Skip to contentSkip to conversation

at91rm9200ek NOR booting is broken

U-Boot Andreas Bießmann 2 participants 4 messages Dec 1, 2010

lore
Share & tools
lore
#0 lore raw

Dear Albert Aribaud,

as mentioned in another thread here is the description of broken at91rm9200ek NOR booting.

I use the patchset posted in
http://thread.gmane.org/gmane.comp.boot-loaders.u-boot/89974 on top of v2010.12-rc2 tag. I configure the code for 'at91rm9200ek_config' and build that with a homebrew 4.5.1 toolchain with 2.20.1 binutils (I think the toolchain does not matter).
The same board configured with 'at91rm9200ek_ram_config' do work with an JTAG. The JTAG configuration is copied from u-boot (just to mention the SDRAM and clock initialisation is sane).

regards

Andreas Bie?mann

#1 lore raw

thanks Andreas,

Le 01/12/2010 13:37, Andreas Bie?mann a ?crit :

Quoted 5 lines

Dear Albert Aribaud,

as mentioned in another thread here is the description of broken at91rm9200ek NOR booting.

Quoted 8 lines

I use the patchset posted in
http://thread.gmane.org/gmane.comp.boot-loaders.u-boot/89974 on top of v2010.12-rc2 tag. I configure the code for 'at91rm9200ek_config' and build that with a homebrew 4.5.1 toolchain with 2.20.1 binutils (I think the toolchain does not matter).
The same board configured with 'at91rm9200ek_ram_config' do work with an JTAG. The JTAG configuration is copied from u-boot (just to mention the SDRAM and clock initialisation is sane).

What is the exact symptom? does the board display anything on its console, or does it stay completely silent?

Quoted 3 lines

regards

Andreas Bie?mann

Amicalement,

Signature · 1 line
Albert.
#2 lore raw

Dear Albert ARIBAUD,

Am 01.12.2010 13:49, schrieb Albert ARIBAUD:

Quoted 6 lines

thanks Andreas,

What is the exact symptom? does the board display anything on its console, or does it stay completely silent?

I see the banner and 'DRAM: 32 MiB' on serial console. If I activate DEBUG in a/a/l/board.c I can see the debug messages up to 'relocation Offset is ...'

regards

Andreas Bie?mann

#3 lore raw

Note: pwclient dies on patch 73644 with the following

albert at lilith:~/src/u-boot$ pwclient apply 73644
Applying patch #73644 to current directory
Description: [U-Boot,v2,1/4] at91rm9200ek: add configure target for RAM boot Traceback (most recent call last):
File "/home/albert/bin/pwclient", line 463, in <module>

     main()

File "/home/albert/bin/pwclient", line 446, in main
action_apply(rpc, patch_id)
File "/home/albert/bin/pwclient", line 263, in action_apply
proc.communicate(s)
File "/usr/lib/python2.6/subprocess.py", line 680, in communicate
self.stdin.write(input)
UnicodeEncodeError: 'ascii' codec can't encode character u'\xdf' in position 605: ordinal not in range(128)

I'd wager it does not like Andreas' last name.

Le 01/12/2010 16:13, Andreas Bie?mann a ?crit :

Quoted 7 lines

Dear Albert ARIBAUD,

Am 01.12.2010 13:49, schrieb Albert ARIBAUD:

Quoted 4 lines

thanks Andreas,

Quoted 6 lines
Quoted 2 lines

What is the exact symptom? does the board display anything on its console, or does it stay completely silent?

I see the banner and 'DRAM: 32 MiB' on serial console. If I activate DEBUG in a/a/l/board.c I can see the debug messages up to 'relocation Offset is ...'

Thanks.

The size of u-boot.bin is right for sections .text to .dynsym inclusive as listed by readelf -a.

The content of the binary u-boot.bin correctly reflects the .rel.dyn and .dynsym sections of the ELF u-boot file.

_rel_dyn_start_ofs, _rel_dyn_end_ofs and _dynsym_start_ofs> all contain correct offsets to their respective targets.

This means that if the u-boot.bin is correctly flashed to NOR, and if nothing touches the .rel.dyn and .dynsym section, relocation is done correctly.

But... with your fixes applied above v2010.12-rc2, as you indicated, you don't have the fix for not using r8 during relocation, which means GD is completely bogus after relocation. Can you apply this fix too?

Quoted 3 lines

regards

Andreas Bie?mann

Amicalement,

Signature · 1 line
Albert.