand, if you use the SAM9260EK.

My concern was that I have configured everything in BR to produce an
image for dataflash, but the output image for at91bootstrap was labelled
dataflashcard. Since I am not using dataflashcard (I found the slot on
the underside of the board :-)), I wanted to make sure I had not done
something wrong. I can't see what, and it appears to build for
dataflash, but just calls the image dataflashcard. I think that is only
because CONFIG_CARD_SUFFIX="card" is in the defconfig.

I'm now trying to build for NAND flash since my final target will not
have dataflash anyway, but that wont build because make MEMORY=nandflash
doesnt #define AT91C_SPI_PCS_DATAFLASH, but the defconfig has
CONFIG_DATAFLASH=y. It seems as well as "make MEMORY=nandflash" the
..config needs to be changed from the defconfig to build.

>You want to use the latest = 2.13.
>2.13-rc# has some issues.
>Need a modification of the recipe in buildroot.

Ok, thanks. I'll try 2.13 shortly.