Re: [ECOS] Help with Redboot an a ARM SA1100 Lart Board

[Gary Thomas <gary at mlbassoc dot com>]

On Fri, 2003-08-08 at 11:32, Steven.Clugston wrote:
> I am trying to get ecos running on a SA1100 variant board called a 'Lart' from 
> http://www.lart.tudelft.nl/
> I have used a setup based on the Brutus template as I cannot get the sa1100mm 
> template to work at all. The lart comes with a custom boot loader called blob 
> that lets you run linux from flash ram. The board has 4Mb of Intel 28F160F3 
> flash ram and 32Mb of EDO DRAM configured on banks 0 and one only. It has no 
> SRAM or any other memory.
> 
> The blob boot loader reports the memory as follows:
> 
> Memory map:
>   0x00800000 @ 0xC0000000 (8 MB)
>   0x00800000 @ 0xC1000000 (8 MB)
>   0x00800000 @ 0xC8000000 (8 MB)
>   0x00800000 @ 0xC9000000 (8 MB)
> 
> I don't really understand why the memory is fragmented into 8Mb blocks and not 
> two continuous blocks of 16Mb per bank with an offset of 128Mb from the base 
> of each bank, but it works with the supplied linux setup and I have used the 
> DRAM and flash waveform/config register settings from the blob loader and put 
> them into the platform_setup.h in place of the brutus ones. I have edited the 
> assembly to make sure that the same actual values are being passed to the 
> registers as the blob loader.
> 
> My problem is that when I try to upload then run code from reboot (running 
> from ROM)it either hangs from the go command (using the brutus memory config) 
> or gives no output except a "$T0a0f:e47e0040;0d:c0a60000;#dc" string everytime 
> I hit a key in the terminal window when using the blob memory settings. This 
> same string occurs after tying 'go' or 'go address' regardless of whether I 
> upload code or not or what ever memory location I try to run it from, as shown 
> below:
> 
> Platform: LART development system (StrongARM 1100)
> Copyright (C) 2000, 2001, 2002, Red Hat, Inc.
> 
> RAM: 0x00000000-0x02000000, 0x0000d8a8-0x02000000 available
> RedBoot> go
> $T0a0f:e47e0040;0d:c0a60000;#dc

Am I missing something here?  I don't see where you loaded your 'hello'
example into RAM.

The $Txxxx string is GDB protocol, indicating that the program has
failed.  If you were to disconnect your terminal emulator and connect
to the board using GDB (I see that this works from below), you could
get an idea where things are going wrong.

> 
> When I use gdb with -w insight to upload and run the examples such as the 
> hello world one, it seems to step through the code OK.
> 
> Please can anyone help, I have been struggling with this for a week or so. I 
> have included object dumps of the hello executable and the redboot rom image, 
> and also the memory mapping I have tried to use in various platform sepecific 
> files.

First glance, the setup files (MMU mapping especially) look fine to me.

-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss