lpc40xx BSPs do not build on master

[Joel Sherrill]

FWIW arm gcc 8 is the culprit. I have confirmed it with an unmodified RTEMS
master.

lpc40xx_ea_r* do not build with gcc 8/master and RTEMS master.

configure:3595: arm-rtems5-gcc -mthumb -march=armv7-m -mfpu=fpv4-sp-d16
-mfloat-abi=hard -mtune=cortex-m4 -O2 -g -ffunction-sections
-fdata-sections   conftest.c  >&5
/data/home/joel/test-gcc/install-master/bin/../lib/gcc/arm-rtems5/8.0.0/../../../../arm-rtems5/bin/ld:
error: /tmp/ccZTj7BL.o uses VFP register arguments, a.out does not
/data/home/joel/test-gcc/install-master/bin/../lib/gcc/arm-rtems5/8.0.0/../../../../arm-rtems5/bin/ld:
failed to merge target specific data of file /tmp/ccZTj7BL.o
/data/home/joel/test-gcc/install-master/bin/../lib/gcc/arm-rtems5/8.0.0/../../../../arm-rtems5/bin/ld:
warning: cannot find entry symbol _start; defaulting to 0000000000008018
collect2: error: ld returned 1 exit status


On Tue, Jan 16, 2018 at 5:15 PM, Chris Johns <chrisj at rtems.org> wrote:

> On 16/1/18 11:17 am, Joel Sherrill wrote:
> > I have most of the architectures and targets building with empty
> bsp_spec files.
>
> It is not clear to me how this change and the no-preinstall changes will
> interact, that is which is pushed to master first.
>
> The BSP spec files themselves are not touched in the no-preinstall branch,
> in
> fact the only place they are referenced is linking the samples and tests
> and
> installing.
>

I think they are orthogonal changes. Unfortunately, emptying the bsp_specs
files requires modifying the built in info in gcc. Some of which was just
right
enough to configure anyway. That's why we had to have bsp_specs to build
real applications.

My plan is that once they are all empty, I can then add the patches to not
use
them and them remove them.

>
> Chris
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20180116/e2e5aef7/attachment-0001.html>