[rtems commit] bsps/arm: Use ALIGN_WITH_INPUT

Chris Johns chrisj at rtems.org
Thu Dec 19 08:17:48 UTC 2013

On 19/12/2013 6:36 pm, Sebastian Huber wrote:
> On 2013-12-19 01:30, Chris Johns wrote:
>> On 17/12/2013 1:13 am, Sebastian Huber wrote:
>>> Module:    rtems
>>> Branch:    master
>>> Commit:    287bbb65afd24ffc6254ae5f328733213f184205
>>> Changeset:
>>> http://git.rtems.org/rtems/commit/?id=287bbb65afd24ffc6254ae5f328733213f184205
>>> Author:    Sebastian Huber <sebastian.huber at embedded-brains.de>
>>> Date:      Fri Aug 30 17:43:16 2013 +0200
>>> bsps/arm: Use ALIGN_WITH_INPUT
>>> This requires at least Binutils 2.24.
>>> ---
>>>   .../lib/libbsp/arm/shared/include/linker-symbols.h |    6 -
>>>   c/src/lib/libbsp/arm/shared/include/start.h        |    7 -
>>>   c/src/lib/libbsp/arm/shared/startup/linkcmds.base  |  194
>>> +++++++++-----------
>> I will be reverting this change ...
>> /usr/home/chris/development/rtems/4.11/bin/../lib/gcc/arm-rtems4.11/4.8.1/../../../../arm-rtems4.11/bin/ld:linkcmds.base:68:
>> syntax error
>> so I am broken and cannot commit the patches.
> It would have been nice to wait some hours to give me a chance to
> comment on this.

I need a current RTEMS with the old tools and I needed to push my 
changes before any dependency on tools and our time zones did not line up.

>> I think you will need to find another way to integrate this change if it
>> depends on a specific binutils. Sure the RSB has the new binutils
>> however this
>> needs to work through into the tool sets in use and that takes time.
> The new Binutils are in the RPMs and the RSB.  Updating Binutils is not
> that time consuming.

Maybe stepping gcc, newlib and binutils was a mistake. Maybe I should 
have stepped binutils by itself and avoided this issue however I did not 
know this dependence existed. The gcc change is what stops me updating 
my release tools at this point in time.

I have almost completed what I need to do.

>> Also please note the commit message states what the patch is not what
>> is fixes
>> or provides. I have no idea what this patch addresses and what it
>> attempts to
>> fix or provide.
> Sorry about this, but my experience of the past is that nobody in the
> RTEMS project except me cares about the linker script copy and paste
> cleanup.  I sent several emails to the list and got no answers.
> This ALIGN_WITH_INPUT is absolutely necessary to make this linker script
> reliable.  We talked about this issue several months ago.
> https://sourceware.org/ml/binutils/2013-06/msg00246.html
> https://sourceware.org/binutils/docs-2.24/ld/Forced-Output-Alignment.html#Forced-Output-Alignment

Thanks. I think the change is fine if I could use a suitable binutils.


More information about the devel mailing list