[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Sven Schnelle]

Hi List,

i'm trying to get u-boot running on the Atmel ATNGW100,
as the preloaded u-boot is rather old (v2010.08). However,
ist doesn't run:

U-Boot 2010.09 (Oct 11 2011 - 14:31:44)

U-Boot code: 00000000 -> 00012c1a  data: 00019d38 -> 0004e7a0
VMR table @ 0x0001a584
<hangs>

I've found the following thread on this list, were another user
reports the same problem:

http://lists.denx.de/pipermail/u-boot/2011-June/094385.html

Does anyone also still have this problem? Or does anyone have a
pointer for me before i start digging into that problem?

Thanks,

Sven

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Sven Schnelle]

On 10/11/2011 03:06 PM, Sven Schnelle wrote:
> Hi List,
>
> i'm trying to get u-boot running on the Atmel ATNGW100,
> as the preloaded u-boot is rather old (v2010.08). However,
> ist doesn't run:
>
> U-Boot 2010.09 (Oct 11 2011 - 14:31:44)
Please ignore the version, this is the log from trying a few older 
version. However,
with the latest git, it hangs at the exact same location.

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Andreas Bießmann]

Dear Sven,

Am 11.10.2011 15:06, schrieb Sven Schnelle:
> Hi List,
> 
> i'm trying to get u-boot running on the Atmel ATNGW100,
> as the preloaded u-boot is rather old (v2010.08). However,
> ist doesn't run:
> 
> U-Boot 2010.09 (Oct 11 2011 - 14:31:44)
> 
> U-Boot code: 00000000 -> 00012c1a  data: 00019d38 -> 0004e7a0
> VMR table @ 0x0001a584
> <hangs>
> 
> I've found the following thread on this list, were another user
> reports the same problem:
> 
> http://lists.denx.de/pipermail/u-boot/2011-June/094385.html

as I have written in this thread the MIMC is the only one with three
entries in the 'simple tlb'. This is the first time one complains about
ATNGW100.

I only have a STK1002 and our not mainline board here (and grasshopper)
here and I can confirm that at least ATSTK1002 and our not mainline
board do run with 2011.06 release. I could re-run the test with 2011.09
but I guess it will still run as before.

> Does anyone also still have this problem? Or does anyone have a
> pointer for me before i start digging into that problem?

I do not have any pointers right now but really interested in your
investigations.

best regards

Andreas Bie?mann

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Andreas Bießmann]

Dear Sven,

Am Di 11 Okt 2011 15:23:00 CEST schrieb Andreas Bie?mann:
> Dear Sven,
>
> Am 11.10.2011 15:06, schrieb Sven Schnelle:

<snip>

>> Does anyone also still have this problem? Or does anyone have a
>> pointer for me before i start digging into that problem?
>
> I do not have any pointers right now but really interested in your
> investigations.

One thing, we had much truble with our toolchain. Which toolchain do you
use? Could you provide me (off list) a u-boot.bin for atstk1002 or
grasshopper?

best regards

Andreas Bie?mann

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Sven Schnelle]

On 10/11/2011 03:28 PM, Andreas Bie?mann wrote:
> Dear Sven,
>
> Am Di 11 Okt 2011 15:23:00 CEST schrieb Andreas Bie?mann:
>> Dear Sven,
>>
>> Am 11.10.2011 15:06, schrieb Sven Schnelle:
> <snip>
>
>>> Does anyone also still have this problem? Or does anyone have a
>>> pointer for me before i start digging into that problem?
>> I do not have any pointers right now but really interested in your
>> investigations.
> One thing, we had much truble with our toolchain. Which toolchain do you
> use? Could you provide me (off list) a u-boot.bin for atstk1002 or
> grasshopper?

I'm using the atmel avr32 toolchain 3.2.3, which contains avr32-gcc 
4.4.3 and binutils 2.20.1.
I can't try the older atmel toolchain due to missing dependencies for my 
OS (Debian).

i've just recognized that ld complains during built:

avr32-ld:built in linker script:15: warning: memory region `FLASH' not 
declared
avr32-ld:built in linker script:69: warning: memory region `CPUSRAM' not 
declared

not sure if this is related to the problem i'm seeing.

Thanks
Sven

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Andreas Bießmann]

Dear Sven,

Am Di 11 Okt 2011 15:37:29 CEST schrieb Sven Schnelle:
> On 10/11/2011 03:28 PM, Andreas Bie?mann wrote:
>> Dear Sven,
>>
>> Am Di 11 Okt 2011 15:23:00 CEST schrieb Andreas Bie?mann:
>>> Dear Sven,
>>>
>>> Am 11.10.2011 15:06, schrieb Sven Schnelle:
>> <snip>
>>
>>>> Does anyone also still have this problem? Or does anyone have a
>>>> pointer for me before i start digging into that problem?
>>> I do not have any pointers right now but really interested in your
>>> investigations.
>> One thing, we had much truble with our toolchain. Which toolchain do you
>> use? Could you provide me (off list) a u-boot.bin for atstk1002 or
>> grasshopper?
>
> I'm using the atmel avr32 toolchain 3.2.3, which contains avr32-gcc
> 4.4.3 and binutils 2.20.1.
> I can't try the older atmel toolchain due to missing dependencies for
> my OS (Debian).

We use a self-build toolchain from some older buildroot which has

 * avr32-linux-gcc (GCC) 4.2.2-atmel.1.1.3.buildroot.1-20100804
 * GNU ld (GNU Binutils) 2.18.atmel.1.0.1.buildroot.1

gcc got another patch to have the build date in version string.

> i've just recognized that ld complains during built:
>
> avr32-ld:built in linker script:15: warning: memory region `FLASH' not
> declared
> avr32-ld:built in linker script:69: warning: memory region `CPUSRAM'
> not declared

So you do not have a *-linux-* toolchain?

> not sure if this is related to the problem i'm seeing.

Maybe ... I guess since I could build working binaries with my toolchain
something toolchain related is going wrong.

Maybe have a look for
http://distribute.atmel.no/tools/opensource/avr32-gcc/ (the script
should build a uclibc toolchain) to get another toolchain.
I had a short look for these patches some days ago cause we are hunting
another maybe toolchain related problem. AFAIK there are some different
linker settings. The avr32-linux toolchain does set the processor to
ap7000 (is it -m?) but the avr32-elf (which uses newlib) does not set
ap7000 as default. Maybe this could be the solution for your problem.

best regards

Andreas Bie?mann

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Andreas Bießmann]

Dear Sven,

Am 11.10.2011 15:28, schrieb Andreas Bie?mann:
> Dear Sven,
> 
> Am Di 11 Okt 2011 15:23:00 CEST schrieb Andreas Bie?mann:
>> Dear Sven,
>>
>> Am 11.10.2011 15:06, schrieb Sven Schnelle:

>> One thing, we had much truble with our toolchain. Which toolchain do you
>> use? Could you provide me (off list) a u-boot.bin for atstk1002 or
>> grasshopper?

> i've just build a version for grasshopper:

> Sven

Built with my toolchain:

---8<---
U-Boot 2011.09-00253-gac77ce1 (Oct 11 2011 - 15:57:07)

U-Boot code: 00000000 -> 000110ac  data: 000177b0 -> 0001e7e0
VMR table @ 0x13ff9170
DMA: Using memory from 0x13f9d000 to 0x13fa1000
Flash: 8 MiB at address 0x00000000
DRAM Configuration:
Bank #0: 10000000 64 MiB
In:    serial
Out:   serial
Err:   serial
Net:   macb0
Press SPACE to abort autoboot in 1000 seconds
U-Boot>
--->8---

Your u-boot.bin

---8<---
U-Boot 2011.09-00251-gf530e92 (Oct 11 2011 - 15:38:39)

U-Boot code: 00000000 -> 00013a70  data: 0001b420 -> 0004fe40
VMR table @ 0x0001bd94

--->8---

As you can see, your u-boot.bin is not working here.
I suspect some toolchain problem on your side.

best regards

Andreas Bie?mann

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Sven Schnelle]

On 10/11/2011 03:52 PM, Andreas Bie?mann wrote:
> Dear Sven,
> 
> Am Di 11 Okt 2011 15:37:29 CEST schrieb Sven Schnelle:
>> On 10/11/2011 03:28 PM, Andreas Bie?mann wrote:
>>> Dear Sven,
>>>
>>> Am Di 11 Okt 2011 15:23:00 CEST schrieb Andreas Bie?mann:
>>>> Dear Sven,
>>>>
>>>> Am 11.10.2011 15:06, schrieb Sven Schnelle:
>>> <snip>
>>>
>>>>> Does anyone also still have this problem? Or does anyone have a
>>>>> pointer for me before i start digging into that problem?
>>>> I do not have any pointers right now but really interested in your
>>>> investigations.
>>> One thing, we had much truble with our toolchain. Which toolchain do you
>>> use? Could you provide me (off list) a u-boot.bin for atstk1002 or
>>> grasshopper?
>>
>> I'm using the atmel avr32 toolchain 3.2.3, which contains avr32-gcc
>> 4.4.3 and binutils 2.20.1.
>> I can't try the older atmel toolchain due to missing dependencies for
>> my OS (Debian).
> 
> We use a self-build toolchain from some older buildroot which has
> 
>  * avr32-linux-gcc (GCC) 4.2.2-atmel.1.1.3.buildroot.1-20100804
>  * GNU ld (GNU Binutils) 2.18.atmel.1.0.1.buildroot.1
> 
> gcc got another patch to have the build date in version string.
> 
>> i've just recognized that ld complains during built:
>>
>> avr32-ld:built in linker script:15: warning: memory region `FLASH' not
>> declared
>> avr32-ld:built in linker script:69: warning: memory region `CPUSRAM'
>> not declared
> 
> So you do not have a *-linux-* toolchain?
> 
>> not sure if this is related to the problem i'm seeing.
> 
> Maybe ... I guess since I could build working binaries with my toolchain
> something toolchain related is going wrong.
> 
> Maybe have a look for
> http://distribute.atmel.no/tools/opensource/avr32-gcc/ (the script
> should build a uclibc toolchain) to get another toolchain.
> I had a short look for these patches some days ago cause we are hunting
> another maybe toolchain related problem. AFAIK there are some different
> linker settings. The avr32-linux toolchain does set the processor to
> ap7000 (is it -m?) but the avr32-elf (which uses newlib) does not set
> ap7000 as default. Maybe this could be the solution for your problem.

I downloaded the avr32-buildroot, and had it built a
gcc-4.2.4/binutils-2.18 toolchain, and the u-boot built with that
version works. so this looks indeed like a toolchain issue.

It's quite interesting that the original atmel toolchain fails, as i've
chosen this one to be one the safe side, but obviously this wasn't a
good idea.

Thanks for your help.

Sven

[U-Boot] u-boot not working on Atmel ATNGW100 (AVR32)

[Andreas Bießmann]

Dear Sven,

Am 11.10.2011 18:15, schrieb Sven Schnelle:
> On 10/11/2011 03:52 PM, Andreas Bie?mann wrote:
>> Dear Sven,
>>
>> Am Di 11 Okt 2011 15:37:29 CEST schrieb Sven Schnelle:
>>> On 10/11/2011 03:28 PM, Andreas Bie?mann wrote:
>>>> Dear Sven,
>>>>
>>>> Am Di 11 Okt 2011 15:23:00 CEST schrieb Andreas Bie?mann:
>>>>> Dear Sven,
>>>>>
>>>>> Am 11.10.2011 15:06, schrieb Sven Schnelle:

<snip>

>> Maybe have a look for
>> http://distribute.atmel.no/tools/opensource/avr32-gcc/ (the script
>> should build a uclibc toolchain) to get another toolchain.
>> I had a short look for these patches some days ago cause we are hunting
>> another maybe toolchain related problem. AFAIK there are some different
>> linker settings. The avr32-linux toolchain does set the processor to
>> ap7000 (is it -m?) but the avr32-elf (which uses newlib) does not set
>> ap7000 as default. Maybe this could be the solution for your problem.
> 
> I downloaded the avr32-buildroot, and had it built a
> gcc-4.2.4/binutils-2.18 toolchain, and the u-boot built with that
> version works. so this looks indeed like a toolchain issue.
> 
> It's quite interesting that the original atmel toolchain fails, as i've
> chosen this one to be one the safe side, but obviously this wasn't a
> good idea.

Well, the AP7000 is a obsolete device. Linux on AVR32 is not longer
supported officially.
Unfortunately Atmel has some special behaviour to the open source
community. AVR32 is still not an official gcc port and I can not see any
effort to get it in, same for libc, binutils, a.s.o.

But you can build a linux toolchain out of the official atmel patches by
using the script from
http://distribute.atmel.no/tools/opensource/avr32-gcc/. I guess their
'official' toolchain (which is a newlib one targeted on avr32 ucX
devices) is also built with this script.
I did use the patches from this site to build my own toolchain some days
ago and can confirm that a
'avr32-unknown-linux-gcc-4.4.3-uclibc-0.9.32-binutils-2.20.1'-toolchain
can successfully build u-boot for avr32.

If you like to play around avr32 is ok, but if you do some serious
things and like to stick at atmel I recommend to switch to sam9 devices
soon. There are also some interesting devices from other vendors:
plugcomputer, beagle, ... to name just two large device families
available for hobbyists.

best regards

Andreas Bie?mann