[uClibc] Problems building for AMD Alchemy processors.

Charles Palmer charles.palmer at acutetechnology.com
Mon May 23 22:18:06 UTC 2005


Hi Erik

I had tried that (it seemed natural!) but make fails if I select both "Board 
support for the AMD development boards " and "AMD Alchemy DBAu1500 
Development Board Support". (I've now done this twice with identical 
results).

So:

1    This doesn't work for me (why?)

2    It is not clear what selecting the AMD board is for - can you tell me? 
If the function of buildroot is to build cross-compilation tools, libraries 
and a root file system, then why does it need to know what my target 
hardware board is? And why (as seems to be the case below) are we building 
an ethernet driver?

3    My board uses an Au1200 chip, rather than an Au1500 chip. The Au1200 
does not have the in-built Ethernet controller that the Au1500 has (though 
my board has an SMC Ethernet chip), so even if the make process did not 
fail, it is not clear to me that it would have built tools and a filesystem 
that is relevant to my hardware.

4    And anyway (back to my original question) why is it that my original 
build process and its resulting cross-tools appear to be building code that 
does not run on my target?

5    Could it be that the reason that make fails with the Au1500 selected is 
the same reason that programs built with the toolchain won't run (when built 
with the Au1500 de-selected)? Is there something wrong with my paths, or 
tools, or Linux distribution, or environment variables or....?

(Sorry to pose these difficult problems!!!)

Here is the error that I got:


/home/acutetech/buildroot/build_mipsel/staging_dir/bin/mipsel-linux-uclibc-gcc 
 -D__KERNEL__ -I/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik/include 
 -Wall -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -fomit-frame-pointer 
 -Os -falign-functions=2 -falign-jumps=2 -falign-labels=2 -falign-loops=2 -I 
/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik/include/asm/gcc -G 
0 -mno-abicalls -fno-pic -pipe  -finline-limit=100000 -mabi=32 -march=mips32 
 -Wa,-32 -Wa,-march=mips32 -Wa,-mips32 -Wa,--trap   -nostdinc -iwithprefix 
include -DKBUILD_BASENAME=au1000_eth  -c -o au1000_eth.o au1000_eth.c
au1000_eth.c: In function `au1000_init_module':
au1000_eth.c:97: sorry, unimplemented: inlining failed in call to 
'str2eaddr': function body not availableau1000_eth.c:1219: sorry, 
unimplemented: called from here
au1000_eth.c: At top level:
au1000_eth.c:142: warning: 'phy_link' defined but not used
make[4]: *** [au1000_eth.o] Error 1
make[4]: Leaving directory 
`/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik/drivers/net'
make[3]: *** [first_rule] Error 2
make[3]: Leaving directory 
`/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik/drivers/net'
make[2]: *** [_subdir_net] Error 2
make[2]: Leaving directory 
`/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik/drivers'
make[1]: *** [_dir_drivers] Error 2
make[1]: Leaving directory 
`/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik'
make: *** [/home/acutetech/buildroot/build_mipsel/linux-2.4.29-erik/vmlinux] 
Error 2

Regards - Charles

----- Original Message ----- 
From: "Erik Andersen" <andersen at codepoet.org>
To: "Charles Palmer" <charles.palmer at acutetechnology.com>
Cc: <uclibc at uclibc.org>
Sent: Monday, May 23, 2005 9:28 PM
Subject: Re: [uClibc] Problems building for AMD Alchemy processors.


> On Mon May 23, 2005 at 08:11:25PM +0100, Charles Palmer wrote:
>> Hi team
>>
>> Has anyone suceesfully built a toolchain for these processors? If so, 
>> how?
>> (Many thanks in anticipation of your help...)
>
> Yes.  I use buildroot, use 'make menuconfig' to select an AMD
> Alchemy then run 'make' to compile things.  Works nicely on my
> AMD Alchemy board...
>
> -Erik
>
> --
> Erik B. Andersen             http://codepoet-consulting.com/
> --This message was written using 73% post-consumer electrons--
> 




More information about the uClibc mailing list