Motorola Shared BSPs not Linking

[Till Straumann]

I just did a git clone of 'master' but was unable to reproduce your
problem. The 'motorola_shared' BSPs build just fine.

I do use an older toolchain (but don't have time to upgrade right now):

binutils-2.20.1
gcc-4.5.2 (with gcc-core-4.5.2-rtems4.11-20110220.diff and 
gcc-g++-4.5.2-rtems4.11-20110131.diff)
newlib-1.19.0 (with newlib-1.19.0-rtems4.11-20110724.diff)

-- Till

On 02/23/2012 05:34 PM, Joel Sherrill wrote:
> On 02/23/2012 10:15 AM, Till Straumann wrote:
>> My last build of 'master' still linked fine but that's been a while
>> (7/2011, I believe).
> That's a while back. :)
>
> I did a git blame and nothing has changed recently.  And
> I saw nothing really changing since we merged the
> qemuprep* BSPs.
>> Could you provide the console messages when 'hello.exe' was
>> linked along with a linker map?
> I will send you my entire build log privately.
>
> But it looks like the beatnik BSP and others reference
> preload.S as part of their startup asm. motorola_shared
> does not. But that has been part of that BSP forever.
> I just don't see what changed.
>> - Till
>>
>> On 02/23/2012 03:17 PM, Ralf Corsepius wrote:
>>> On 02/22/2012 05:54 PM, Joel Sherrill wrote:
>>>> Hi
>>>>
>>>> Just came across this in the build logs and unfortunately
>>>> don't have time to investigate. This was on the mcp750
>>>> but every motorola_shared variant fails the same way.
>>>>
>>>> powerpc-rtems4.11-objcopy -O binary -R .comment -S hello.exe rtems
>>>> gzip -vf9 rtems
>>>> rtems: 55.1% -- replaced with rtems.gz
>>>> powerpc-rtems4.11-ld -o hello.ralf
>>>> ../../../../../mcp750/lib/bootloader.o --just-symbols=hello.exe -b
>>>> binary rtems.gz -T ../../../../../mcp750/lib/ppcboot.lds -Map hello.map
>>>> &&    chmod 755 hello.exe
>>>> ../../../../../mcp750/lib/ppcboot.lds:97: undefined symbol
>>>> `__rtems_start' referenced in expression
>>> I can reproduce this.
>>>
>>>> Anyone have any ideas or fixes?
>>>>
>>> The only place in rtems.git which provides a symbol named __rtems_start
>>> is c/src/lib/libbsp/powerpc/shared/start/vectors_entry.S
>>>
>>> However, this file doesn't seem to be referenced by the mcp750 BSP nor
>>> by the motorola_powerpc:
>>>
>>> # grep -R shared/start/vectors_entry.S
>>> c/src/lib/libbsp/powerpc/*/Makefile.am
>>> c/src/lib/libbsp/powerpc/beatnik/Makefile.am:EXTRA_DIST +=
>>> ../../powerpc/shared/start/vectors_entry.S
>>> c/src/lib/libbsp/powerpc/beatnik/Makefile.am:vectors_entry.$(OBJEXT):
>>> ../../powerpc/shared/start/vectors_entry.S
>>> c/src/lib/libbsp/powerpc/motorola_powerpc/Makefile.am:libbspstart_a_SOURCES
>>> += ../../powerpc/shared/start/vectors_entry.S
>>> c/src/lib/libbsp/powerpc/mvme3100/Makefile.am:EXTRA_DIST +=
>>> ../shared/start/vectors_entry.S
>>> c/src/lib/libbsp/powerpc/mvme3100/Makefile.am:vectors_entry.$(OBJEXT):
>>> ../shared/start/vectors_entry.S
>>> c/src/lib/libbsp/powerpc/mvme5500/Makefile.am:
>>> ../../powerpc/shared/start/vectors_entry.S
>>>
>>>
>>> Seems to me, as if some of the powerpc exception/irq "generalizations"
>>> broke something.
>>>
>>> I am not sufficiently familiar with how this is supposed to work to be
>>> able to provide a fix.
>>>
>>> Ralf
>>> _______________________________________________
>>> rtems-users mailing list
>>> rtems-users at rtems.org
>>> http://www.rtems.org/mailman/listinfo/rtems-users
>